Power supply build quality pictorial. part 2

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

I will dig out this post of mine (No. 700 in this thread if you wanna the pictures).

Been playing arround with my new loader and clamp meter for a while. I have succesfully done more or less complete rebuilt of this thing some time ago. It should now be capable of it's rated power. One thing I have noticed is that it seems to HAVE some kind of protection despite how crappy it looks. The thing is, it sits on the least important rails - +3,3 and +5 V I have been able to squeeze only arround 130 W combined from these rails, than it shuts down.

On the other hand, it really is able to provide 15 A for 12V power-hungry systems (well, plus some minor load on +5 V from fan). Voltmeter does not show any crossloading when loading 15A on that rail ONLY.

Will try to do some starting review on this thing, stay tuned for more

I can't wait

[Behemot]

Re: Power supply build quality pictorial. part 2

Just one question since I am still not really sure in this area. I have tracked this to be a half bridge, is that correct? Not center-tapped transformer, symmetric transformer output to rectifiers. Two transistors, ones base connected to others collector.

There is film capacitor between Vin and one transformer pin and small ceramic cap with resistor between the same pin and 0V. This is not very clear to me but somehow it corresponds to the schematic of half bridge. Putting this all together I think that's half bridge

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

Behemot, it sounds like a half-bridge. In a half-bridge one side of the O/P transformer primary is connected to the center of the two input electrolytics and the other side of the primary is connected through a film capacitor to the center of the two transistors. The film capacitor - typically 1uF or 2.2uF - keeps the transformer core from walking into saturation due to slight imbalances in the on times of the two transistors. Is the PWM controller a TL494-KA7500 type and on the secondary side?

[Pentium4]

Re: Power supply build quality pictorial. part 2

I recapped this entire PSU except for the primary cap. Replaced the 2200uf 16V Teapo on the 5VSB to a 3300uf 16V UCC KY cap. I replaced all the 1000uf 16V CapXon's with 1500uf 16V UCC KY's (and there were 7 of them) so I'm hoping the ripple will be in check with this thing

I'm having issues with my main computer and wondering if this thing would be adequate for my current build in my sig for a few weeks tops? I'm mainly worried about the ripple being too high

[Behemot]

Re: Power supply build quality pictorial. part 2

Behemot, it sounds like a half-bridge. In a half-bridge one side of the O/P transformer primary is connected to the center of the two input electrolytics and the other side of the primary is connected through a film capacitor to the center of the two transistors. The film capacitor - typically 1uF or 2.2uF - keeps the transformer core from walking into saturation due to slight imbalances in the on times of the two transistors. Is the PWM controller a TL494-KA7500 type and on the secondary side?

Yeah, exactly, KA7500. Than LM339N for secondary compensation.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

An LM339 is a quad general purpose comparator. It's probably used for Power Good and some combination of protective functions.

[Behemot]

Re: Power supply build quality pictorial. part 2

Yeah yeah, tracked one pin to PG, they also talk about being directly suitable for 12 V in datasheet so I think it may somehow drive +12 V. Than typical things like short I guess.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

The outputs of an LM339 are open collector - the collectors of transistors which have no internal connection. This lets the user use whatever positive collector voltage they want, within the maximum voltage and the practical minimum voltage, by connecting a suitable pull-up resistor between the voltage and the output pin. Thus, an LM339 can adapt between TTL and higher analog voltages, in either direction. It's one of those 1970s vintage parts that is still a workhorse in modern usage.

[Behemot]

Re: Power supply build quality pictorial. part 2

Oh, I was battling with the osciloscope more than taking measurement yesterday, but I noticed the voltage regulation is not good, really, I blame the way how they are trying to acomplish that, I think they could take much better way than 4way comparator.

With no load, there are strange negative spikes on every rail. For example on +12 V, peak-to-peak ripple is arround 60-80 mV no load and jumps to over 200 mV with load. If I add even minor load on +3,3 and +5 V, it drops to 80-100 mV at 120 W load on +12 V. Obviously they didn't use minimum load resistors yet the supply needs it to be within specs.

