PcBytes wrote on 2022-04-20, 07:15:

The one with JEE caps I suspect is some Raidmax or something similar, and the OEM behind that one is Sun Pro. That one is to avoid at all costs.

The two I disposed were branded as "IT" and "Codecom".
Can't tell you which was which. Already dismantled them, to prevent people from fishing them out of the garbage.

RaiderOfLostVoodoo wrote on 2022-04-20, 07:06:

It also had this thing attached to the top:

The funny thing is that the pic accompanying this reminds me of some other PSUs where this was also a thing. Made it more clumsy to take a look inside and clean it. Sometimes the fan is attached the same way iirc.

Btw the soldering on that last unit looks horrible 🤣

But I'm considering to add them to my Ebay inventory. I don't want to sell PSUs which are a potential risk to the users life.

Isn't any PSU potentially hypothetically possibly risky to users lives? I mean some are so heavy you could break your foot dropping the PSU onto it 😜

Tetrium wrote on 2022-04-20, 07:52:

Btw the soldering on that last unit looks horrible 🤣

That's two different units.

RaiderOfLostVoodoo wrote on 2022-04-20, 07:58:

Tetrium wrote on 2022-04-20, 07:52:

Btw the soldering on that last unit looks horrible 🤣

That's two different units.

My bad, I was still editing it (looks like I picked up some bad habits of a certain Error_Too_Many_Edits user named after a round thingy 😜 ).
But you're right, both look horrible.

RaiderOfLostVoodoo wrote on 2022-04-20, 07:06:

Combined with the professional soldering skills of Little Xiu from Shenzhen:

Little Xiu from Shenzhen has been breathing acid slag at three Euro per hour for the last forty years to afford the West an previously unfathomable standard of living, so maybe let's not piss on his merits the first chance we get?

RaiderOfLostVoodoo wrote on 2022-04-20, 07:06:

Yeah no, I'm not gonna use el cheapo PSUs for my valuable retro hardware.

They're not "el cheapo" they're "ching cheap chang". STB Voodoos are "el cheapo".

Let's not kid ourselves, this isn't about value, we're group OCDing here.

There's no such thing as "valuable retro hardware", it's obsolete junk in the real work - that's what makes it retro.
If it were about actual value - at least in economic theory, we could invest a mere tenth of the time we spend here reading the tea leaves about potential PSU quality into a minimum-wage job and that money into a PSU failure insurance against further hardware damage with really good coverage, since risk is the product of damage times probability.

In the time an average PSU will statistically malfunction in a way that kills a 100 € Voodoo, you'll have spent much more money on electricity.
These things aren't unique. They've been mass produced with the same individual passion as the average PSU. They're rare, that's it.
But this doesn't merit obsessing about PSUs. That's not economy, that's retrophilia.

Please don't hang the messenger! 😉

RaiderOfLostVoodoo wrote on 2022-04-20, 07:06:

But I'm considering to add them to my Ebay inventory. I don't want to sell PSUs which are a potential risk to the users life.

The eight different proof logos on the label say that it's not a potential risk to the users life.
What are they going to do? Bite them? Send a sudden, fatal power surge through the wireless mouse cable?
In strictly technical terms, the only official reason why they might be a hazard is if they've been opened by a non-professional and if so, that's also the only good reason not to sell them on ebay. Because now you're liable for anything out of the order. That's why there are seals on those thing, among other things.

Let me remind you of you own words:
Re: crappy old power supplies
- It's part of the experience of using old parts to use old parts. Don't let other people's timid control needs ruin that for you or you'll end up rewiring your whole house like audiophiles do.

PcBytes wrote on 2022-04-20, 07:15:

The Golden Field one is pretty anorexic in the heatsinking section

Have you tried using a pendulum?
Unlike the other one, the Golden Field has a temp controlled fan.
So, while still keeping a low noise ceiling, that's what affording it to use simpler heatsinks that won't put unnecessary stress on the main components.
Components that also vary vastly in efficiency - so consider that the nameplate power of those things is not what the heatsinks are for, it's the friction losses while providing that. 😀

I can only reapeat once more: The only PSU that ever did minor damage to any of my hardware was a friggin Enermax in regular, completely on specification, non-retro use.

