>>3007000
Literal page 2 bake? That’s barely half the thread’s full lifetime. It’s a general, not like front page bumping matters at all. Waiting until page 8-10 is how it’s been done for at least a few years.

>>3007000
I'm looking for a new Multimeter and I can't decide between the Amazon Commercial 90DM619 and the Kaiweets HT118E.
I've also taken a look at the UNI-T UT139C, but that's not available right now.

Any meters that I'm missing? Budget is 50€ tops.

>>3008477
90DM610*
Missing my Numpad right now.

File: 1789557029472558.png (82 KB, 1659x1154)

82 KB PNG

>>3008497 >>3008514

>>3007000
>>3008491
>Diodes pointing towards FB means you get the greater of the two feedback signals, diodes pointing away from FB (wth a pull-up resistors) means you get the lesser of the two feedbacks signals.
i see this now (in ltspice.) thanks for saving me a frustratingly stupid board revision.

>Even the small amount of ripple that gets through your low-pass filter will be hugely amplified by your error amplifiers, resulting in a bunch of output ripple.
are PWM DACs more difficult than i'm perceiving? settling time doesn't really matter to me. 100 khz PWM through a 30K and 100nF RC filter should yield roughly 1.65 mv ripple peak to peak on a 3.3V input. that's close to 11 bits? afaik DACs don't have output ripple but shouldn't this ripple be functionally the same as the precision/accuracy of a 10-11 bit DAC?

>Seems like Q1A is a bit superfluous
Q1A exists to shut off the charging power path no matter what happens. i selected them to withstand VCC (20V from USB-C.) normal charge termination? shut off charging. buck converter fails and the output rail sees 20V? shut off charging.

>If it’s tied to ground in a single-supply circuit, it will clip any negative voltages
i don't want to charge a dead battery since it's unsafe. off-board comparators check for a minimum voltage, like 3.0V for an 18650. a reverse polarity cell would read 0V, as would an absent cell. a dead cell would read <3V. comparators trigger SD LATCH, no current flow can occur because Q1A/Q1B are held off.

>>3008496
typo:

Q1B exists to shut off the charging power path no matter what happens. i selected them to withstand VCC (20V from USB-C.) normal charge termination? shut off charging. buck converter fails and the output rail sees 20V? shut off charging.

Q1A exists to shut off the discharging power path in the case of the IRFP250N blowing up.

>>3008497
> IRFP250N
Why not use bipolar, they’re more robust and usually handle around 60 V

File: images.jpg (31 KB, 447x447)

31 KB JPG

>>3008516

>>3008498
>Why not use bipolar, they’re more robust and usually handle around 60 V
i have a dozen IRFP250N on hand. i've used them in pic related to 100W and 10A sustained. they should be more than adequate for 10-20W (heat sink limited).

>>3008496
>that's close to 11 bits
Yeah but the R2R DAC isn’t oscillating by half an LSB thousands of times per second. If your error amplifier gain is like 0dB at the PWM frequency it’s probably not an issue, but I’d definitely calculate that.

Q1A won’t shut off charging, only discharging. If your buck converter outputs 20V, you’ll have to catch that in software and shut off Q1B.

I guess worst case with that voltage sense amplifier is it starts clipping the signal, so you detect a 0V battery or a backwards battery as the same. But I’d still want to be able to tell the difference between those, and avoid any issues with the amplifier not behaving well close to its negative output rail. I’d want to be able to measure down 2.5V at least, for LFPs and DW01-protected cells.

>>3008501
I’m looking at the datasheet, you’ll be in the thermal instability range at anywhere below 45A in the VCCS operating mode. So I’d be cautious.

How does an op-amp work?

>>3008550
> how opamp work
It’s like a super-good transistor.
The “op” is that you can configure it to do summing, difference, integration, inverting, etc etc ad nauseam with a bit of extra circuitry.
The “amp” is that it amplifies with configurable gain.

>>3008550
Long-tailed pair (possibly with darlingtons) as an initial differential amplifier, plus a few additional (I think common-emitter) gain stages, followed by a push-pull output stage. The long-tailed pair has a very high effective input impedance, and the push-pull stage has a very low effective output impedance. With all that gain in the middle, any nonlinearities and parasitic behaviour flattens out when you use negative feedback.

File: 20260917_035516.jpg (3.52 MB, 4000x3000)

3.52 MB JPG

>>3008583 >>3008587 >>3008615

This is what I am being taught in college.

It's flat out 2 amps and I have calculated the math all other resistors. Am I retarded?

File: 27629_50 (1).jpg (183 KB, 540x548)

183 KB JPG

>>3008615

>>3008582
Here is another one, one typo means the problem is wrong, even spaces count the problem as wrong and it does not ask for the math. I did it in fractions, not decimals.

>>3008582
I don't know why but my screenshots are always sideways when they are vertical in ny gallery. Anyways, I ain't ask for homework, these are already counted against me.

File: 71xTcOOe6XL._SL1500_[1].jpg (162 KB, 1500x1500)

162 KB JPG

>>3008619

>>3007000
Can I run 24v through a 12v KDC4 switch. 12v isnt giving my motors the torque I want and the electronics housing is designed for the kdc4. Now Im reading that kdc4 switches are designed for AC and not really meant to handle more than 12v but im confused because some switches say up to 36v (despite have a 12v label).

>>3008587
>>3008583
>>3008582
It is because you are posting from a phone. Phones assume all their users are retarded, when only most of their users are retarded. Post from a PC. You know, like the one you are taking the picture of? You can screenshot whatever you want using Windows Key + Shift + S. If that is too hard, and you use Google Photos, you can access your phones on the web by going to photos.google.com. You can then download them and use an image viewer like IrfanView to make sure they are rotated properly before posting them here.

If you are on an iPhone, well, just ignore everything I just said and deal with the fact that your pictures are sometimes rotated.

>>3008613
Usually you can, but it will have it’s current derated.
DC has an sustained arcing problem, that’s why the DC current rating is so ridiculously low.
There’s probably a formula for it based off the AC rating, but it’s likely non-linear and they can’t print every DC voltage/current rating on it.

>>3008619
Is there any way I can figure out the dc/current rating. I realized I forgot to include the current. The switch I have is rated for 12v 40a or 250vac 30a. Im trying to run 24v 8~10a max.

File: get-rotated.gif (210 KB, 220x219)

210 KB GIF

>>3008615
Fair

>>3008628
At absolute worst it will be 50% of the 12V rating, it’s probably closer to 75%. So 10A will be fine if it’s rated for 40A at 12V. One technique is to find comparable rocker switches on places like DigiKey that give you more thorough datasheets.

File: pepe-is-caught-in-the-cen(...).jpg (61 KB, 1200x1200)

61 KB JPG

>>3008666

>>3008615
I actually did try doing a screenshot but the screenshot takes up too much space to be used on 4chan.

File: images[1].jpg (33 KB, 429x466)

33 KB JPG

>>3008696 >>3008824 >>3010640

>>3008641
How to prove you are a Zoomer without outright saying you are a Zoomer.
You can change the file format of the image. Use JPG instead of PNG. It takes up much less space. Also, don't take a full screen capture. Use 'snipping area' to only capture the part of the screen you need.

>>3008636
Thanks. I was afraid I was going to have change to a different switch model and have to design the housing from scratch.

File: ssr.jpg (55 KB, 735x587)

55 KB JPG

>>3008693 >>3008699

They don't seem to make pcb-mountable SSRs at the current rating I need (5-6A) plus they're kinda pricey anyway. So I'm thinking of implementing it myself.
Is it just an opto and a triac? Just wire the micro into a zero-crossing opto then wire that up to the triac?
Do I need to care about zero-crossing on the micro, maybe to align the control signal it emits?

>>3008692
If they were to implement SSRs today, they’d probably use two mosfets back-to-back in opposite directions to handle the AC.

I don’t think you have to worry about zero crossing that much in this application unless the noise spike is causing problems.

Why solid state? Lots of things still use regular relays for switching heating elements (stoves, toaster ovens). Some even manage to attempt to control the relay at the z/c like mr carlson did on his refrigerator to increase the lifespan of the little pcb relay… in addition to the snubber network.

>>3008693
>they’d probably use two mosfets back-to-back in opposite directions to handle the AC
I've seen this in diagrams too, any advantages to this approach?
>I don’t think you have to worry about zero crossing that much in this application
Space is pretty tight so I was hoping it'd reduce heat somewhat not having to switch at the peak. I may be misunderstanding this though.
>Why solid state?
Fast switching to vary power to a heating element since it's heating a very small mass and has to stick closely to a temperature.
(Specifically it's for a coffee machine)

>>3008666
I wish I was a zoomer lol, grew up without a phone. But good tip.

>>3008696
What's growing up with a phone got to do with anything? When I was a kid, phones came in two types: Rotary and Touch-Tone. When smartphones came out I hated them. I only got one because a new job I got in 2014 required one. I still hate them. Laptops and tablets too. 104 keys, a full sized mouse, and at least two big screens is the only way to do it. Gotta have a sound system too. Headphones are junk.

>I wish I was a zoomer lol
Are you saying you are were born before Windows 95 released yet don't know how to use a computer? Fuck dude, what have you been doing? Not getting ripped and fucking chicks, that's for sure. You are on 4chan.

>>3008692
If you want easy-mode, buy a digitally controlled dimmerlink board from rbdimmer.com. Cheaper than aliexpress. You don’t need to care about zero crossing synchronisation, there’s a little MCU that handles that for you, just speak UART or I2C to it and say what dimming percent to set. They make both monolithic boards with a triac and optotriac, and controller boards that have just the MCU on it. That said, my reading suggests they don’t do pulse skipping, only phase-angle control. So good for loads that need high speed, bad for situations where EMI matters.

A conventional circuit with zero-crossing is definitely doable yourself, just make sure you’ve got enough timers and interrupt time for it. Maybe the RP2040’s PIO system can handle it for you without interrupting the CPU. If you want to do phase angle control, I believe optotriacs with built-in zero-crossing circuits won’t work, they wait until zero crossing before turning on the output. Just learnt this today.

If speed REALLY doesn’t matter, as in lots of full AC cycles between turning on and off, then forget the zero crossing and just switch it on and off open-loop. A zero-crossing optotriac will make for low-transient switching operations.
But if you want something faster, especially PID control, then pulse skipping, if not full phase-angle, with a zero-crossing detection circuit is the way to go. They’re pretty easy, just an optocoupler and some diodes and resistors and stuff. It’s basically an established design. The code is the harder part for me.

>>3008693
I’ve also seen people put a bridge rectifier in series with the AC load and put a single MOSFET in there. Easier to drive, but in both cases they need isolated driving circuits that are more difficult than an optotriac. At least you can just arbitrarily PWM the MOSFETs without caring about zero crossing, but whether you should is another matter. Seems like a recipe for even more EMI than phase-angle control.

>>3008697
Based cellphone hater. I’m waiting for an open source hardware dumbphone that’s actually good. Calls, texts, maps, ideally an end-to-end encrypted messaging platform, and maybe an authenticator for 2-factor things. Headphone jack. Can act as a 4G router for your cyberdeck. A week of battery life because it doesn’t do anything in the background, besides polling for messages and calls. Let people write custom executables for it in Fortran.
>headphones
You should take the chifi IEM pill. When the cables break, you replace just the cables. I recommend the Truthear Gate, though lately I got the Salnotes Zero 2 and it’s been great.

