I recently needed an adjustable PSU, so I opted in for an Aliexpress solution – PCB containing LM317T IC, capable of 0-30V@1.5A.
At least in theory.
In reality, ehh…. I don’t even want to discuss that sorry mistake of a board design. Fake 317T, small traces, and 1A rectifier diodes.
So, I decided to make my own entire PCB. And I overengineered it, as I designed it for LM338T.
And, also ragefixing, a CRT film removal and attempted motherboard repair.
Ragefixing
OK, so let’s start with less interesting one first: Dell P791 CRT. My first CRT ever. It’s only one month younger than me, but it probably works better than I do 🙂
What was wrong? Scratches. A lot of them.
Now, You probably ask yourself: Scratches? On a glass surface?
Yes, scratches, but only on the ARAG (ARAS?) surface – on the anti-glare material.
This is what I was putting with originally.
At first I started scrubbing with 100% acetone, but results were… underwhelming.
At that point I decided go on the internet. Majority of Trinitron tubes had a plastic film over them, way, way easier to remove than what I was trying to do.
Unfortunately it has a Diamondtron tube, but that got me thinking.
How can You check if it’s glass only or plastic?
Scratching the surface with a knife.
No, I’m not joking. Have You ever heard of Mohs scale of hardness?
Basically there is no way that ordinary kitchen knife could scratch a glass surface. But it easily can sratch a plastic surface.
It fortunately made a scratch on the surface. I was off to disassemble this thing in a matter of seconds.
There is one thing though. This plastic surface is connected to GND of the monitor and it prevents the charges from anode from building on the front glass. The monitor will be 100% safe to use, but there when You touch the glass it will hiss a bit, exactly like it was in older television sets.
This stuff is hard to remove, but IPA really helped in my case.
Now, how about the colors without it?
In the daylight blacks are a bit more grayish, but nothing unusable. At night though, this tube is like new.
Alright, that’s about it for the CRT, how about the motherboard?
OK, we’re starting HOT!
Insert *ba dum tss* sound here.
It’s an Abit IC7 with a hot SB – shorted.
Old one out and new one in.
And a massive fail.
It worked for about a second before smoking from the CPU VRM area and shorted the SB again.
Now, I haven’t told the whole story – I already have checked whole VRM section, even replaced the controller and MOSFET drivers, as well as one MOSFET which was shorted. Only after that have I replaced the SB.
I have removed the MOSFET which was faulty previously, but this time it was OK. No sign of other damage.
If I had a supply of these SBs then maybe I could diagnose it more, for example by disabling one driver, but unfortunately I have to deem this PCB unfixable.
I think after that sad ending it’s high time for a main course: engineering at its finest.
DIY PCB production
As I said previoulsy, I needed a variable PSU for one of my side-projects, so I got myself one.
Too bad it’s seriously BAD.
I had to mod it out of the box with bigger capacitors, as these supplied were not only of poor quality, but also RATED AT 25V!!!
ON A PSU CAPABLE OF 30V!!!
And I’m not even mentioning capacitance or ESR.
Ehh… You have to always be on high alert with anything that’s Chinese.
I believe it’s a high time for a tangent about LM317T/338T ICs. I bought a couple of them, some new, some used, some from Poland and some from China, some cheap, some not. I also desoldered one from some old board. You know which were original?
Used 338T from China and desoldered mine. That’s all. Funny thing is, these 5 LM338T cost me about as much as one fake one from Poland.
This is the IC supplied with the Aliexpress kit. If You look really closely You can pinpoint the exact place whey rip You off.
The engraving is in fact paint.
From left to right:
- desoldered, LM317T
- used, Chinese LM338T
- new, Polish LM338T
- used, Polish LM338T
- new, Chinese LM317T
Fake Chinese 317T in short circuit caps at 200mA. I’m sure this is supposed to be a funny joke, but I’m not laughing.
Used Polish LM338T. More like a LM317T.
New Polish 338T. Again, not even close to a real IC.
Desoldered LM317T.
Why is it more than the 1.5A in the datasheet? Because it can. Literally, since the temperature is low it can supply more current, when temperature rises current decreases. 1.5A is just a guaranteed, reliable amount You can work with.
Also, I’m pretty sure it could pull more power, but my own bench PSU capped out at ~3.2A.
Used Chinese LM338T. Possibly a real IC, no way to actually test it. My bench PSU is screaming for mercy at this point.
Man, I should make myself a 10A bench PSU 🙂
That’s about it. Don’t buy anything from Polish retailers and buy only used ICs from China. I have more similar stories about used ICs from China, especially audio op-amps, but these have to wait for a different post.
Ok, so I ditched the supplied POS 317T and installed the used 338T. And?
It burned.
Oh gee, who could have seen it coming? 1A rectifier diodes (1N4007) couldn’t keep up with the rest of the module and released the magic smoke. If only I could refill them…
But seriously now, this kit is full of these bad design choices – bad rectifier diodes, crappiest voltage regulator they could possibly find, traces that are about 1mm wide, and capacitors which would immediately blow once powered.
Since I can’t rely on already available solutions I decided to go all out – learn the EasyEDA program and etch my own PCB.
It took me a while, but I painstakingly learnt how to design a circuit, convert it into a PCB design and print it onto a paper. Getting the right paper for thermal transfer was a journey on its own, but I ultimately got it.
My PCB design. Probably not the most optimal, but it works. As a bonus I designed it so that I wouldn’t need to buy any more parts.
Shame I didn’t take a picture before the PCB was already etching, but You still can clearly see the transfer result. Pretty good, If I do say so myself.
There was a tiny bit of a trace missing, but that’s not a big deal. I can fix that with solder.
PCB with drilled holes. There was also a short between LM338T traces, but I fixed them with a knife.
Since I was limited to one-sided design I had to make sacrifices. There were 2 jumpers needed – one for the potentiometer and one for output ground. Traces that handle the power are 2.5mm wide, so I figured that 1mm solid core copper wire would do the trick.
All traces were tinned and not cleaned from flux, since my flux for sure doesn’t corrode over prolonged time (checked by myself), but clean copper does.
I designed it for continous operation with 5A flowing through it. I don’t have a PSU capable of such power, but it maxes my bench PSU no problem – obviously tested later, as on the picture the voltage regulator doesn’t even have a heatsing attached.
I have a weird tingling sensation in my right foot, it probably means that You’re going to see more of my amateure design choices down the line 🙂
Thanks for reading!