Bad design examples

[starfury1]

Re: Bad design examples

The circuit you have supplied is incomplete in many connections making it difficult to for example see where the supply voltage to pin 12 of the TL494 comes from, it is connected to nothing!!! Clearly it must be connected to a supply independent of the main switcher so that it can start and drive the switcher.

haven't fully read this at the moment just have other things to take care of but looking at the schematic and bear in mind no guru here

I agree Davmax that there should be an independent supply for the controller...now not sure how this all slots together
from what I can tell T6 is feed from the direct mains rectification DC
I am assuming that the circuitry on the primary side windings (T6) are some form of discrete build OSC or chopper circuit to produce AC
(transformers don't work well with DC!)

if you look at the secondary winding D30 is connected to pin 12 of IC1 but then were is the filter cap?
the closest thing I can see to that is C21 22uF 16 Volt shown schematically on the far side (other side) of the pwm controller IC1
This seems rather small but I suppose that really depends on how much current it needs to supply.
(bold used to high light connection components)

you are right it is a bit hard to workout and sometimes schematics don't convey the flow of functions very well at all.

anyway yes it should have a DC supply and thats about the closest I can see that might be it represent on the schematic.

its open for comment of course

the thread its self is interesting and your input Davmax is always of interest.

Points and counter points are always needed in a post like this as it opens other avenues up that might need to be highlighted and further examined.

If you can nail what caused a failure you are in a better position to improve the situation against this type of failure again.

BTW aren't Green coloured resistors fusible resistors?....I dont know if this is defined as a code or not but I thought if they were green body they are fusible resistors...?

forget it googled a few photo seems not the case

does raise a point thought were when are these likely to be used in a switcher ?

Anyway when I get a chance i'll give this a good read through

Thanks guys

Cheers

[i4004]

Re: Bad design examples

i4004. The circuit you have supplied is incomplete in many connections making it difficult to for example see where the supply voltage to pin 12 of the TL494 comes from, it is connected to nothing!!! Clearly it must be connected to a supply independent of the main switcher so that it can start and drive the switcher.

what about d30?

What the circuit diagram does show is that the second power supply is entirely seperate from the main switcher and obtains it's power from the primary DC input rails. It also shows that it is most likely a free running oscillator without any feedback regulation because it only has to supply the 7805 for a 5Vstb. There is no relationship as I pointed out to the secondary or primary circuits of the main switcher and cannot therefore screw up the main switcher as you stated. Or the other way round, the switcher screw the secondary power supply.

dav, beefy of this will probably be that you're the guy that can't read electric diagrams, yet you know everything about failure modes of psu, huh.

it is supplying power to ic. it is hooked to pin12 of ic via d30.
circuit of psu 4 and 5 differ from this scheme(i didn't say they're completely the same) and they have extra cap after the diode to smooth out the supply to ic.

Looking at the circuit the most likely failure would be the high voltage transistor switching this small transformer.

yes, it failed.
like i already said.

There is no way a massive main switcher primary current can all be transferred across the transformer. If you do not understand my explanation please study transformer theory and operation or get somebody who knows to explain it.

you're again using that tone. i don't want "i know, you don't" types of answers. you need to elaborate your standpoint because experience is proving you're wrong.
also, who ever said all current will be transferred? current won't be "transferred" at all. magnetic field will, and it will be proportionate to current and it will induce proprotionate voltage(and then current) in the secondary.
those are trafo operating basics.
if you wanna prove small trafo can't produce too high voltage on secondary when switching transistor shorts, show me graph of flux vs. current and you're done. if you can't prove this, you proved nothing.

If for example it required 30% more primary current to saturate the core the increased and maximum stored energy available increases by 69% to be precise.

so now it's 69? swell!