How do I make my own drill battery charger if the one that came with my drill craps out? Does it have those microcontroller thingies that would detect you're using a counterfeit one? It's 20V and I just wanna hook it up to a regular 20V power supply

File: images[1].jpg (19 KB, 415x739)

19 KB JPG

>>3008724

>>3008700
Sorry bro, they ain't gonna hold a candle to my 5 speaker surround sound system with full range speakers. When I upgrade my rig, once the AI bubble pops, I'm going to get a sound card that can handle 7 speaker surround. I already have the speakers for it. Earbuds are even worse than headphones. Their tinny little dogshit drivers cannot handle proper sound output. Especially when trying to reproduce wildly different frequencies simultaneously. Headphones are super bad about it and earbuds are worse.

At least they don't put a dent in your head. Picture related.

>>3008707
Get a CC/CV DC-DC converter module. Probably a buck converter module like an LM2596 (the one with two or three trimpots), run it off a 24V supply. Set to probably 21.0V and some sensible amount of current (less than the converter’s maximum, and less than your 24V adapter’s maximum too), verify the settings with your multimeter. Wire the output up to one of those tool battery adapters. I think you can typically just feed power into its pins without any communication, though I’d probably seek a second opinion from an authority on your battery brand in particular.

>>3008709
I like having channel separation and no destructive interference between channels in the event my head is 5mm off the centre of the room. Proper IEMs often use multiple types of drivers to cover a wider range of frequencies. Because the required SPL at the driver is so low, the drivers end up being significantly underdriven, with plenty of headroom. If the frequency response is still a bit off, EQ can cover for that so long as you’re not running into distortion. Also I only have two ears, and the vast majority of music is only mixed in stereo anyhow.

Now I’d you’re arguing that the phase relationship isn’t accounted for by standard EQs, you’re right. But:
>human ears can’t really hear phase offsets in a note, though this isn’t an absolute I suspect there are cases where it might be noticeable
>a digital EQ could be designed to compensate phase too, it’s quite possible that such a thing already exists
>the phase relationship between multiple speakers that are spatially separated is even more significant and harder to control

File: 69a020e4-cc2e-11e2-9a4a-0(...).jpg (32 KB, 440x330)

32 KB JPG

>>3008697
This was my first phone, so you're gate keeping a person who is already further down the path than you.

Anyways, I had the instructor run through the math, this is an electrical engineer, only to figure out that 6/3 is 2.

File: images (53).jpg (29 KB, 300x562)

29 KB JPG

>>3008700
The vacuum tubes are EMP resistant for when the government electrocutes your balls

>>3008724
I spend most of my time on video games and movies/TV, so much of it does have surround sound mixes. I'm in a computer chair connected to a mouse and keyboard so I don't have a problem with moving out of the sweet spot. Plus, my amplifiers allow for multiple speaker configurations. If I wanted to, I could just get another set of speakers, mount them in an idea position for music listening only, and then turn a knob on the amp powering the front channels to have them power the 'music' speakers.

As for EQs, I listen to so many different genres of music from so many different eras that programming an EQ to be able to handle every album would be a nightmare. Everything from Anamanaguchi to the ZZ Top. Brian Bennett to Chuck Berry. Motörhead to Matthieu Chedid. Then add all the different games, movies, and TV shows I watch? Nah. I'd rather have a system that just sounds good no matter what I'm doing. Plus I like to feel it in my chest when shit blows up. You can't do that with earbuds and you don't even have to have the volume up that high to do it with full sized multi-driver speakers.

>>3008768
No you just EQ to make the speakers or IEMs sound neutral (e.g. harman target, or perfectly flat), then listen to the audio as intended by the album. The audio chain between the .mp3 bits and the sound pressure should be transparent, it shouldn’t colour the sound one way or the other. Swapping between EQs each time you listen to a different genre is absurd. If the mix is so bad you have to EQ it differently to make it sound good, then bake a modified frequency response into your .wav so it sounds like god intended without having to fiddle with your EQ settings.

I found a 2tb external hard drive with 30% health. Is it still possible to repair it?

>>3008795
Are the platters ok? Are the heads safely parked? The spindle motor/bearing?

>>3008699
>rbdimmer.com
Looks like cuckflare cucked them out of potential visits. Is "dimmerlink" a generic name or their own product?

>>3008666
>Be GenX
>Be completely forgotten
Also first half of GenZ is millennial-Lite and better with computers than actual millennials (I use Debian btw).

t. 28 y/o comp-sci Zoomer

>>3008819
I think it’s their trademark. I haven’t seen an equivalent from any other brand, except for arguably the Kemo M150.

Also I’m pretty sure you could replace the variable resistor in a normal analogue dimmer circuit with an LDR and blast it with an LED controlled by a microcontroller, the only issue being its maximum voltage rating. Maybe you could put a few LDRs in series and balance them with TVS diodes in parallel with each. Otherwise, looks like wider diameter photocells have higher voltage ratings, with the 20mm ones being capable of 500V.

>>3008695
> mosfets, any advantages to this approach?
Modern mosfets (hexfets) have insanely low “on” resistance and need less heatsinking which can be a major space saver.

>>3008699
> a bridge rectifier in series with the AC load and put a single MOSFET in there
That’s brilliant, especially if you have some spare bridge rectifiers around… same power, just pulsed DC.

File: 7cca0cadf0320353c02114390(...).jpg (126 KB, 682x976)

126 KB JPG

>>3008873

>>3008824
Ah, but the first half of Millennials (and the back half of Gen-X) are Xenials!

>Xennials are described as having had an analog childhood and a digital young adulthood

We are a post-computer generation but a pre-internet generation. Early Gen-Xers were already preteenagers/teens when computers started entering the home but many later ones, and early Millennials, grew up with early these computers. Remember, many of those home machines booted up to a BASIC interpreter or had one built in. That started you off in a very different mindset then starting from a GUI. Even business machines, like the IBM and its clones, generally booted to a command line (DOS). Plus all the best games ran in DOS.

You had to know how to use a computer to install games. It was a completely different experience to what came later. Check this shit out:
https://youtu.be/V-t-zTkNka8?t=90
https://www.tiktok.com/@geekjam/video/7397922669213076779
https://youtu.be/cioyLQ2O6yc
When you put a disk/disc into your PC it didn't do shit in DOS. You have to know what the drive letter was, how to list the contents of the directory, start the installer, configure your own sound & graphics settings, maybe even fuck around with extended memory, and then know how to navigate to where you installed the thing. Only then could you play. Lord fucking save you if you needed to setup a boot disk (fuck you Ultima VII).

Then came Windows and Direct X and installers became this:
https://www.youtube.com/watch?v=TCymoV8jVkw
Pop in the CD, it would autorun (this guy had to do it manually because modern PCs turn that shit off), chose install, and hit next a bunch. Done. If you were born too late you simply had no idea.

Even old consoles didn't hold you hand. No cartridge? Here's a flashing screen. Figure it out, dumbass. Better have it tuned the correct channel and have the switches set correctly. Some of you have no idea what I'm talking about.

The PlayStation gave you this:
https://youtu.be/r_Ifccgl05A

File: todays-find-sony-kv-9400-(...).jpg (161 KB, 1080x1080)

161 KB JPG

>>3008894

>>3008824
>>3008867
Xenials are the technology Precursors. Elder Gods. I've used a rotary phone. I have watched a film... on film using a projector that I operated. I have swapped reel to reel tapes (for music and data). I have programmed a VCR. I have tuned a TV. Not changed a channel, *tuned* each channel button using switches and dials like a DJ. Picture related. No one can do that anymore. Analog TV is dead. I've been on the roof of a house fucking with an antenna. I know why the save icon looks like that because I lived it.

Before USB was a glint in Joe Decuir's eye I battled with parallel ports, serial ports, Centronics ports, PS/2 ports (fuck you IBM), and game ports. I used to keep a floppy disk with backup copies of all my computers drivers for when shit hit the fan. I've had to manually configure IRQ ports. Copyright protection used to mean looking up a line in the manual to type in or using a Dial-a-Pirate:

https://youtu.be/7tMGNgofeyU

Not connecting to the internet. Fancy programs used a thing called a 'dongle' to prevent piracy. No, I'm being serious. That is what it is called.

Troubleshooting the office copier that has brought everyone to a standstill? That doesn't even work up a sweat. I can deploy that at the server level using a GPO, preconfigure and lock down the settings so no one can fuck it up, and even make sure the Mongos can see two entries when printing; one for black & white only and one for color, just to make it as simple as possible.

Comparing early GenZ to an early Millennial is laughable. I remember when Linux wasn't even a thing. I remember when Windows Updates weren't even a thing either. Shit, we used to reinstall our operating systems once a year just to keep the performance up. I've been using emulators and virtual machines to run different OS's and systems on their non-native hardware longer than you have been alive. I went to a vintage gaming show and bought a T-shirt there back when you wert still in diapers.

>>3008873
> Centronics ports
USB C connectors have 24 pins.
We’re back baby, we’re back!

File: 6A334C40-A7F4-433E-BF4D-E(...).jpg (1.28 MB, 2574x1687)

1.28 MB JPG

Needed to modify a w1099 temperature+humidity control module. By default its humidity regulation would vent all the hot air I’m building up and prevent it from getting up to temperature, so I had to ensure the humidity controlled load would only activate when the heater was off. Next, because this is a 12V module and I want to run it at 24V, I had to add a 12V regulator, linear because I don’t have any 24V capable miniature buck converters lying about. Then that was dissipating too much power, so I swapped the relays for MOSFETs. I was going to have to add an additional BJT to invert the signal going to the FET gate, until I realised the 5V MCU outputs were capable of driving a logic-level power FET (IRLZ44N) by themselves. So I desoldered the onboard BJTs and ran a jumper wire to the FET gates. Hopefully this setup should work more reliably than the original relay design, I’m only switching 2A@24V, but there’s going to be a bit of heat nearby.

I might need to blob something in to mitigate vibration, but it feels pretty rigid.

File: NBR-253-Full[1].jpg (142 KB, 1800x1199)

142 KB JPG

I need some BNC to RCA adapters to use a PVM I picked up with more devices. Is stuff from Hosa (hosatech.com) worth a damn or is it just rebranded Chinese trash like everyone else's shit?

I've decided I want to spend more time learning the actual theory/math behind this shit. I have tinkered and got things to work on my breadboard, but I don't like having such a big knowledge gap.

For people who know their shit (even on a smaller scale like just small circuit stuff), how did you best learn it?

>>3008956
University courses that recorded on YT, edx and mit. And also free pdf uni textbooks from web. That's how I did it

>>3008956
Once you get ohm’s law, KCL/KVL, complex impedences, and the common components (passives, transistors, op-amps), you should have enough of an intuition to get by self-taught. When you need to learn how to make a good feedback loop, you look into control theory. When you need to do some dac or adc work, you look into nyquist sampling theory. When you need to handle high frequencies over significant distances, you look into transmission lines. That’s how I did it, having learnt the basics from a university course.

>>3008975
We can tell you haven't been to a university because of how you write.