Cuttoon wrote on 2022-04-20, 09:53:

Have you tried using a pendulum? Unlike the other one, the Golden Field has a temp controlled fan. So, while still keeping a lo […]
Show full quote

PcBytes wrote on 2022-04-20, 07:15:

The Golden Field one is pretty anorexic in the heatsinking section

Have you tried using a pendulum?
Unlike the other one, the Golden Field has a temp controlled fan.
So, while still keeping a low noise ceiling, that's what affording it to use simpler heatsinks that won't put unnecessary stress on the main components.
Components that also vary vastly in efficiency - so consider that the nameplate power of those things is not what the heatsinks are for, it's the friction losses while providing that. 😀

I can only reapeat once more: The only PSU that ever did minor damage to any of my hardware was a friggin Enermax in regular, completely on specification, non-retro use.

That label on the Golden Field is a lie. If it was thermally controlled, you'd at least have an thermistor present to the heatsink, or a separate PCB that does the thermal regulation - I'm seeing none, so the label is a lie.

I'd however expect the CRS unit to have a thermistor, especially since the fan seems to actually have a removable plug, unlike the Golden Field unit.

Eh, I'm starting to think some people posting about "crap PSUs" haven't actually seen a low quality PSU. This is what the bottom of the barrel looks like:

And yet it works, even in this state, so I'll probably end up using it for less important projects, of which I have many.

BTW, the size or weight of the heatsinks is not a good indicator of the quality. Well, not with the modern digital PSUs, these things are so efficient that there isn't any need for a lot of heat rejection surface. It does hold true for the AT era stuff though, and ATX up to a point. If you see massive heatsinks then ask yourself why are they needed, does the PSU high current output demand more cooling or is the design crap and inefficient. Or both. I had a pretty decent Tagan 570W modular PSU that worked for like 15 years with little downtime before if finally died - and it was heavy with massive heatsinks but that was also due to the rather poor ~70% efficiency. These days I wouldn't even touch a PSU rated below 80%, and my current PSU is 94% and thus I've yet to see it spin the fan (except in test mode). Which is great, no noise and no dust.

Anyway, back the the photo, this is supposedly 200W unit and the PCB is also marked as such, but frankly I'd be worried trying to pull 150W out of it. Should not be a major problem considering most ot the stuff I build tops at 50W or so (and more like 30W for 386). The two input capacitors are marked 330u/200V 105C but both are measuring about 220u so I suspect these were fakes from the factory (especially considering their size). There is no input filter, I've added a 0.1u X2 cap, might actually look for a used common mode choke later as well. Output caps are 220u for auxilary voltages and 1000u for main 5V and 12V. No pi filter for 5V rail, in fact the choke wasn't even in the design. I'll add some 100nF ceramic bypass caps for the new low-ESR electrolytics but frankly this is not a PSU I want to spend money on, so I won't be putting my best Panasonic or Nichicon caps in it.

Also, no fan speed control, though perhps that is a good thing looking at the minimalistic heatsinks. But again, it works, and it was "free" since it came with an AT case I got pretty cheap.

PcBytes wrote on 2022-04-20, 09:57:

That label on the Golden Field is a lie. If it was thermally controlled, you'd at least have an thermistor present to the heatsink, or a separate PCB that does the thermal regulation - I'm seeing none, so the label is a lie.

Seriously mate? A well established OEM supplier will lie about the one feature that any user can actually test for, with a standard issue pair of ears?
That thing has a PCB. Why would it need another one?
And, why would you put the temperature sensor on the heatsink? The heatsink doesn't mind to be hot, as a matter of fact.
There's no electronic component that gives a reading of "this PSU component x is hotter than it should be".
It changes it's electric resistance, that's all. Everything from there is a matter of a some other resistors and probably two simple transistors that have been tuned to maintain a certain target temperature.
To calculate a probable offset between any area of the PSU and the most critical parts of a few degrees into that is well within the means of engineering, won't you agree?

I own a temp controlled Seasonic 150 W AT PSU, probably from around 1990. Wanna bet if that has an extra PCB and a cable and a thermistor somehow fixed to a heatsink?
If not, think they'd have managed to integrate that by the time of ATX.
That would be a completely useless extra effort in a very competitive market.
I'm not paying for that and neither should you.