The principle I need you to grasp is the fact that a low energy transformer transfers energy in small energy blocks at the clock frequency (say 50-100Khz) so if one block is even twice it's normal energy it is still not going to have a major impact. A switching transistor short circuit will produce only one pulse energy block to saturation level, any current beyond that point will not increase the core flux density and therefore stored energy, inductive reactance to current flow will cease and the short circuit current will rapidly increase blowing anything in it's path.

if one block is twice its normal energy(btw. that is your speculation: you don't know how big is it, or how long does it last, or what current is needed to saturate the trafos we discuss) it produces 2x higher voltage on secondary.
you say this is not enough to make some damage?

offcourse, you didn't provide any proof this current can't do 3, 4 or 5(or more) times nominal saturation, and ergo, same amount of voltage increase on secondary.

I am not going to give you a long explanation of transformer theory, if you do not know you should be very careful not to mislead yourself and others.

you don't need to. i just asked for graph showing flux density vs. current for a typicall smps trafos. you being so fluent with it shouldn't have troubles finding it.
you must of seen it somewhere, because otherwise you wouldn't be making such claims, right?

also, nice to see you're making some progress: here
https://www.badcaps.net/forum/showpo...9&postcount=51
you didn't saw how can feedback loop fail at all. good thing you now believe it, but this time it's not about feedback loop. smps have more than one failure mode.

by your account, we don't have explanation for a thing that happened:
too high primary current(on small trafo that supplies ic) can't do wrong on secondary(so pwm ic didn't affect main switchers in a bad way), and then we're only left with feedback-loop failure, but that also is not a proper explanation in our case(because it suggests we should discard burning of one circuit and concentrate on circuit that's fine).
your standpoint is that catastrophic failures cannot happen: primary short circuit won't be the cause, and feedback failure is not something applicable here: yet such disasters happen, which means you're wrong.
i say short pulse(or long pulse...it's obvious short circuit lasted some time here) of high current on primary is enough to burn stuff on output; why not?
it's not like stuff hooked to secondary requires a particular time to burn. it requires a particular energy.

also, i say that if it can happen at this circuit where the trafo is REAL small, it also can happen on main switcher and its trafo, with a probable difference that it's primary windings won't burn like they did on this small trafo.

so you're willing to attack me and my explanation without providing anything more substantial. this is not appreciated at all.
if you have a better explanation, i'm listening. thus far you provided nothing but distraction.

[i4004]

Re: Bad design examples

starfury, i think i love you!

(you were swifter to reply about d30...also, i had the unpleasant task to discuss trafo some more..hehe)

c21 is well spotted.

on my psus (no.4 and 5) i actually have 47u/35v, and it's fairly close to the small trafo.
like i said, this scheme is not full identical to psus in question, but is pretty close. it shows the general principle of this cheap psu.

[starfury1]

Re: Bad design examples

Thanks,
well I'll have to leave the larger discussion on this to you guys I really dont know enough
to comment too much.

With that schematic it is hard to spot how it is derived, but it must exist
(or else not shown)

got to run

Cheers

[davmax]

Re: Bad design examples

The details fo SMPSs will be continued in seperate thread here:
https://www.badcaps.net/forum/showthread.php?t=5108

[starfury1]

Re: Bad design examples

since the TL494 controller is used in the schematic diagram posted

I found this doc on it that covers designing with it.
fairly well covers each block of it by the looks of it

https://cdn.badcaps-static.com/pdfs/...3bce4b5456.pdf

Also another Doc

https://cdn.badcaps-static.com/pdfs/...67fcc5c1b6.pdf

General and somewhat sweeping comment..meaning not totally accurate

This controller was fairly commonly used in XT AT ATX designed PSU's (1980's 90's era)
produced by many manufactures and also goes by the number KA7500B
what differences (if any) may exist I don't know



There are other controllers used as well and the 90's these would have been used lesser extend, more likely to be found in older, cheaper supplies.

most (if not maybe all) supplies build for PC use today,
(meeting current regulations)
would not be based on these controllers or the schematic as shown.
(as far as I am aware that is)

Thought I post this here due to schematic being in the thread.