>>3008956
A lot of the app notes and circuit descriptions have a “theory of operation” section, so just stop ignoring those. That way, if you’re using a component or IC, you’ll have a reason to go in-depth on things that you don’t know that was mentioned in those app notes.

I would like to see that wonderful photo of an ATX power supply turned into a benchtop supply. Barrier terminal block on top of the housing with voltages written in Sharpie.
Do any of you know what I'm talking about?
Please share it. Also do me a huge favor and have a great day. Thank you.

File: 1600639300334.jpg (759 KB, 2048x1365)

759 KB JPG

>>3009285
I have the pic, but my testicles are visible and I'll get banned with a rubber hammer by an unpaid intern if I post it. Just use your imagination.

Speaking of benchtop, I think a benchtop multimeter is preferable to a handheld. Especially if it has two independent inputs

>>3009297
Benchtop makes sense if you already have two handheld meters, or need it to do kelvin 4-wire testing or to have more than 4-6 digits. So, less than 0.1% of multimeter owners. That said, I want to buy some of those symbol Nixie tubes (Ω, V, A, F, H, and m, μ, M, n, p, k) and make a benchtop meter for that.

File: high side current sense.png (76 KB, 1437x621)

76 KB PNG

>>3009345

I got a pretty good accuracy with my 0-500mA high side current sense circuit. Should certainly be accurate enough for my project (fan controller thingy). Schematic coming up.

File: High-side current sense B(...).png (74 KB, 1435x985)

74 KB PNG

>>3009349

>>3009344

>>3009345
I don’t get the principle of operation. I see circuits like this inside op-amp internal schematics, what is that arrangement of transistors called?

Found an old router. What parts can I salvage from it? Would like the antennas but they're probably going to be too big for my laptop's wifi card.

File: breakthrough.jpg (366 KB, 960x946)

366 KB JPG

>>3009448
Turn it into a flash drive NAS that holds OS install media, drivers, and docs.
or install Doom. Once you start salvaging components it will be useless, so make it do something useful.

File: image.png (20 KB, 325x229)

20 KB PNG

>>3009575

Anybody got some good recommendations to study these topics?

>>3009349
AFAIK it has two current mirrors which force the currents on both legs of the circuit to be equal. R3 adds a small offset to the output voltage so it would start at 75mV when the load current is 0mA.

>>3009538
I'm familiar with the normal 2 transistor current-mirror, but the 4-transistor doesn't seem to be a trivial extension of that. Even then, all the 4-transistor current-mirrors I see people using have 4 NPNs, not 2 NPNs and 2 PNPs. It doesn't seem obvious how they'd be different, but that just makes me more curious why that choice was made.

Also, how a current mirror makes for a good differential amplifier, especially one with a single-ended output relative to ground or 12V, isn't obvious to me either. I guess a long-tailed pair would be less linear, and maybe the current mirror is much better at being immune to common-mode noise? But I'd want to test what fluctuations in that 12V rail do to this circuit, seems to me that the current through both brances of the mirror is fundamentally dependant on that 12V rail, and so any voltage measured will be proportional to the 12V rail (e.g. Vout = k * V1 * I * (R1 + R2)).

>>3009504
ETA-International study guide.

>magic smoke escapes the power tool
>look inside
>0.2uF 5 connector capacitor from west germany
they sure don't make em like they used to
what now?

>>3009640
It's a dual capacitor meant to reduce arcing thus extending brush life, as well as noise filtration. You can still run the tool without the cap, but you should check and/or replace the brushes first.

File: gyrator.png (24 KB, 1723x707)

24 KB PNG

>>3009837

>first time ever building something that uses a gyrator
>it actually works
amazing
for those wondering it's for the supply side of a bias tee. I want to present >1 MOhm impedance to a 700 Hz signal coming back up the coax. the gyrator simulates a 1 kH inductor
I had to add a diode to bring the overshoot under control during startup, but other than that it's been behaving excellently
now to solder this up and put it in a nice 3D printed box

>>3009809
>>3009810
>>3009811
Look at this butthole talkin shit.

>>3009793
What sensor needs 1kH? Is the output signal really low frequency?

>>3009837
I could bump the frequency up with the current choice of opamp. but I also need to spin a new design that uses a less power, because 50 mW is way too high. the most extreme is MAX40018 which only pulls 1.6 µA quiescent current per channel, but it has a pathetic gain-bandwidth product of 9 kHz. the AD8505 is more reasonable, but still only 95 kHz. and it uses a whopping 27.5 µA per channel

>>3009837
oh and I designed the sensor myself if that isn't clear

>>3009844
What kind of sensor is it? What aspect of reality does it measure? Temperature? Radiant intensity? I've only seen bias tees used in RX antennas.

What can I make with a solenoid water pump salvaged from an old steamer?

File: driver.jpg (1.31 MB, 1486x2230)

1.31 MB JPG

>>3009940

>>3009868
not sure how much I can reveal at the moment. that said, here's the realization of the circuit
I had issues with the output sagging seemingly at random which I traced to leakage into the 10 MOhm series network. letting the resistor hang in the air, and drilling out all copper around the high impedance node solved the issue. all my multimeters are too low impedance to actually measure there

If you guys were to recommend ONE book that everyone interested in electrical stuff should read, what would it be? I'm talking extreme basics: what is current, voltage, amps, etc. Essentially something like an electric for dummies handbook.

>>3009911
Forrest Mims

>>3009907
Spray that circuit with some conformal coating to prevent atmospheric dust and grime from causing leakage problems.

File: 20260929_140445.jpg (1.9 MB, 2880x2160)

1.9 MB JPG

>>3009960

you wanna see what a motherfucker looks like?

>>3009953
Oof, did that failure damage anything else? Overvoltage to a CPU or RAM?

File: 20260929_152254.jpg (1.24 MB, 2880x2160)

1.24 MB JPG

>>3009963

>>3009960
luckily no.
i measured 8 ohms between atx 12v and gnd. i was scared it was going through silicon because it was low but not zero. so i removed the cpu and measured again: 8 ohms.measuring closer to ground zero was 2 ohms or less, measuring across the blown cap was dead short.

basic cleaning with alcohol actually removed the short, but the smug jackass in me wanted to clean it properly, so i put flux and solder and wicked it back out. the fucking short came back.

i added solder and wicked it out again and again till there was no solder left, but the short stayed. i was mad and disappointed,
the pads were gone, the nearby resistor came out, the mosfet underneath was starting to feel the heat. i hated myself for wanting to do it right, and mostly didn't understand where the connection was happening. i knew it wasn't the mosfet but where else.

File: 20260929_162144.jpg (2.19 MB, 2880x2160)

2.19 MB JPG

>>3009962
i ended up digging under the surface and severing the two sides physically. which removed the short. i double checked components against a boardview to make sure they were happy, and put it back together. it worked thankfully.

>>3009940
good idea. I even got some PCB laquer recently not that I expect dust to be a huge problem here since it's going in a box

File: IMG_20261001_100514~2.jpg (148 KB, 2382x878)

148 KB JPG

>>3010077 >>3010081 >>3010092

What to do with a dead graphics card

>>3010076
If the silicon is dead then it's trash, otherwise it's practice.

>>3010076
Might be some intact high bandwidth memory on it? That shit’s worth more than gold these days.

>>3010076
snip the edge connectors and sell them to a gold refiner. you'll probably want to collect at least 1 kg of them first

are you really supposed to just pray you don't bridge the pins when measuring a laptop charger ic? are there like, one sided probes available?

>>3010194
yeah it's a risk. you gotta use test points, and finer probes. i burned an ic inside a tv because of clumsy measuring.

>>3010194
What do you mean by measure? Where are you measuring?

>>3010194
There are SMD probing stations with needle-like probes held in articulated mounts. Usually designed around a microscope, they look pretty expensive but maybe you can get some aliexpress probe holders to use alongside a PCB holder. You can also get micro gripper probes, they have little claws that extend like a pearl-clutch, and are small enough to grab a TSSOP pin without shorting to nearby pins. Without any fancy hardware, maybe you should try soldering tiny wires onto pads you want to measure. You can solder them on while it’s not powered, and inspect the joint closely for shorts before you risk turning the power on.

File: IMG_20261003_112418~2.jpg (89 KB, 1174x899)

89 KB JPG

>>3010262 >>3010282

Do I need thicker wires for this one?

>>3010248
It's 16W, so 68mA minimum. It might have a terrible power factor, but if your wire can handle 0.5A you should be fine. In other words, you could use wire as thin as 30awg and you'd probably be alright, though I'd probably use something like 24awg. More important would be finding wire that's got mains rated insulation, if that means 10A mains wire then so be it. Though I'd want to use either very rigid wire (well supported >1mm strands) or very flexible wire (<0.1mm strands) if there's any chance that the vibration of the pump could work harden and embrittle the wire. Fastening the wire down tightly with cable clamps and isolating the vibration of the pump as much as possible is what I'd attempt.

Looking for some great needle nose pliers for soldering, anyone? Shopping from Amazon only.

>>3010248

power = voltage x current
so current = power / voltage
so current = 16W / 220V = 0.072A
meaning the thinnest wire you have rated for 220V is more than adequate

>>3010194
>bridge the pins when measuring a laptop charger ic?

if the cable is made using 2 wires side by side, then stick needles in the 2 wires and measure
the holes will close when you remove the needles
alternately, salvage a female connector from some old equipment

File: IMG_4997.jpg (25 KB, 700x700)

25 KB JPG

>>3010490

>>3010278
I recommend a set of small stainless forceps. They clamp onto the part or wire so you don’t need to provide constant tension with your fingers while soldering fiddly parts, and they provide more clamping force than reverse springloaded tweezers. Otherwise, anything with a return spring, thin jaws, and grippy teeth, pic related.

>>3010282
Watch out for power factor!

>>3010283
He’s talking about the IC, the thing on a circuit board, probably on the laptop board not inside the charger. Presumably he’s measuring more than just the output voltage.

>>3010092
> sell them to a gold refiner
Those edge cards have an expected plug-in cycle life of approximately 1 time.
For this reason, they’ve perfected getting the gold plating as close to 1 atom thick as possible. You’d need 10’s of thousands of these things to get a negligible amount of gold out of them. The cost of cyanuric acid alone far outweighs the amount of gold you you’d get.

>>3010352
ok just give all connectors to me if you don't want 'em

File: Capture.png (42 KB, 1222x643)

42 KB PNG

>>3010388 >>3010428

i'm trying to turn an open collector comparator into a latching signal. when the comparator output goes low, the latch goes high. the latch can only be reset if the comparator output returns to high.

this seems to work in falstad. as far as i can tell, "setup and hold" constraints don't exist for SR latches because they aren't clocked circuits? would the only edge case therefore be some sort of minimum pulse time for the circuit to latch?

https://www.falstad.com/s.php?s=zTQ17G

Is there any benefit to desoldering smd components and IC chips

>>3010386
after some more reading, apparently these do have setup and hold constraints. my "enable" signal (analogy to the clock in a D latch, i guess?) will always be pulled up, and so i am guaranteed to meet the setup and hold constraints. is that right?

File: micro currency.jpg (210 KB, 640x640)

210 KB JPG

>>3010392

>>3010387
>any benefit to desoldering smd components

when society collapses, these will be used as the standard exchange currency for hookers and dope

>>3010391
Them some funny lookin' chips ya got there, boi.