Deunan wrote on 2022-04-20, 10:04:

Eh, I'm starting to think some people posting about "crap PSUs" haven't actually seen a low quality PSU. This is what the bottom of the barrel looks like:

BTW, the size or weight of the heatsinks is not a good indicator of the quality.

Thank you, good Sir. This was really beginning to get a little silly. Bunch of nerds obsessing about size... 😜

Cuttoon wrote on 2022-04-20, 10:13:

why would you put the temperature sensor on the heatsink? The heatsink doesn't mind to be hot, as a matter of fact.

The heatsink might not mind, but the diodes (or switching transistors) bolted to it do 😀

Deunan wrote:

And yet it works, even in this state

Heck, it even has a fuse!

Deunan wrote on 2022-04-20, 10:47:

Cuttoon wrote on 2022-04-20, 10:13:

why would you put the temperature sensor on the heatsink? The heatsink doesn't mind to be hot, as a matter of fact.

The heatsink might not mind, but the diodes (or switching transistors) bolted to it do 😀

Oh come on, this isn't highschool. Abstract thinking will not be frowned upon.

As I did explain further on, my point was, the one heatsink isn't every critical component in the unit. So when factoring in that offset, you may as well factor in the offset of any point in the case. It's thermodynamics, not magic.
As in "as long as this part here where we conveniently situated the fan control circuitry doesn't get any warmer than x, the most vulnerable bit of this PSU will not get warmer than x+n, based on testing, calculation and experience".
Do I need to draw a picture?

Deunan wrote on 2022-04-20, 10:04:

Eh, I'm starting to think some people posting about "crap PSUs" haven't actually seen a low quality PSU. This is what the bottom […]
Show full quote

Eh, I'm starting to think some people posting about "crap PSUs" haven't actually seen a low quality PSU. This is what the bottom of the barrel looks like:
IMG_1530.JPG
And yet it works, even in this state, so I'll probably end up using it for less important projects, of which I have many.

BTW, the size or weight of the heatsinks is not a good indicator of the quality. Well, not with the modern digital PSUs, these things are so efficient that there isn't any need for a lot of heat rejection surface. It does hold true for the AT era stuff though, and ATX up to a point. If you see massive heatsinks then ask yourself why are they needed, does the PSU high current output demand more cooling or is the design crap and inefficient. Or both. I had a pretty decent Tagan 570W modular PSU that worked for like 15 years with little downtime before if finally died - and it was heavy with massive heatsinks but that was also due to the rather poor ~70% efficiency. These days I wouldn't even touch a PSU rated below 80%, and my current PSU is 94% and thus I've yet to see it spin the fan (except in test mode). Which is great, no noise and no dust.

Anyway, back the the photo, this is supposedly 200W unit and the PCB is also marked as such, but frankly I'd be worried trying to pull 150W out of it. Should not be a major problem considering most ot the stuff I build tops at 50W or so (and more like 30W for 386). The two input capacitors are marked 330u/200V 105C but both are measuring about 220u so I suspect these were fakes from the factory (especially considering their size). There is no input filter, I've added a 0.1u X2 cap, might actually look for a used common mode choke later as well. Output caps are 220u for auxilary voltages and 1000u for main 5V and 12V. No pi filter for 5V rail, in fact the choke wasn't even in the design. I'll add some 100nF ceramic bypass caps for the new low-ESR electrolytics but frankly this is not a PSU I want to spend money on, so I won't be putting my best Panasonic or Nichicon caps in it.

Also, no fan speed control, though perhps that is a good thing looking at the minimalistic heatsinks. But again, it works, and it was "free" since it came with an AT case I got pretty cheap.

Wow...there's literally barely anything inside this PSU 🤣
I have seen empty ones, but I don't remember having ever seen a PSU this empty.
Couldn't they have saved some more money by making the PCB a bit smaller? 😜

How about that chunky boi? Weight - 2.5 kg. Lots of Koshin caps. 2 fans, semi-modular. Unknown brand, long dead by now.