Cheers

[i4004]

Re: Bad design examples

davmax, some good news for you (not so good for me, but..anyway):
turns out you were right, and i was wrong.

just talked about it on a particular local forum and the guy says mostly auxillary power supply goes bonkers and makes main supply go out-of-specs, which results in mess on output. just like we saw in my example.
(bad lytic messes this oscillator that feeds primary of small transformer and that messes the pwm ic....perhaps the small-transformer disaster didn't destroy the pwm ic at all (my original guess was that supply went too high), but the supply voltage was just going down and caused a breif period of pwm ic producing wrong pulses...because supply voltage is used to produce stable +5v inside the pwm ic that is then used to run internal oscillator, or perhaps this fading supply made internal comparators(error amplifiers atc.) go bonkers....plenty of crap can happen if ic's supply fails, wether going too high or too low....and this circuit can go either way as it doesn't have much of a regulation...).

also, i've found few examples of primary switchers breaking but still not harming the output (i think willawake's recent experience smells like it
https://www.badcaps.net/forum/showpo...6&postcount=15
also this by per
https://www.badcaps.net/forum/showpo...10&postcount=6
and some examples on some other forums).
if i was right, then all of those would have destroyed the mobos, and whatnot, but they didn't.


so..sorry about that.

[starfury1]

Re: Bad design examples

So a fact finding mission supported what davmax was saying....

Your posts were based on the information and knowledge you had gathered and then the theory you developed from that said information and knowledge
Thats fine...

you obviously took Davmax's information on board and then investigated what he had said
Then redraw your theory and conclusion based on that and the further investigation...thats good.

Its not so much we make wrong conclusions or theories, as its a lack of information or some aspect we miss (or miss understand) when drawing them....
(I do it often enough here, and then have to fix with a more accurate information or redefine what I was saying in proper context)

So its good that you revised your theory & conclusion and were big enough to post on it
As its accurate information we need to exchange whenever possible.

As I said the thread its self is a good idea
(and The new thread, were we really should continue details of designs)

Happy to see Davmax has chimed in on this too

Thank you I4004

I would love to have more input but like Davmax, have a family thing so wont be able to concentrate too much on the real technical aspects
(which I just don't know enough about anyway and wont really have time to research in detail)

As you know, I like to be somewhat sure of what I am saying before I go shooting my big mouth off
(like USA having 2 Actives...I totally didn't know that)

Cheers

[davmax]

Re: Bad design examples

i4004. Thanks for your input. I am working on covering the understanding that typically is not published. To cover statements like "more power in the primary gives more output power" this is very misleading in that in terms of controlling the MOSFET ON pulse width under normal operation this is true, but it does not cover the limiting values or what happens when things go wrong.

I will get it altogether. During the process I was able to see the destructive path provided by the secondary power supply delivering a supply voltage to the switcher controller chip. If that supply goes wrong anything can happen. This is a weak point. Same as any break in the feedback loop, if that happens the outputs will shoot up. However good PSU designs cover this with an over voltage detection circuit.

So you have stimulated me produce info on how energy moves across the transformer and what happens if things go wrong. Looking at what is written to date it takes quite a volume of words cover the processes involved in the ON & OFF switching cycles.

[i4004]

Re: Bad design examples

yes, the problem is this is really poor covered in publications, which means we should go with experience, not just our experience, but other folks' stuff too.
and if you can find many examples of "switchers blown, but pc ok", or even better, like "my psu was barfing for days before it finally destroyed my pc" then you see a pattern.

this also means pwm ic supply circuit is something that should be considered as critical for safe operation of psu. designs as explained above(without chips, with crappy components) should be avoided.

somewhat surprisingly, i have found same crappy circuit in my old enermax 365(one i posted pix of, the one i payed waay to much), but it didn't fail, probably because of better cooling, and maybe the fact that i turn it off completely overnight, as usually this circuit is on all the time.
a crappy circuit that's on all the time: can't be good.

one more interesting thing; in one of the pdfs starfury linked suggested power supply for this chip is linear power supply. that would make much more sense and be much more reliable design.

lastly, looking at this cheap design i got this feeling; somebody finds some old psu electric diagram (on the internet) and starts making such psus to make some money fast and in the process even further decreases quality and number of components.
sure such supplies will fail more.

overall, any design that uses such a ridiculous method to supply pwm ic must be branded "utter crap".