>>3010387
Yeah, it’s the same benefit you get from desoldering thru hole components.
Most commonly, you can get cool regulators, tvm diodes, polyfuses, current shunts… things like that. I make things like oscillators with discrete transistors, I got about 30 of them from an old sony TV… used 2 of them to make the speed control signal for a delta fan.

>>3010392
it's gay-ai. the uncanny wall is proving to be impenetrable.

>>3010386
So you want it to be manually reset, but only for this to be able to happen when the comparator isn’t still trying to set it? Yeah, an SR latch with a tiny bit of combinatorial logic on the input is the way to go. It doesn’t need to be a gated latch, that enable input is superfluous. Then you can DeMorgan your circuit to make a simpler one out of NOR gates.

While flip-flop ICs already have edge triggered clock inputs, I can’t think of a situation where this would be overridden by the comparator signal. Maybe a JK flip-flop, with K tied high and J tied low, reset into CLK, and the comparator going into an asynchronous SET pin? I’d want to browse some datasheets to find one with asynchronous set/reset pins, and where the async set being high overrides the resetting clock input.
I guess the same idea would apply to a D flip-flop with async set/reset, if you tie D low.

>>3010387
To replace dead chips, obviously.
Sometimes a chip will have firmware programmed onto it, or it will be otherwise hard-paired to work with specific other serial numbered components, so you’ll have to desolder it from a bad PCB and transplant it on the replacement PCB.
And some chips are just valuable enough to be worth the labour even for reselling.

File: Screenshot 2026-10-04 230713.png (1.05 MB, 876x1027)

1.05 MB PNG

Why use any of those convenient pre-made LCD adaptor boards when you can spend the time and money to make your own, using much bigger PDIP ICs?

File: Screenshot 2026-10-04 233903.png (1.63 MB, 845x1206)

1.63 MB PNG

I think it was last thread someone helped me with the layout of this awful Frankenstein board with its mix of THT and SMT components

>>3010314
With serrated or smooth jaws?

>>3010490
>serrated or smooth

you need both for diff jobs
serrated is a better choice 90% of the time, tho

>>3010497
Different anon, what do you use smooth jaws for? I find their low gripping power basically makes them useless. If someone made pliers with grippy vulcanised rubber or hard silicone jaw covers though, that could be handy.

>>3010519
>what do you use smooth jaws for?

straightening things like wires, component leads, and chip pins without damaging them
also forming wires and sheet metal into shapes, again without scratching
adjusting links in jewellery chains; metal glass frames; grabbing tiny plastic lenses

>>3010520
I don't see how it matters for component leads and chip pins. Insulated wires, especially with soft insulation, I can understand. Same for jewellery, glass, optics, etc.

>>3010521
nta, but first thing that would comes to mind are gold plated pins for example. They are not gold plated for fun, nor for the great conductivity of gold, but because you don't want them to corrode.
If you scratch those you could drastically reduce their life span. Also silver wire or magnet wire. The point is to being able to grab and bend stuff without scratching thin coatings for electrical reasons and not just for having a pretty surface finish.
>just my 2 cents

>>3010524
Ah I see where you’re coming from. If I was going to solder to them anyhow who cares, but if it’s part of a connector then yeah it would be valuable to preserve the coating. Though even then, the corners on some flat jawed pliers could bite into the coating, especially thin magnet wire. I’ll see if anyone has made rubber jawed pliers or forceps.

>>3010527
>rubber jawed pliers or forceps
Just get some round-nose pliers.

Has anyone here gone through the process of lab testing a design for CE compliance? Do you have any advice?

Can I post here without being bullied? I've always like the idea of making little things with components. I've just no idea where or how to get started.
It's mostly because I have a little device and the button stopped working. When I took it apart it the little button/switch inside had snapped from the solder. I bought a replacement switch, solder iron, all of that shit...and that was pretty much it. It's sort of like a tactile switch, but the more rectangular kind and it's absolutely tiny which just makes it more difficult for someone like me.

Where do I start bros? What do I practice on first?

Pins 4 and 5 have no continuity on my VGA cable. I'd get a new one but I also want some learning experience. Where should I begin.

>>3010565
> little device and the button stopped working
Secret Plot Device.

>>3010573
Do you realize how rare it is to have a broken VGA cable?

>>3010573
>Where should I begin

for a clean-looking result, you'd need like $50 in parts and tools
but a used one at the thrift store is like $2
for a learning experience, i'd suggest convincing your sister to experiment
they're the best teachers

>>3010565
>What do I practice on first?

youtube videos

"Basketball players who only visualized shooting improved their free-throw accuracy by 23%, nearly matching the 24% improvement of players who practiced physically" -- Alan Richardson, 1960

Any recommendations for an inline mains plug power meter? I want a compact one to chuck in my toolbox to help me troubleshoot electronics. Having no screen and a Bluetooth app instead would be fine, but I’m looking for one that accurately measures power factor on nonlinear/ switching loads. My current cheap one is measuring 80W from a 50W supply, and browsing about it’s hard to tell which ones are measuring phase offset for reactive loads, and which ones are actually continually multiplying the instantaneous voltage and current values and taking the rms of that.

I’ve considered making my own, if it’s just measuring current then it’s pretty trivial to cram a current transformer inside a piggyback plug and run wires to my DMM. That would be useful for 99% of uses, so I’d consider making one of those for on the road, and having a nice Hopi meter or equivalent at home. Adding power-factor would require a microcontroller, I’d probably take a series resistor (plus TVS) from the mains input and run it into a little audio transformer to make a power supply and an isolated voltage measurement waveform. Maybe I could cram it all in the housing of a smart-plug. But the coding sounds like a pain.

Any recommendations? AU plug btw.

>>3010565
>snapped from the solder.
If it was just a cold joint it's probably an easy fix.
If the pad lifted it could be significantly more difficult.

>>3010573
IIRC many VGA cables don't have all wires to begin with, if you haven't yet it would be good to make sure they're meant to be there before assuming it's broken

File: 1791331431792180.jpg (78 KB, 960x739)

78 KB JPG

>>3008666
I love how Gen X doesn't even exist at all. Either too tied up around work or collecting their Funko!Pops. Strange nobody even mentions them ever.

>>3007000
are there any off the shelf solutions for low voltage ideal diodes? most ideal diode controllers use the bus voltage to power the IC, meaning it won't function with low voltage. i need it to function with a bus voltage as low as 0.7V.

I'm starting a community on making our own RAM. It doesn't matter if it's 512mb or 4gb. If my RAM sticks crap out I ain't shelling 100 bucks for 1 stick.

>>3010650
i guess i mean to say, other than LTC4353 which is pure ass rape at $12 ea.

Don't know anything about electrical work but it looks like my first project is going to be replacing outlets. Got almost all the tools I need: screwdriver (obviously), voltage tester, pliers. About to go to Ace and buy a set of wire cutters. Anything in particular I should look for?

>>3010670
The only tool you need is the screwdriver. You can test for voltage with any device that plugs into an outlet. A nightlight for example. Pliers are unneeded. Same with wire cutters.

Replacing an outlet laterally consists of turning off the breakers, checking the outlet with a lamp, unscrewing the faceplate, unscrewing the outlet from the box, pulling the outlet out, unscrewing the screw terminals to remove the wires, putting the wires on the new outlet terminals, tightening the screw terminals, placing the outlet back in the box, screwing it to the box, placing the faceplate on it, and screwing it down.

>>3010670
Get a square drive bit if your outlets will accept it (most do). Phillips sucks dick.

>>3010706
You mustn't have ever replaced backstabbed outlets. They get extremely hot under load and you end up having to cut off at least an inch of wire to get to clean copper and uncooked insulation when replacing them.

>>3010706
Like I said, I'm new to all of this. I watched several videos, one of which suggested having wire cutters handy in case the wires were inserted into the back of the outlet and became a bitch to get out (like the anon below you said). I bought a pair of Klein cutters as I've had good experiences with that brand previously. Unfortunately it was all for naught. I get to the house, plug in the wire tester, then go to the breaker box in the basement. The panel is not labeled (for the most part). Some switches ARE, with things like "sump pump," "microwave," other appliances. So I start at the bottom, flip, go back upstairs, check. I repeated that process for every single unlabeled switch on the panel (30 or so), and none of them killed outlet I was testing. By then I was tired from going up and down the stairs a hundred times so I got fed up and left. Later this week I'll go back with my wife and skip the marathon.

As for the actual replacement, I'm pretty confident I can do it. Every video I watched echoed your steps. Got the tools, got the replacement parts. In the meantime I'll just curse my FIL for not labeling the panel. It's funny because he was one of the most particular, organized people I've ever met. We're talking about a guy who saved everything. I was going through a desk in his office and found receipts for every single electronic/appliance/material/home good he ever purchased. Example: the Zenith 32" color TV set (with remote!) back in 1978. I found a notebook detailing all of the materials he bought to build the house. All this is to say that I'm surprised a guy like that didn't label his panel. Who knows. Maybe the paper slipped off or is in a drawer somewhere.

>>3010711
The way Phillips is designed had to be some guy fucking with us. The shape itself seems tailor made to make stripping the screw easier. What the fuck was he thinking?

>>3010711
>Get a square drive bit if your outlets will accept it
Which won't help with the faceplate or stab in connectors. You can generally release stab ins with a flathead unless they are melted in. You could get a multibit driver or just a second screwdriver but, eh. Robersons are very nice otherwise. Their only real problem is their heads are ugly. So, not for finish work that people will see but great for everything else.

>>3010713
>The shape itself seems tailor made to make stripping the screw easier. What the fuck was he thinking?
>>3010711
>Phillips sucks dick.
Agreed but there was a method to the madness. They were designed to be used in manufacturing with early powered drivers that had primitive clutches or no clutches at all. It was preferable for the head to cam out than either overdrive the screw into the material or have it snap at the shaft. Specifically, it would throw out the driver, which often cammed out the screw, instead of driving the screw further into the material and potentially causing bigger problems. This 'feature' of the Phillips was known at least as early as 1942. They caught on and became so entrenched that they still dominate even though better clutches and better head designs have been around for decades.

>>3010650
At some point, you just gotta use a comparator or long tailed pair to switch the MOSFET. Have you tried looking at the energy harvesting field?

>>3010651
You’ll be lucky to get 64M. What’s the plan, those share-a-wafer custom silicon providers? I want someone to make Optane again, or just some form of high reliability (single level) flash storage.

>>3010723
The worst part of Phillips is probably not the camming out, but the presence of JIS and PoziDriv and one or two other too similar looking bit types that people get confused. Additionally, some narrow screws in plastic enclosures use PH2, but are down holes thinner than the standard 5-6mm PH2 shaft. Had to go out of my way to buy 4mm PH2 and 3mm PH1 bits from aliexpress, 150mm long.
You’d ever have that problem with hex keys. Probably torx too, I think they’re usually not much wider shanks than the torx shape itself?

>>3010732
I stripped a torx screw on a smart plug a while back because it was like a T6.5

>>3010670
Turn off the breaker and follow the code where you live.

>>3010723
>Their only real problem is their heads are ugly. So, not for finish work that people will see but great for everything else.
I know it's a regional thing, but it's kind of funny to see this. Exposed robertson heads are everywhere where I live and nobody thinks twice about it. Though, faceplates are still often godawful flat for some reason.
While torx is in theory stronger, from what I've seen, robertson generally does better in less optimal situations (which is usually the case). Either one is way better than phillips.