[Behemot]

Re: Power supply build quality pictorial. part 2

Oh my. Preliminary results do not show any good. The -12 V rail oscilates like hell no matter what I do with the rest. I guess it may have something in common with the coil I have put in instead of wire? Maximum ripple was 310 mV

About the rest, well. The +3,3 V rail got boost from replacing superfast with shottky and is now over maximum tollerance in ATX - it went as high as 3,6 V. Rest is good voltage wise, but the ripple on +12 V is too high, from some 9 A out of norm. It has no Pi coil (just shares core of common output inductor) and there is space for only one cap (I have put in a 3300 uF/16 V Samxon RS).

Here are some pics, do you think there is possibility of adding something to the +12 V rail to calm it down? As for the -12 V, I will remove the coil (the blue one) and see what happenes. Oscilloscope measurement is typical for what too strong coil does.

[Th3_uN1Qu3]

Re: Power supply build quality pictorial. part 2

I noticed the voltage regulation is not good, really, I blame the way how they are trying to acomplish that, I think they could take much better way than 4way comparator.

The comparator has nothing to do with voltage regulation. It is part of the protection (OVP/UVP) circuitry and the power good signal generator.

[Behemot]

Re: Power supply build quality pictorial. part 2

So you say the KA7500B is responsible for primary driving and secondary voltage controll?

Anyway, interesting fact is, the LM339N is most likely responsible even for OPP. I have not been able to suck labeled current from it, it has shut down at some point every time, no matter what combination of load I have chosen. I do not know whether it is ood or bad. It does not burn itself, yes, but…is it good they made it the way you CANNOT suck labeled power from it even if you tried?

[tom66]

Re: Power supply build quality pictorial. part 2

Behemot, you got a Rigol now? How did you get that nice graph out -- I can only get screenshot-like images??

[Behemot]

Re: Power supply build quality pictorial. part 2

This one is still not mine - it's here under one project (for all those ppl who need to do their electronics stuff). Older model, 1062E so it has been some fight, but we managed to get the Ultrascope SW running again. I just took screenshots from it with Print Screen

It just was not running great unfortunatelly, I had to unconnect the scope, set it on the scope manually, connect it and take screenshot. And than again. When the program is running as it is supposed to be, you can controll the scope settings in the program.

[Th3_uN1Qu3]

Re: Power supply build quality pictorial. part 2

So you say the KA7500B is responsible for primary driving and secondary voltage controll?

KA7500B = TL494, classic PWM chip. Developed in the 80s, still being used because it's just good at what it does. A little googling will find you some schematics of ATX PSUs based on the TL494/LM339 combo. Other versions will use a LM393 and some more discrete components, or a different supervisor altogether. But the TL494 remains a basic building block for a lot of power supplies out there.

Anyway, interesting fact is, the LM339N is most likely responsible even for OPP.

OPP is likely connected to pin 4 of KA7500B. It's coming from the driver transformer, via a 1N4148 diode, a small capacitor and a resistor divider.

I do not know whether it is ood or bad. It does not burn itself, yes, but…is it good they made it the way you CANNOT suck labeled power from it even if you tried?

It shows that labeled power is simply false, which is common practice in cheap, "no-name" PSUs.

[Behemot]

Re: Power supply build quality pictorial. part 2

But the thing is, most el cheapo PSU's don't have any protections. They just burn.

[Behemot]

Re: Power supply build quality pictorial. part 2

…

I will continue in separate thread to not polute this one anymore

[Th3_uN1Qu3]

Re: Power supply build quality pictorial. part 2

But the thing is, most el cheapo PSU's don't have any protections. They just burn.

They do. The OPP is just configured too high to act as more than a short circuit protection. Because in the process of removing parts/substituting with cheaper ones, they usually forget to reconfigure the protection circuitry.

[tom66]

Re: Power supply build quality pictorial. part 2

Another possible explanation is to get xxxW peak power, so it works initially in the users computer, passes 28 day return time then goes bang as they try to use it for longer at excessive power.

[Behemot]

Re: Power supply build quality pictorial. part 2

You can do RMA than anyway. Here is 14 days return and 2 years minimum warranty so it won't help.

But since in my case it shuts down instantly, I cannot even believe the talk about "maximum" power for short periods of time which our local business inspection told me. I tried to get this thing out of market, but talking to official organs is like talking to doorkeeper on main train station. Why do we pay them than? Another reason for me to want minimalistic state desperately.