[i4004]

Re: Bad design examples

here's another example of that cheap design
https://www.badcaps.net/forum/attach...achmentid=7405
(https://www.badcaps.net/forum/showpo...&postcount=288)

one transistor oscilator that drives primary of 'ad-178 t3'.

[davmax]

Re: Bad design examples

I see that when you refer to crappy design you are talking about components, quality and mechanical layout. Whilst these are certainly major contributing factors to lack of reliability I get really interested in how well the circuit design will respond to error or failure ie how damage can be avoided/minimised.

one more interesting thing; in one of the pdfs starfury linked suggested power supply for this chip is linear power supply. that would make much more sense and be much more reliable design.

I am not sure about this statement. Where is the linear design? For the secondary power circuit? It certainly cannot be achieved for the main power.

[i4004]

Re: Bad design examples

https://cdn.badcaps-static.com/pdfs/...3bce4b5456.pdf
page26.

it is an example circuit, but i find the idea of using small linear supply rather good! its simplicity guarantees much higher reliability than either this cheap&cheesy-design transistor OR design with pwm ic supplied from 300v rail.

when i mention crappy design, yeah, i mention quality of the components, but also design itself is poor because it relies on components that are prone to failure.

one could easilly envision ultimate psu:
1-secondary power supplied by linear supply: this is almost bulletproof.
2-all components around pwm ic top quality, from lytics to resistors, same for output smoothers
3-power mosfets and their coolers(and everything on primary side) grossly overengineered, use parts with much higher ratings than will it ever be needed(for example if 15A would suffice, put 30A part) , put a lot of metal for cooling.

that would be nice.

[davmax]

Re: Bad design examples

Unfortunately the linear regulator you refer as an example is not a linear reg. It is a switch mode working at 20Khz and supplied by 32 volts. Now I understand why I could not find it.

The example is a 5V output job that is low powered (50W) and has the advantage of a step down transformer to get away from high voltage. This approach is just not feasible for high powered ATX SMPSs.

Unfortunately to maintain ATX PS efficiency there must be a 300V+ capacitor and this is the only source available for the designer to create second power supply. On earlier ATX SMPSs I have seen a small stepdown 50/60Hz transformer used for secondary power. But as the 5Vstb power specification has increased for modern PCs the transformer has to be much larger so designers have gone for a small switching PS off the main DC input to keep size down and efficiency up.

[i4004]

Re: Bad design examples

It is a switch mode working at 20Khz and supplied by 32 volts.

and how is it supplied by 32v, and how is ic itself supplied?
by what sort of power supply?

This approach is just not feasible for high powered ATX SMPSs.

did i ever say we should use linear supply to supply ALL the voltages from psu to pc? why do you insist on repeating it?

But as the 5Vstb power specification has increased for modern PCs the transformer has to be much larger so designers have gone for a small switching PS off the main DC input to keep size down and efficiency up.

lookin' at the manual of my new asus board reveals 1A on 5vstb to be used for 'pwr on by ps/2 keybard'. and this bios option is disabled by default.
if that's the only thing where one needs more power on 5vstb rail, then it's stupid to design psu around it.

i would say it's more about cost savings, because psu makers are just saving in any way they can. i think linear supply for 5vstb would suffice for most system built even today, given they'll probably be consuming less than 5w in standby.

fact that you've seen it on older designs probably proves why some of them are in service even today, which will be harder to see with todays' models.