>>3010732
>At some point, you just gotta use a comparator or long tailed pair to switch the MOSFET. Have you tried looking at the energy harvesting field?

pretending for a second that i had a charge pump or floating supply to address the highs side NMOS drive requirements, would this stand a chance at working IRL?

https://www.falstad.com/s.php?s=KN2O0k

i think i would need a linear solution (op amps) rather than hysteretic. initially i thought TI LM74610 would have worked as an off-the-shelf solution then i read the details of its hysteretic operation: it won't clamp reverse current until V(drain,source) > 20 mV. this is several amps of current flow for a low RDS_on FET, with a potential indefinite reverse current flow if the 20 mV threshold is never crossed.

What do they mean when they used to build gadgets from off the shelf components?

>>3010765
>high side NMOS
Aww, you can’t have low-side NMOS? And your current is high enough that PMOS isn’t an option? If your current is high, you’re probably best off using a dedicated gate driver with at least sn amp of source and sink current. You can probably find such a driver with a built-in charge pump, then control it from a comparator.

You don’t want something linear, you want your FET either hard on or hard off. Anything in between and it dies to ohmic losses or spirito instability. And op-amps are slow when used in saturation, though with some diodes you can mostly avoid this. You want both speed and minimal hysteresis, I guess you should experiment for yourself what parts and assemblies might be suitable for this.

>>3010740
> like a T6.5
That should be illegal.
Next they’ll “invent” T6.25 and T6.75

I was already through with torx, the screws simply don’t stay on the driver bit when you have to put the screw in a recessed box, then I end up cutting my hand open trying to fish out dropped screws in the box already overfilled with live wires and splices.

>>3010723
> phillips… preferable for the head to cam out
That is a myth. It was never designed to cam-out. It was designed to make it very easy to set the screw in the bit, and can be done in a factory when dealing with 1000s of screws. It’s self-setting. Any drywaller will tell you that, and it’s why they’re used for sheetrock (with clutches) because they don’t cam out without the clutch… especially on a new ph2 bit. But I can do 1 every second because I don’t have have to wiggle the bit back and forth to get the bit seated in the screwhead like all others.

Desoldering SMD components isn't worth it imho. Boomers were lucky they could get free components back then.

>>3010808
It's possible it's Torx Plus, but it's not like it would be easy to tell at that size.

>>3010821
> Torx Plus
Another torx standard? roflmao… i hope you’re pulling my leg here… that’s insane

>>3010818
do you not have a heat gun?

>>3010834
They introduce a new variation every time the patent expires. Gotta keep the royalties flowing somehow.

>>3010818
It is worth it if the chips are worth saving. CPUs, RAM, and flash memory chips come to mind.

>>3010834
There’s also Torx Ttap. And it’s also similar looking to Polydrive.

>>3010850
and the torxheads keep crying that the entire planet hasn't taken up the standard yet.
i wonder why?

File: IMG_20261009_180605~2.jpg (115 KB, 540x1752)

115 KB JPG

>>3010945

How do I turn this phone on without a battery? It vibrates when I hold the power button but there's no display.

>>3010856
Idk man, what voltage are you feeding it? It looks like that resistor goes from V+ to V-, but usually the thermistors are a third pin to ground. If you have a real battery, you could measure the voltages on its pins and compare them to your spoofing setup.

How easy is it to fix a TV? Is it as simple as opening it up and seeing something with a burn mark on it that needs replaced?

>>3010959
No. If the problem is something nonspecific, like it won't turn on, it can take hours for a skilled tech to figure out what is wrong.

>>3010959
Very easy. I'm not even trained nor can read schematic

>>3010962
Fuck sake. I think one of them might need the screen replaced because there's sound and no picture, but the other one just doesn't turn on.

>>3010964
Sounds dubious.

My next project is going to convert an old Samsung into a janky droid deck with a rudimentary controller or mini PC. Probably without a battery.

File: Inked20260905_072909_LI.jpg (1.99 MB, 2160x2880)

1.99 MB JPG

>>3011137

>>3010959
opening the back without breaking the panel takes: 5-10 minutes. longer if the back is clipped on and have to be pried off.
confirming power supply outputs correct voltages: 5-10 minutes
diagnosing why it's not outputting correct voltages: 5 minutes to never.
confirming if the backlight is functioning by detaching motherboard: 2 minutes.
confirming no short on main board by probing inductors: 1-2 minutes
fixing a shorted power rail by removing or replacing component: 10-20 minutes .. zero minutes if you use a screwdriver to knock a shorted cap
confirming correct voltage on all power rails: 5-35 minutes depending on how quickly you can trace lines and see the small text.
fixing missing voltage by replacing mosfet, or ic: 10-30 minutes
confirming no shorts on t-con: 1-5 minutes
replacing component on t-con: 5:20 minuets
replacing a COF chip: 30 minutes and about $15,000 for the tap bonding machine
replacing panel: 20 minutes if it's a matter of ordering a new one and wiring mainboard, psu and speakers ..etc to it

Just tried to solder a SoM onto a prototype PCB and shit doesn't work. Don't know if it's because I messed up the soldering or because the board itself is defective. Either way that's probably a few hundred dollars and weeks of time wasted to produce a new one.

>>3010791
i found this from TI, using a nice yet affordable comparator that i wasn't familiar with (TLV1805.) https://www.ti.com/lit/an/snoaa23a/snoaa23a.pdf

i think i'll try building the nmos version, although i still don't quite understand the point of the "anchoring" diode they use (DB2430100L), even after reading the description.


>The output diode D4 is used to "anchor" the output during light or floating loads. At light or no loads (less than reverse current threshold), there is a possibility the MOSFET could turn on due to the comparator offset voltage. The diode provides enough of a negative leakage to turn the MOSFET off.

>>3011015
Looks like a reasonable circuit. Ultimately there’s some amount of time and voltage slack in your circuit, the amount of current you’ll be seeing will determine whether the delay will cause that body diode to cook.

What’s the application? Solar?

If I power a 12V brushed DC motor with 48V or even more, but with current limiting to within its design current, will anything go wrong? Maybe I’d need a fairly conservative current limit, or a constant power circuit, but it seems reasonable to expect it won’t die by overheating if this is done properly. The speed will slowly creep up, to a much higher top speed than possible at its design voltage. Does high speed damage the brushes? Would the higher voltage cause excessive arcing at the commutator? Is there a fluid you can submerge a brushed DC motor in to prevent arcing, I wonder?

>>3011036
>What’s the application? Solar?
battery charging

File: 1791427336276991.png (62 KB, 806x804)

62 KB PNG

>>3011128

This logic chip can debounce 6 mechanical switch contacts

>>3011084
So can a 74HC14 with some caps and resistors, but neat to know there are dedicated ICs out there.

>>3010670
Following up on this. Got some time to go to the house and get into it. I got a whopping four outlets and one switch changed. Very small in the grand scheme of what needs done, but I'm feeling good about it, maybe even a bit proud. Never done anything like this before. One of them took me nearly an hour; what I THINK happened is the ground wire wasn't long enough when they initially wired everything, so they spliced another piece to it, which resulted in the box being extremely cramped. After I got it wired I was having a hell of a time making the outlet not be cocked to one side so I spent a stupid amount of time straightening it. In the end I was 95% happy with it (pic related).

Like anon warned of here >>3010711, they used the "insert the wire into the back of the outlet instead of using the screws" method, and they were indeed stuck like hell. I clipped them as close as I could and re-stripped them, then screwed them down. This is pretty tedious work but it's actually been very gratifying being able to DIY something and having it be successful.

This house has so many damn outlets. That room in my picture has nine alone, and it's only about 12' x 14'. Total for just the upstairs is probably 70-80 outlets. Will take a while but paying a professional I'm sure it'd cost me hundreds in just labor. After I finish this I will work on replacing the doorbell chime. Need to research that one a bit more first and figure out the transformer issue.

>>3010959
First step is probably determining if it's a TV nice enough to bother repairing.

>>3010975
What's the reasoning behind blurring the photo?

File: 20261010_142742.jpg (464 KB, 3000x4000)

464 KB JPG

>>3011429

>>3011135
Forgot the picture.

File: IMG_20261008_224023.jpg (395 KB, 3264x1468)

395 KB JPG

>>3011141 >>3011147

1998 sony handycam shuts off when flipping the switch to camera mode to open the lenses. It still has current but the entire thing goes off and the lights on the playback buttons go black. When switching to playback mode the entire thing goes back to life.

>>3011140
Also forgot to mention that it's hooked up to its own power adapter. Don't really have those specific sony alkaline batteries available.

>>3011140
It’s probably drawing too many amps for one reason or another. Dodgy motor, or dodgy motor driver. Maybe it requires the battery to buffer its power because the adapter can’t source enough instantaneous current. For a similar reason, maybe capacitors that handle inrush have gone bad.

Gonna take some nostalgic images with CCD artefacts?

File: series-122-relays-on-bric(...).png (2.22 MB, 2560x1440)

2.22 MB PNG

>>3011189

Debounce THIS

>>3011159
For what purpose does that exist
I know reed relays are used in signal switching, but that seems excessive

File: Capture.png (33 KB, 781x430)

33 KB PNG

>>3011199 >>3011207

i'm trying to layout a current sense resistor and current sense op amp. pic related is the optimal solution, but my current flow is in the opposite direction. this means the left side of the resistor has to connect to (IN+), and the right side of the resistor has to connect to (IN-). i don't want do deal with two sided assembly. what's the preferred approach?

>keep op amp positioned as the picture shows
>route one shunt trace on top of board
>route another shunt trace on bottom of board
>use vias on both traces to match impedance
OR
>rotate op amp 180 degrees
>route shunt traces on top of board
>route REF and OUT on the bottom layer with vias

i don't need the utmost accuracy or anything, i'm just wondering if one option is materially better than the another. FWIW it will be a 4 layer board and the middle 2 layers are both ground pours.

>>3011194
I don’t see why the sense input lines would need to be short, length-matched, or just on one side of the board. They’re low impedance signals after all. Current sense amps are only so fast (I’m surprised there aren’t low pass filters on the sense lines), and you need to have extremely fast signals for an extra few mm to matter. Vias to flip the input wires seems trivial and with no downside.

That said, flipping the part and running more traces underneath it seems relatively elegant. Are you using REF, or just tying it to GND?

File: semiconductor-diagram.png (91 KB, 1920x1400)

91 KB PNG

>>3011194
Some INAs may have alternative pin orderings available. Maybe check that first

>>3011189
Probably a switch matrix for some huge automated testing systems

File: IMG_20261014_080653.jpg (301 KB, 1468x3264)

301 KB JPG

>>3011340

Can I use whatever I solder on different projects

>>3011331
what are you trying to ask?
are you asking about harvesting parts from dead electronics?
you can use whatever parts you could possibly find, but knowing how to use them properly is something you have to figure out.

I wanted to replace LM3914s with LM3915s, since the former are unobtanium. So I looked into making an exponential converter to cancel out the logarithmic converter of the 3915. Using a PN junction, it’s definitely possible to do that, but temperature coefficients mean drifting by an LED or two within a normal temperature range. Then I looked into incorporating another diode in the voltage reference, to see if the tempcos would cancel out. But exp(Vin/T) / exp(Vref/T) does not cancel the two Ts. So that was a waste of time. So, discrete resistor ladder + comparators, or MCU it is.