CWT (Channel Well) OEM, looks pretty sturdy. What are the connectors on it? (besides normal ATX, that is 😁)

Cuttoon wrote on 2022-04-20, 13:31:

As in "as long as this part here where we conveniently situated the fan control circuitry doesn't get any warmer than x, the most vulnerable bit of this PSU will not get warmer than x+n, based on testing, calculation and experience".

This is all fine in theory but the conventional wisdom is to place temperature sensor as close to the hotspot as possible. One reason would be the airflow forced by the fan will make a distant sensor pretty much cool all the time, no matter what temperature the semiconductors are at. This is less of an issue in a closed case with no vent holes, but we are talking about PC PSUs. Another reason is you can detect a spike much faster with the sensor on the heatsink, so you can start cooling earlier and have less delta between temperature points. Which is important as thermal cycling is stressing both the parts and the PCB (cracked solder joints are a thing).

So, while it's not an unbreakable rule, most sensors will be placed either very close the heatsink on the PCB, and perhaps even be bent and even glued in place to touch it, or will use wires to reach it. BTW, a good example of the correct sensor placement is an old, power transistor based final stage of an audio amp. These use silicon diodes to compensate the bias voltage, which needs to change as the transistor heats up and it's BE junction drop gets lower. It's a very effective feedback scheme but it needs to have the diode follow the transistor temperature with no delay.

Obviously there are other factors as well, for example one might actually want to have some delay in the fan response, so that it doesn't change the speed too often an make annoying sounds. But that can be also done with an RC filer, and would probably be much more predictable than having the sensor further away. Just my 2 cents.

Deunan wrote on 2022-04-20, 20:57:

This is all fine in theory but the conventional wisdom is to place temperature sensor as close to the hotspot as possible. One r […]
Show full quote

Cuttoon wrote on 2022-04-20, 13:31:

As in "as long as this part here where we conveniently situated the fan control circuitry doesn't get any warmer than x, the most vulnerable bit of this PSU will not get warmer than x+n, based on testing, calculation and experience".

This is all fine in theory but the conventional wisdom is to place temperature sensor as close to the hotspot as possible. One reason would be the airflow forced by the fan will make a distant sensor pretty much cool all the time, no matter what temperature the semiconductors are at. This is less of an issue in a closed case with no vent holes, but we are talking about PC PSUs. Another reason is you can detect a spike much faster with the sensor on the heatsink, so you can start cooling earlier and have less delta between temperature points. Which is important as thermal cycling is stressing both the parts and the PCB (cracked solder joints are a thing).

So, while it's not an unbreakable rule, most sensors will be placed either very close the heatsink on the PCB, and perhaps even be bent and even glued in place to touch it, or will use wires to reach it. BTW, a good example of the correct sensor placement is an old, power transistor based final stage of an audio amp. These use silicon diodes to compensate the bias voltage, which needs to change as the transistor heats up and it's BE junction drop gets lower. It's a very effective feedback scheme but it needs to have the diode follow the transistor temperature with no delay.

Obviously there are other factors as well, for example one might actually want to have some delay in the fan response, so that it doesn't change the speed too often an make annoying sounds. But that can be also done with an RC filer, and would probably be much more predictable than having the sensor further away. Just my 2 cents.

Fair enough.
What people seem to under appreciate in real world application: Theory has REALLY LOW marginal cost.

this 500w psu is JAM PACKED! in fact it is so FULL that when plugged in and turned OFF the voltage is leaking out. i measure 1.8v on all the red wires and 4v on the yellow wires. when actually powered on it reads within 5% of spec.

Great example of what I meant - big heatsinks but poor PSU. AFAICT it's an old analog design with medicore efficiency - in fact it looks like most of the AC input filter is missing to save a few cents on the manufacturing costs, and there isn't any PFC I can see either. Because it's modular but the case was kept standard size, there isn't much space to let the air in so it mostly just uses the top, horizontal surfaces of these heatsinks to cool things down. Output capacitors will have a hard life in this unit, I would not be surprised to see them degrade or even bulge/leak within a year or two.

As for the "power leak", I guess its design is too dumb to properly supply standby voltage for USB so it kinda runs all the time but at much reduced cycle duty. And, because there isn't a proper separation of the standby circuit from the main one, there isn't any space for that, the main output will be energized and without some load to bleed it you will see the weird voltages. It might even get worse if you load the standby rail but not the main ones.