[i4004]

Re: Bad design examples

what's more, if most catastrophic failures are caused by 5vstb problems (and this indeed seems to be the case) that alone would warrant putting linear supply as mandatory in the atx specs.

critical circuit (and control loop surely is) should have reliable supply.
but we can't expect it on psu that costs 10$ to boot.

overall, such mod wouldn't be too tough to do:
employ supply usually found in small ac/dc adaptors (offcourse, put good caps in it) and i think finding space for it in the psu case wouldn't be a problem.
if it would, hell, attach it on the outside of the psu.

this and one 7805 would suffice.
(or find adapter with 5v output, so you need no 7805).

cheap psu moded in this way(also all of its caps replaced with better ones) would probably yield reliability of much more expensive units.
offcourse, power output wouldn't change. it would still be 500w on the box and god knows how much really.
(1/2 of stated power usually...hehe)

[davmax]

Re: Bad design examples

First of all I must apologise in that I now understand that you were referring to the 32VDC power supply. When you quoted linear power supply I was looking for a linear regulated power supply or any circuit that has linear active components.

I have read your suggestions and think that we should really focus on what is being produced by the million and accept that this is where technology is at. The failures that occur are few and possibly relate to faulty components rather than the supply design. It does not matter what circuit is made there will always be failures. Switch mode power supplies are and can be very reliable. Why try to change for a few failures?

I did give you a reasonable explanation

Unfortunately to maintain ATX PS efficiency there must be a 300V+ capacitor and this is the only source available for the designer to create second power supply. On earlier ATX SMPSs I have seen a small stepdown 50/60Hz transformer used for secondary power. But as the 5Vstb power specification has increased for modern PCs the transformer has to be much larger so designers have gone for a small switching PS off the main DC input to keep size down and efficiency up.

I do not like the term second power supply so from here on I will refer to the "Standby Power Supply"

For you to quote a 1 amp motherboard is not a good way to go.

1. The current ATX spec for 5Vstb is 2 amps. Why would manufacturers produce less?
2. It is not wise to take 1 amp from a 1 amp supply this would increase risk, something you are trying to avoid.
3.The Standby power supply also supplies the TL494 controller chip somewhere in the range of 7 - 40V (quite a tolerance) clearly it cannot be taken from 5Vstb.
5.The proposal to use a plug pack would be best applied to the controller chip power. However it should be emphasised that we should simply ensure that quality power supplies are used and that is all. If you take the size of a plug pack it becomes clear that to supply 5V @ 2A and the controller supply the size would be too great to apply this technique inside the normal SMPS.

To summarise we need to work with what is available and not rush in to modify because a few fail in this circuit area. Cap failure has been more prominent. QUALITY POWER SUPPLIES. Not mandatory changes!!!

[PCBONEZ]

Re: Bad design examples

I see both your points and I feel they both have merit.

I'm tired of things changing all the time and I'm tired of crap parts.

But on the other hand I think questioning what is going on and looking for alternatives is something that should be continuous and never ending because if it weren't for people doing that our PSU's wouldn't be glowing from those *^%#@!! LED's, it would be from the vacuum tubes.

.

[gonzo0815]

Re: Bad design examples

I think the usual Tospwitch stand by smps is more then adequate, efficient and reliable, unless underrated capacitors are used.


From what i have seen, most quality psu and many of the cheaper ones are equipped with it.
In most cases, a failing 5v sb supply is indicated by start up issues (PSu has to be disconnected from the mains to get it running again).
This lead me to believe, that the smps controller chip is supplied by it.
If this is archived by a second winding or by simply using other controller chips is beyond my knowledge.

Cheap wall plugs and transformers are from my point of view not much more reliable then those if ever.

[Scenic]

Re: Bad design examples

(1/2 of stated power usually...hehe)

the guy who had this in his cheap-ass case had following running with it..

Athlon 64 3400+
1GB DDR400/PC3200
2x320GB HDD
DVD-ROM
DVD-RW
ATI Radeon X1650 (AFAIR)
and lots of modding crap (LED fans, CCFL lamps.. bla bla..)

it's a wonder that this 240W thingy survived for 3 months.. (then he had BSODs all the time and brought his PC to me..)