>>3011138
Turns out not just one but two outlets in this room are linked to a single switch. Is unfucking this something an amateur can do?

>>3011429
Sure they’re not outlets that switch on with a light switch, for you to plug lamps into?

>>3011432
I might've used the wrong terminology but that's exactly what they are.

File: 2026-10-14-234803_1273x88(...).png (129 KB, 1273x880)

129 KB PNG

>>3011463 >>3011468 >>3011495 >>3011501

>>3007000
Im pretty new to pcb design, but ive made some pcbs and know a bit about electrical engineering. recently i have designed an ESC for a brushless motor, especially to be field oriented control capable. This includes an external "smart" gate driver, inline current sensing, and lots of feedback to the mcu. i took great care in choosing the parts, ratings, mcu pins with dma, and so many little things that i really want to have done perfect, but face difficulties making decision about the layout. I looked at a lot of examples but each time i learn about a new phenomenon and how to not deal with it. i have no idea if or how to deal with switching noise, how to wire my analog current sense to not be influenced by the switching noise and many similar examples. i have become stifled and don't know what to do; maybe this project is to ambitious for me, but i really find it interesting.

>>3011454
I'm ignorant of this and can't help, but I wanted to say I think your schematic is really cool. You drew that up? Awesome, pal.

File: converting switched socke(...).png (68 KB, 566x555)

68 KB PNG

>>3011429
>Is unfucking this something an amateur can do?

yeah
except it's not really fucked, it's a useful feature in a bedroom, for example
anyway, just do what the pic says
how do you determine which wire is red and which black, as they wont actually be those colors?
if you have no instruments, just use trial and error
be sure the breaker is off while mucking with it

>>3011433
Buy some lamps.

>>3011454
Cool project. Using a CH32 means I guess you’re not using existing codebases like SimpleFOC or VESC? Ambitious, but I’m not much of a coder so that part is what scares me. CAN and RS485 are interesting, what are they being used for? In-line current sensing is definitely easier mathematically, but you need to ensure your sense amps can handle the high common mode voltage noise. Maybe that voltage regulator could benefit from an anti-backfeed diode, assuming it doesn’t have one built-in.
I’d add the option to use one of those radial magnet hall sensors, the sort you put on the shaft of a motor to measure position with high accuracy.

For the board layout, because of all the power rails, I’d want to go for 6-8 layers. And act to minimise switching loop area as much as possible.
Also it’s cheaper to get JLC to solder the SMDs on your boards than it is to get a stencil with them.

>>3011454
>switching noise
minimize the area of the current loop where switching occurs. the area of the current loop directly correlates to the noise created because it acts like a big antenna. so, short traces between gate driver and FETs, FETs and power holes, uninterrupted ground layer underneath those traces.
for the 5V buck, maybe put a pi filter on it targeting 700khz, sim it, then tweak when you get the boards. kinda overkill though
switching noise is inevitable with an ESC, best you can do is mitigate with intentional placement and routing. there's only so much you can do without biting the bullet, paying JLC to build it, testing, and then going from there
>how to wire my analog current sense to not be influenced by the switching noise
follow the layout guide in the current sense IC datasheet. ground via fence around A_SENSE and C_SENSE traces. surround the IC with ground vias cause why not. maybe bias the current sense amps with a dedicated voltage reference IC/zener rather than 3v3 from the LDO, probably overkill though
really, a lot of the filtering for switching noise specifically will have to be done by the ADC. sync sampling to PWM signal, time sampling to bypass the majority of the switching noise waveform, etc.
>power path
there's no supply favoring, so if USB and VM are present, you're probably gonna pull current from both. not necessarily bad, you wont backfeed either, but considering how your 3v3 rail is your vref for both your ADCs and current sense amps, that'd make me a little nervous. a cleaner solution would be a power path selection IC like the TPS2117, which can cleanly cut off one or the other

otherwise, the schematic looks very well thought out. you get a gold star
add some testpoints on analog/switching signals so you can probe when fine tuning the ADC firmware. maybe TVS on the various connectors and the motor holes if you want extra brownie points

>>3011454
>>3011495
>your 3v3 rail is your vref for both your ADCs and current sense amps
I'd definitely want to separate out an AVDD node, maybe with a small ferrite bead or just an RC filter on it to decouple noise from the DVDD rail. Splitting off the AVSS pin from the MCU would be good too, though just as an excercise in star-grounding, in other words preventing power currents pulling the ground reference of the ADC and INA amps by using a net-tie. I'd also want to add an external voltage reference that could be used both for the ADC and the current sense amps, though I don't know if that's possible with the MCU he's chosen.

>180nC total gate charge
That's a lot, what kind of frequency are you going to PWM them at?

>>3011501
>Splitting off the AVSS
>star-grounding
horrible idea. do not do that. splitting ground planes always creates more emi problems than it fixes. the only reason you ever do it is for HV isolation, which is not necessary here

>>3011502
You'd keep the main ground plane intact, just ensure current can only flow between the current sense amps and AVSS before it gets to the main GND net. If current to the gate driver could drop voltage between AVSS and the amps, then that's an offset you'll measure. Same reason you split off AVDD.

>>3011503
>just ensure current can only flow between the current sense amps and AVSS before it gets to the main GND net
so route the ground? that's hardly an improvement. any hole in the ground plane cannot have any traces going over it, especially ones with AC, since that'll cause undefined return paths, which then creates large area current loops, thereby nuking your emi/emc
the far better option is to be smart with your placement so large currents never even flow through the ADC section of the board. the current sensing voltage readout can dogleg off the main gate driver -> FET -> motor hole chain and no high current will flow under it since it'll be an AC signal that has a current return path that follows the signal trace.
as long as the current sense output trace does not go directly over or under the high current components of the board, and the ground plane under the trace is not interrupted, everything will be kosher and work just fine.
do not route your grounds, do not put holes or cutouts or whatever else in a ground plane.
everyone uses solid, unbroken ground planes for a very good reason. you dont stray from that unless the conventional ground plane has been proven to cause EMI issues and you know exactly what you're doing. it would be completely unnecessary to try to do some weird star ground fuckery in this application

>>3011532
Use a solid unbroken ground plane on one layer. Then on another layer, add an analog ground pour that goes underneath all analog traces. Connect them together at a single point near the MCU. I call it a 3D star ground.

>>3011468
>CH32 means I guess you’re not using existing codebases
I initially got interested in this because of the code and control theory and really want to try to implement field oriented control for BLDC myself. Ive found lots of resources like visualizations and libraries, so I think I can do it.
>CAN and RS485 are interesting, what are they being used for?
Well they are pretty common in robotics, especially CAN. The CH32V303 supports it and i had spare gpios, so this increases the usefulness of my ESC. There are also sensors that communicate in RS485 and I saw it on an ODrive ESC.
>but you need to ensure your sense amps can handle the high common mode voltage noise
They are specifically for current sensing and im pretty sure they ll be ok. I feel like im missing something because i have seen inline sensing on very few open designs. I also need very accurate ADC measurements as the 100V/V gain means its pretty senstive, something like 20mA (assuming 80A range) per bit - I think this is a bit questionable.
>I’d want to go for 6-8 layers
How should i distribute the layers? I obviously don't need a whole layer for high voltage, so Ill designate a specific area for it. As far as I know the power connectors should also be near the mosfets to minimize transients due to trace inductance?
>Also it’s cheaper to get JLC to solder the SMDs on your boards
All my parts are jlcpcb/lcsc. Ive chosen most passives and minor ICs like the CAN and RS485 transeivers to be jlc prefered extended so they don't charge extra for the loading fees. I actually have 2 BOMs, one for jlcpcb smd assembly, and the other for lcsc so i can solder parts like larger capacitors and connectors as the loading fees can become quite steep. I chose SMD GH JSTs because I saw them on the ODrive but I maybe I should have chosen some other type of through hole JSTs.
>radial magnet hall sensors
Ive seen them on ODrive controllers, but i don't know if the extra assembly fee is worth it.

>>3011495
>short traces between gate driver and FETs, FETs and power holes, uninterrupted ground layer underneath those traces.
Yeah, thats what I though. I am using the TI Application Note and example layout in the gate driver datasheet for everything related to the actual switching.
>for the 5V buck, maybe put a pi filter on it targeting 700khz
Wouldnt the 3v3 LDO smooth out the transients anyway? I obviously don't know too much about these things, so i am again relying on the application note for the buck.
>switching noise is inevitable with an ESC, best you can do is mitigate with intentional placement and routing
And the problem is that good routing and placement is the thing I know the least about (along with proper grounding)
>a lot of the filtering for switching noise specifically will have to be done by the ADC
I read an application note about ADCs and put some low pass filters but I mostly pulled the values out of my ass. Ill need a good state observer / kalman filter along with proper sampling time in sync with PWM to make the measurements reliable. Ive already made sure the timer that generates the PWM can be used to trigger the ADC. Also, as I said im not sure at all about the amp gain of 100V/V and the measurements may end up unsalvageable.
>ground via fence around A_SENSE and C_SENSE
does the via size or spacing matter here?
>dedicated voltage reference IC/zener
shouldn't the LDO be pretty stable and quiet?
>a cleaner solution would be a power path selection IC like the TPS2117
I agree, but I don't know if its worth it for the extra loading fee for jlc. At this stage i think ill add a little header to manually cut usb 5v to the LDO.
>add some testpoints on analog/switching ... maybe TVS on the various connectors
yes, i must have forgotten
>you get a gold star
thank you

>>3011501
>>180nC total gate charge
That seems to be a mistake, thanks for pointing out. I guess I would use frequencies definitiely between 20 to 45kHz as thats what Ive seen in libraries like SimpleFOC and in the gate driver application example. The example TI FET is CSD18536KCS, but it seems to be wayy too expensive in lcsc and I can't find a proper replacement. The best I found is clones of the CSD18540Q5B and they have just under 100nC total gate charge.
>ADC and the current sense amps, though I don't know if that's possible with the MCU he's chosen.
The CH32V303 has an ADC voltage pin. Ive thought about it but it increases my changes of fucking up the routing and grounding.
>do not route your grounds, do not put holes or cutouts or whatever else in a ground plane.
understood. Ive read some articles about "proper grounding" and became more confused. I wont fuck with this and leave a normal ground plane
>to be smart with your placement so large currents never even flow through the ADC section
Thanks a lot for all the feedback. I learned a lot and ill continue working on this ESC with more motivation.

>>3011559
>Ive seen them on ODrive controllers, but i don't know if the extra assembly fee is worth it
AliExpress module, wired up to the MCU on the ESC via cable. Unless you were planning on mounting your ESC directly o the motor.

>>3011561
>Wouldnt the 3v3 LDO smooth out the transients anyway
Check out its datasheet, most voltage regulators specialise in ~100Hz rectified mains ripple rejection, and aren’t very good up at 100kHz SMPS frequencies.

>The CH32V303 has an ADC voltage pin
Unfortunately there looks to just be an AVCC pin, with no additional AREF pin. If it had an AREF pin, you could use a precision voltage reference IC, and feed that and to both the ADC and the INAs, that way you could be more sure that 2048/4095 on a 12-bit ADC would be exactly 0A. But even without that, some DC offset and low frequency drift is very solvable in software.
>Ive thought about it but it increases my changes of fucking up the routing and grounding
A simple 10 ohm RC filter going to your AVCC and maybe INAs is worth considering. Or even another independent 3.3V LDO. The INAs have PSRR, your ADC only has that if you’re not using 3.3V as your reference. If you want to switch the MCU to an internal reference, I’d want your INAs to at least be roughly centred around that, so adding an extra footprint or two around those two REF pins would better allow you to keep signals within the ADC’s full-scale range.

>>3011559
>How should i distribute the layers?
signal ground power signal for 4 layers (probably enough). signal ground power power ground signal for 6 layers. powers can be swapped for signals, you shouldnt need more than 1 power plane for this desu. just make sure that there's at least 1 ground layer adjacent to every non ground layer
>20mA (assuming 80A range) per bit - I think this is a bit questionable
it's probably fine. i know nothing about ESC design but an amp of accuracy would be more than enough to make me happy @ +/-40A
>solder parts like larger capacitors and connectors as the loading fees can become quite steep
imo either have them assemble it all or do it all yourself. solder stencil, hand placing everything, and putting it in the oven is far easier and faster than manually hand soldering random SMD components. then again, i dont care about cost so i'd happily spend a couple hundred bucks extra to get fully assembled boards rather than having to waste an afternoon populating them
make sure you set thermal reliefs properly for the hand soldered components, you will hate yourself if you dont
>I chose SMD GH JSTs because I saw them on the ODrive but I maybe I should have chosen some other type of through hole JSTs
GH is fine, buy a kit with pre-crimped wires and connector housings

>>3011561
>Wouldnt the 3v3 LDO smooth out the transients anyway?
yeah, it's fine as is. if you cared about radiated emissions and your 5V plane was large, then you'd might want to filter on the output of the 5V, but it's completely unnecessary for this sort of board
>the problem is that good routing and placement is the thing I know the least about
good routing is short, direct routing and good placement enables good routing. put your noisy switching stuff on one end of the board and sensitive stuff on the other. sensitive traces go away from noisy ones. it doesn't have to be perfect, just use your brain and try your best
>along with proper grounding
proper grounding is a single, uninterrupted ground plane and vias to everything that needs it. you can ground fill the unused parts of the rest of the layers and via stitch if you're ok with thermals making rework harder.
several layers that are ground planes is ok, but the board should only have one ground net. split ground planes should only be used when galvanic isolation is needed for safety reasons, and this is not an application that needs it. dont look into the void that is "grounding strategies", it's an abyss of horrible ideas like >>3011545 that over complicate things and make your board perform worse.
if you cant help yourself, altium makes great guides (such as https://resources.altium.com/p/what-return-current-path-pcb) on grounding theory and noise mitigation in mixed signal applications. once again, it generally boils down to "use a ground plane and dont worry about it"
>does the via size or spacing matter here?
not really, pick something that's within JLC's constraints. at several hundred kHz, the wavelengths you care about are far too long for via spacing to matter
>shouldn't the LDO be pretty stable and quiet?
ehh, it's probably good enough. as long as your ADC reference and current amp bias are the exact same voltage, i don't think you'll run into issues caused by a less accurate vref.

>>3011585
There is literally nothing wrong with running an additional analogue ground plane on another layer, besides needing that room in the first place. So long as it covers all the relevant signal return paths.

File: 20261016_153128.jpg (647 KB, 3000x4000)

647 KB JPG

>>3011754

Back again with another outlet question. The swaps are going pretty well. Got several rooms finished. The current room I'm working on is wood paneling over drywall. One of the boxes seems to be pushed further back than the other couple I've done so far in that room. This makes it so I can't install the switch because it goes so far into the wall that I can't mount the plate and still be able to flip the switch. Pic related shows how far back into the wall the box is.

What's the best solution here? I'm a complete amateur. Is there some sort of spacer I can put between the switch bracket and the box so that I can still screw it down without having to push it back into the wall?

I have the motor but nothing else from a crappy 400w 3 phase wind turbine. Can i run the output through a 3 phase to DC rectifier then use that as the input into a solar charge controller assuming the DV output is within the limits of the charger?

Thinking about getto rigging a water wheel.

>>3011678
I assume the MPPT algorithm might take issue with a pulser input, so I’d try to filter it with a bunch of capacitance, though I guess you can try without that.
You may also find that the voltage range is too restrictive, most MPPT controllers are step-down only, so if your turbine makes 60V at maximum speed and you have a 24V battery, when the speed drops below 40% it won’t be able to gather any power from the turbine at all, as the voltage will be below 24V. Turbines seem like they’ll have much more voltage variance than solar panels. If your situation is more like 90V down to a 12V battery, I think you’ll be fine.

Also consider making an active clamp circuit to short the windings or dump the power if the voltage ever gets too high. MOVs might be fine if it’s just spikes to worry about.

>>3011682
>Turbines seem like they’ll have much more voltage variance than solar panels

Wouldn't it be less if the turbine was being driven by water instead of wind? I'm talking about converting the wind turbine into a water wheel.

Basically we have a massive roof with a 20 foot drop that we are going to put a tube gutter on, the amount of water that comes off it on rainy days is more than enough to spin a 4 foot bucket wheel.

Since the only time we have solar issues is when it rains it seems an obvious solution. The charge rate for a 400w turbine compared to the size of the battery bank means it would never stop charging and it isn't wind driven so a dump load probably isn't needed.

Small wind controllers/chargers aren't too expensive, i suppose i could get one of those but i already have spare solar chargers and a rectifier somewhere.

>>3011684
I guess that seems reasonable. If the head is constant the speed would also be pretty constant, instead of something like an undershoot water wheel. Definitely worth a shot at least.

>>3011661
Look up 'electrical box extender'.

https://www.youtube.com/watch?v=UH-8qv-qlK8

>>3007000
>3007000

File: IMG_7708.jpg (406 KB, 2046x1536)

406 KB JPG

>>3011907 >>3011908 >>3011918 >>3011919 >>3011927

I'm trying to replace a USB-A plug. One of the holes won't clear. It's like there's some kind of jam in there or something. I've tried wick, solder sucker, pushing a solid core wire through, pushing a fine tip iron through. Nothing works.

It's actually two holes that were blocked. Only one I managed to push a solid core wire through. The bottom left one remains completely blocked. Nothing is working.

Is there anything else I could try?

>>3011902
The solid core wire actually went through but the solder remained, or is it blocked? If it didn't go through, I can only suggest a stronger wire. Maybe a micro-drill-bit, whether pushing through while molten or to drill through it while solid. More flux, more fresh solder, being careful not to overheat it.

>>3011902
If flux and heat really aren't clearing it I'd probably carefully try a very small diameter drill bit. It kind of looks like it just needs more flux though.

>>3011902
Take it into your garage, heat the joint, and blast it with compressed air from your lungs through a drinking straw in your mouth.

>>3011902
Flux then solder wick while pulling with long nose pliers

>>3011902
We usually use stainless steel needles (medical ones) of various sizes and we blunt the tips somewhat. The stainless doesn’t stick to the solder.
…that’s usually for very small stuff… but usually you want a fine-weave solder wick… yes there is different kinds.

Also clean that gunk off with a swab and 99% isopropyl before attempting again. Sometimes you have to add a bit more solder to get the wicking action started… it looks a bit “dry”

I've found a new lease on my broken stuff. Instead of throwing them away, I just remove everything and add them to my SMD component box.

File: s-l960.jpg (202 KB, 960x960)

202 KB JPG

>>3011995 >>3012022 >>3012027

>>3007000
anyone ever buy/use one of these """pure copper""" pin heatsinks? are they actually made of copper -- or is it some fake bastard metal that performs worse than copper or aluminum?

i'm trying to heat sink ~12 W from a TO-247 package. this https://www.digikey.com/en/products/detail/fischer-elektronik/ICK-S-25-x-25-x-6-5/26281936 would appear to work but there's no way in hell i'm spending $5 each.

File: screenshot-2026-10-19-130524.png (141 KB, 473x408)

141 KB PNG

>>3011981 >>3011982 >>3011993

>>3007000
>Mims III, Getting Started in Electronics
Does a version of this book exist with a font for adults?

>>3011975
>a font for adults?

the series was called Engineer’s Notebooks, and each sold over a million because that quirk made them accessible to noobs, reformed strippers and other debauchees

>>3011975
>just go to radio shack and get this
Was it really that easy back in the day? I'm often inactive here because my time is spent waiting for shipping from China.

File: 813633EC-A566-4866-9C20-A(...).jpg (222 KB, 766x940)

222 KB JPG

>>3012027

>>3011975
> font
It’s his actual handwriting.
Back in they day, people knew how to wield a pen.
Do likewise, gents.
Early editions of Frank Ching’s books used his handwriting which, in later editions, became the tekton font after digitizing his handwriting.

>>3011973
Copper is still not that expensive in great scheme if things. If you want the best, use 925 silver but plate it with copper and nickel.
I reccomend you over-spec the heatsink, and then make sure you test it (regardless of where you got the heatsink) in any case.

File: 20261019_180608.jpg (133 KB, 739x709)

133 KB JPG

>>3012008

is this a retarded idea? i need to branch from that fuse holder to 3 GFCIs, but the bar clamp is only big enough for a single double ferrule

>>3011998
Probably fine, but if I was an inspector, I’d probably want to see a pigtail, splice, or two wires crimped into a single O lug

>>3011973
>https://www.digikey.com/en/products/detail/fischer-elektronik/ICK-S-25-x-25-x-6-5/26281936

Obvious question: Why TF are you even looking at that style of heatsink when it's completely wrong in both construction and size for a TO-247? Also, for 12W you basically "need" either active cooling or a conductive case that you can dump heat into, unless you can spare a bunch of space in your enclosure for a fairly large heatsink and can allow for plenty of passive ventilation.

The $5/ea price is because you're buying a German product from DigiKey. It's trivially easy to find purpose-made heasinks for all the common packages. TO-247 sinks run like $0.50-$2.00 depending on how big you need them to be.

https://www.ebay.com/itm/278352887707
https://www.ebay.com/itm/224125890777

>>3011973
You’re usually better off paying for a bigger heat-sink in aluminium than a smaller one in copper. Or a smaller aluminium heat-sink with a fan. The metal usually isn’t the limiting factor as much as its contact with the air is. Maybe if you’ve got forced air through a heat-sinking duct, AND space is a major constraint, copper will be a better option. Like in laptops and mini PCs. They often use heat-pipes too, thats something to consider using, I saw DIY-Perks using some custom-bent heat pipes in one of his big flashlight builds.

>>3011993
That’s very nice.

>>3011927
A sewing needle did it. Thank you, guys.
The holes aren't fully clear, but enough for the pins to fit. Like there's some roughness around the inner cyclinder surface. Did the through holes deform when I pulled the original connector or something?

File: Screenshot 2026-10-19 190458.png (1.51 MB, 907x1012)

1.51 MB PNG

>>3012082

spending hours revising my PCB layouts to maximize ground plane continuity only to go online and see people proudly uploading stuff like this

>>3012022
>>3012027
i need a thermal resistance <10 C/W and i have a lateral space constraint of 25 mm.

i could bolt this to a TO-247 by drilling a hole through the middle gap between the inner 2 fins. https://heatsinkonline.com/products/1496-wide-heatsink-1661as the spec quotes 3" (76.2 mm) at 2.73 C/W. for a 25mm section, that's 8.3 C/W for a 38mm x 25mm surface area x 50mm height.

compare this to a pin style heat sink, like what i linked: a 25mm x 25mm surface area x 7mm height (much smaller) and 5.8 C/W thermal resistance (materially lower.) i'd need to use thermal adhesive so add 1 C/W to that, but it's still 20% better while being much smaller.

>>3012050
It might work, but consider a 25mm thick blower fan?

File: circuits dont care about (...).jpg (153 KB, 640x640)

153 KB JPG

>>3012104

>>3012048
>spending hours revising my PCB layouts

So let them stand, these two in sight:
Beauty brings delight, function gets it right.
For what looks good may not suffice,
And what works well is worth the price.

how good are electrets from high end vintage video - cameras?

>>3012082
yeah well that's just, like, your opinion man

File: 1v5LedFlasher (4).gif (4 KB, 315x240)

4 KB GIF

>>3012169 >>3012204

Hey, this should be rather simple. I want to modify this circuit to only turn on at night to hopefully extend battery life. I'm thinking another PNP + LDR?

>>3012157
Perhaps a solar cell and a mosfet with -1 V gate threshold. Most (effectively all) nightlights using a LDR actually shunt the light off during the day using *more* power.

>>3012169
A photodiode alone would work to switch a FET.

File: LM3909 with D-cell = 12 y(...).jpg (44 KB, 699x392)

44 KB JPG

>>3012157
>hopefully extend battery life

you need an LM3909 with a D-cell
this guy claims 12 years of flashes
i got about 3 on my fake-car-alarm flasher
LM3909 is hard to find, but a 4-transistor equivalent that claims longer life is out there

Anyone know if there’s a way of measuring real (not reactive) AC power with a digital multimeter? Making a ring diode mixer is possible I guess, but could be bulky and would be tricky at 50Hz. I want a compact little device I can plug into mains and measure power from. Just being able to measure current using a current transformer is mostly what I need, but if I could measure power factor too that would be ideal. My current kill-a-watt clone isn’t trustworthy.

>>3012220
Turns out mains current transformers aren’t common, nor very feasible to wind your own as they need pretty high-end core materials. Guess I should go for the shunt resistor method and a microcontroller, or maybe a dedicated IC. Shunt resistor and an audio transformer might be measurable by my meter while being isolated.

Don’t suppose there’s a compact inline (piggyback) power meter open source project I could build, is there?

>>3012231
> current transformer
There is one inside a every GFCI outlet.

>>3012251
They don’t need to be proportional or linear, nor do they care about phase accuracy, they’re just characterised with a tripping point.

I realise I could use a diy rogowski coil instead of a shunt if I’m looking at powering it and shucking a microcontroller in there either way. Just add an OP07 integrator with a sensible minimum and maximum frequency.

File: NHS100-8R-F_sml(200x200).jpg (5 KB, 200x200)

5 KB JPG

>>3012403 >>3012409 >>3012417 >>3013024

Look I'm not the smartest when it comes to this stuff but I need some help with building a dummy load rig.

I need to test up to 300W per channel for HiFi amplifiers so I'm thinking about what I need to build the circuit. obviously I need the resistors, which I'm looking at getting 2 of these 8 ohm, 300 w resistors.

https://www.digikey.com.au/en/products/detail/ohmite/NHS100-8R-F/17038359

then from there I'm genuinely not what what I need to do. obviously they need some form of active cooling and like some banana plug terminals and then something to mount them to as well. ive seen some dodgy videos on these but none that look genuinely safe.

could I please have some help or if an existing pre made product exists for a couple hundred dollars I'd much rather just buy that but the cheapest I can find is in the thousands. i already have amps to test and an ADC for measuring the output (though its snr is only 109db so I plan to upgrade to something more sensitive soon.

ive seen guitar dummy loads that look good but ive since learned theyre not the best because theyre for reactive loads rather than non inductive?

>>3012402
sorry forgot to add I know I'll need to wire a smaller resistor bank to make a voltage divider right? as to not blow up my adc

>>3012402
You don’t necessarily need active cooling, if you only use it for a few seconds while you trigger the ADC measurement. Alternatively, get some immersion heaters or other 120V water heating elements (15A at 120V is 8Ω), they’re safe to dunk directly in a bucket of water to sink their heat away, and they don’t have temperature coefficients like incandescent bulbs do. Even a length of resistive wire or power resistor dunked directly in water can be fine, the wire will have lower resistivity than the water so I don’t think you’ll see any significant electrolysis, though this assumes the wire doesn’t double back on itself.

A PVC pipe full of mineral oil like a ham radio dummy load could also work, if you need continuous use.

If you do have to use a heat sinking and fans, look into those long heat-sink extrusions that slot into each other to make a square channel that you can screw a fan to. I’ve got one that takes a 90mm fan, and you can get some pretty powerful 90mm fans.

I wouldn’t overthink the design unless it has to be robust enough to be reused often, or portable. Add mains rated thermal fuses or thermostatic switches to parts that get hot, if you’re concerned about fire risk, they’ll handle 50V 6A audio AC just fine. Use mica and ceramic where needed to hold any very hot components, electrical crimps instead of solder. Screw things solidly to a board of plywood or MDF or a wooden frame if you’re concerned about things tipping over and spilling water or burning flooring/carpet.

>>3012403
That’s not a bad idea at all if your ADCZ can’t handle the input voltage, though it’s also an option to use an audio transformer to step down the ratio, as this will additionally isolate the signal in case you might have shared grounds between the amp and ADC that could cause ground loop noise. Either way I’d want to add some clipping diode protection if the ADC doesn’t already have overvoltage protection.

>>3012409
yeah its for a setup that needs to be pretty robust, as its essentially going to be used for multiple amps multiple times. while I'm sure that it would work I'd rather not play around with water and mains as I'd guarantee I'd find a way to kill myself
good advice though especially in the second part, I'll take it on board

>>3012402
does this need to be a "professional" design? if not, i would just buy the cheapest chinky low inductance power resistors you can find, chain them together with spade terminals and jumper wire until you reach a nominal 300W rating, and submerge them in a 5 gallon bucket of water. that way you don't have to fuck around with buying a big heatsink and fans.

File: Capture.png (501 KB, 1454x731)

501 KB PNG

>>3012420

>>3012411
in that case, buying used CPU heatsinks off ebay would probably be the most compact and cost effective option. bolt the resistors to a piece of aluminum or copper, and then mount the CPU heatsinks to that piece of aluminum/copper.

i've used these to sink 100W continuously @ 100C. https://www.ebay.com/itm/389697000053 if you looked around you could probably get them for much cheaper, they're basically e-waste. if you're fine with drilling and tapping copper (it's a pain in the ass because copper grabs taps) you could bolt the heatsink directly to the mounting plate.

File: Capture.png (727 KB, 1486x664)

727 KB PNG

>>3012411
>>3012419
something like this could probably do 200-300W easily @ 100C. just don't go crazy and push the temperature too high or the heat pipes will rupture. https://www.ebay.com/itm/227284369569

which AI bot works the best for you?
I realized that Google AI (not gemini chat but actual google AI mode) is surprisingly decent at reading and explaining basic circuits and building blocks. Claude is great, too. I never really think over anything myself as in the past. I ask AI to plan every detail and no longer take any notes. If I forget, I ask again. Oh and I have them generate per-task micro calculators so I no longer have to use shitty half-assed online calculators. Great times to do great things. The latest example: I wanted to build a growing light with a specific spectrum that approximates McCree that is popular in horticultural applications. AI built me a graphical mix and match tool and recommended a minimum set of LEDs to achieve a certain curve and a separate calculator to design a diffuser that would provide enough lux of uniform light. Next step is to hook up a $150 chinese spectrometer that would provide immediate feedback: color temp, CRI, lux, etc so any profile could be built automatically by adjusting current via individual strings of different colors. Considering how slow /ohm/ and /diy/ have become, I am assuming most people have switched to AI.

>>3012537
I ask 4chan and then bots give me the answer

>>3012537
Local model, not like I’m using my GPU most of the time. Besides not being able to access the internet (I’m keeping it that way) it seems no worse than the Google and duck.ai free search models.

I use it like Wikipedia or Google, a broad knowledge base with low trust as a basis for initial research. Within its memory are things I’ve never heard of, keywords I don’t even know to search. I’ll ask it about something novel to me, then do more detailed search for appnotes and articles written about what it tells me.

Claude coding is so powerful because it can be set up in an automated feedback loop, where it writes code and executes it and analyses the result. While that’s not really doable in hardware, an AI model that can create good spice models from datasheets, and a model that can convert between schematic images it’s been shown and netlists, could close that loop for electronic design. Throw every schematic of reverse engineered product and diy project on the internet at the model to train it.

I don’t like the idea of losing mental autonomy, so I try to be conservative with my use of the bots. I’ve seen people who live depending on GPT, it’s freaky.

File: Zrzut ekranu 2026-10-26 o(...).png (73 KB, 744x914)

73 KB PNG

>>3012585 >>3012591

I fucked up bad. I'm trying to build a Chinese ATC radio kit and damaged this part. R8 was shorted, so I started to remove components, looking for the faulty one and unsoldered T1. The short is gone, but now I realized it might be the transformer windings working as intended. What's worse, I damaged one of it's legs. The manual calls it:
>7X7-10.7MHz
and adds:
>CT has been welded in T1 weeks inside, no need to install.
ChatGPT says:
>It most likely means:
>“The CT component is already built into T1 internally, so you don't need to install it separately.”
>Here, CT most likely means a center tap (a connection to the middle of a transformer winding). The wording is poor machine translation: “welded in” means pre-installed, and “weeks inside” likely means “winding inside.”
>The schematic shows T1 as a 7×7-10.7 MHz transformer with multiple pins. The note is telling you that the relevant internal connection is already provided by the transformer assembly, rather than requiring a separate component on the PCB.

https://dn760109.eu.archive.org/0/items/Chinese-producten-manuals/Aviation_Band_Receiver_Kit_Manual.pdf

File: IMG_20261026_004454~2.jpg (436 KB, 1836x1589)

436 KB JPG

>>3012583
This is how it looks.

How do I enable bridge mode in my shitty ISP locked extra router without shelling out for a new tenda or tp link router just so I have wifi on the 2nd floor or without calling another retarded technician who's going to charge me

>>3012583
The picture shows a capacitor between pin 1 and 3, I think that's what CT is. If you have a fast enough scope and function generator, it should be pretty doable to measure the self-resonant frequency of the LC network between pins 1 and 3. Same for if you've got a VNA. That resonant frequency should change as you screw the core in and out.

Which leg did you damage? If you can measure a finite resistance between pins 4 and 6, and between pins 1, 2, and 3, I think you're fine. I'd guess anything from 10k to 10 ohms with an ohmmeter.

If you need a replacement, I'm seeing similar parts under the search terms "tunable IF transformer 10.7MHz". 7mm seems to be a pretty normal size.

>>3012590
ask /g/
their linux or home server generals might have some knowledge on those matters