The whole Two Transistor Thing

[KeriJane]

Hello.

I've been following the PSU threads for awhile and noticed a lot of discussion about the "2 Transistor 5v Standby" circuit common to cheap PSUs such as Bestec.


I'm currently upgrading a system with an Allied SL 8320BTX 300w PSU. It actually survived a 300 or so watt load test on my load tester (KeriProductions ATX338), and the ripple was only about 30-40mv, maxing at around 60.

This particular PSU of ill repute doesn't seem too awful, so I look inside. "YC" brand caps all over except for that one 47uf/50v near the two transistors...
It's an ASIA-X! Isn't that "Fuhjyyu"'s new name?
I put in Nichicon PW all over the secondary and in this highly suspect 2-Transistor location. It still looks pretty decent under load with no change in ripple or 5VSB and maybe less probability of a Bestec meltdown.

I've noticed similar 2-transistor designs for standby power in LCD monitors as well.

My big questions are:

1- Is there a schematic I can examine in hopes of understanding how a 2-Transistor Standby Circuit functions? I realize if that little capacitor fails it results in uncontrolled Standby power resulting in major damage. I would like to know how the thing works, but can't seem to quite grasp it by looking at schematics of complete PSUs.

2- would it be wise to put a Zener Diode across such a 5V circuit? say maybe a 5.6v or 6.2v 5w one right after the filter capacitor(s)? Would this have a hope of blowing the fuse before the magic smoke gets out? Wouldn't there be some leakage during normal operation?



Thank You to all of the VERY SMART people here that share their knowledge and wisdom,
Keri

[Hugh Chardon]

Re: The whole Two Transistor Thing

I don't have a background in designing circuitry, just in repairing it.
The following website http://www.smpstech.com/problems.htm has
been VERY helpful in isolating (pun) problems in Switching Power Supplies

[everell]

Re: The whole Two Transistor Thing

For question #1, here is a generic (typical) schematic of the two transistor 5vsb psu I have worked on. It shows optocoupler feedback which sometimes works.

For question #2, I recommend that a new thread be started to make a list of those psu that have a PWM chip instead of the two transistor design. These chips do a much better job for overvoltage and overcurrent protection. Then buy a "reasonably priced" power supply from the list of those with the better 5vsb design.

The psu with two transistor design have been around for a long time, and some of these psu last many years. I ran a POWMAX for about six years without knowing how bad they are. I was just fortunate that mine didn't blow up. But now that I know, it is history! Likewise, if you don't want unexpected trouble, the PWM chip design has much better protection.

[Wizard]

Re: The whole Two Transistor Thing

This doen't fail too easily if designed right but it leaves out overvoltage/overcurrent protection. I have this type in RCA CTC203 and ATC113, and they're not blowing up as cap gets old.

But CTC169, boy they will even it has oooold IC but it depends on 2 caps. One 2.2uF and odd size 15uF both 100V they go high in ESR (these caps sometimes I see 63V). When this small two caps go dry, TEA2261 shoots to 200V on 134V supply and IC cuts out and stay shutdown, due to the outputs, they were not in use (stays in standby) and on low voltage outputs they don't overshoot as much and through reguators is 99% safe but it takes a beating on HV multipler (direct view) and HV divider if it is used in projection tv when voltave is above 143V for awhile (this design have adjustable potimeter for standby but fixed potimeter on cold side for run voltage)
and turn on transistion from above 143V to run 143V takes little life from the HV multipler (built into flyback) or stress the HV divider little by little eventually they short out with a ozone stink and shut down.

Fix was replace four caps, and if it requires new fly or hv divider, then tv is working fine.

Cheers, Wizard

[KeriJane]

Re: The whole Two Transistor Thing

Thank you for the helpful replies.

Everell, Thank you for the greatly simplified schematic. It is still over my head but I will attempt to figure out both how it's supposed to work and how a cap failure in it could cause a Bestec meltdown.

Maybe the new thread could include things like 5VSB design, capacitor brand, EMI filtering, presence of Dummy Weights, or other obvious flaws or strengths ? This could give a little proof for assertions that PSU "A" is good or PSU "B" is bad. A quick look with the lid off or even sometimes through the fan can show a lot if one knows what to look for.

Wizard, you are aptly named!
I think maybe I might have possibly understood a little bit of that.


Feeling really stoopid,
but maybe not as stupid as before.
Keri

[Toasty]

Re: The whole Two Transistor Thing

The cap failure may be a result of, or perhaps only a contributing factor of the meltdown, not the root cause. Recent "experiments" by Everell showed that even removing the capacitor did not cause 5vsb "runaway" to 12 volts. There is evidence of other components failing (resistor and diode) that may in fact be closer to the actual cause. See the "Bestec Revisited" thread: https://www.badcaps.net/forum/showthread.php?p=69003

Don't feel stoopid... we can't all be Wizards

Toast

[i4004]

Re: The whole Two Transistor Thing

>For question #2, I recommend that a new thread be started to make a list of those psu that have a PWM chip instead of the two transistor design.

or just a suggestion
"don't buy cheapest psu you can find in the shop".
that would probably suffice...

>The psu with two transistor design have been around for a long time, and some of these psu last many years.

the thing is that MOST of them last and never cause any damage....
but when they do, it's spectacular....

that's why they're still being able to sell this crap: most work...
(ie you overall don't have many cases of psu destroying the mobo even with such crap...just think about all machines that come with cheapest psu possible...and still work after 4-5 years)

all that we can do is try to spread the word that such psus are unacceptable rubbish...

>Recent "experiments" by Everell showed that even removing the capacitor did not cause 5vsb "runaway" to 12 volts. There is evidence of other components failing (resistor and diode) that may in fact be closer to the actual cause.

no, not really....removing the cap alltogether is not interesting because that never happens...what DOES happen is those cases with ripple when emulating leaking cap...that ripple heats everything(caps included, deteriorating them further) and eventually whole circuit spins out of control to produce much higher voltage than designed...

but i agree it's bad design...design should be such that voltage actually falls as caps gradually go bad...but perhaps that would also be bad for mobo..hehe...
or better to say, design shouldn't have lytics at all...

[Toasty]

Re: The whole Two Transistor Thing

Go see the experiments before making that conclusion. It wasn't just removed. Look at the scoped readings.

[i4004]

Re: The whole Two Transistor Thing

i saw them...6ohm resistor produces most ripple...
https://www.badcaps.net/forum/showth...5&page=4&pp=20
images on previous page also show much ripple when cap is leaking(ie with r instead of cap)...
what did you see to point to semisconductors?

if they failed i would say 5v would probably be no more...
let him pull THEM out and see if it works...
<wink>
not that they use best semis they can find or anything, but bulging cap is a clear sign...

[Wizard]

Re: The whole Two Transistor Thing

i saw them...6ohm resistor produces most ripple...
https://www.badcaps.net/forum/showth...5&page=4&pp=20
images on previous page also show much ripple when cap is leaking(ie with r instead of cap)...
what did you see to point to semisconductors?

if they failed i would say 5v would probably be no more...
let him pull THEM out and see if it works...
<wink>
not that they use best semis they can find or anything, but bulging cap is a clear sign...

i4004,

It is only a experiment to characterize the circuit design under various conditions (this is all grey not black and white world). I have seen many cases where certain parts that was not quite bad but still seems to be sufficently upset the function of circuit so MUCH especially if poorly designed.
Buy Art of Electronics book and see what I mean. A good circuit should not get out of range and use common values and also in a way that it will simply stop working completely or self-limiting but very erratic if certain components gets out of range.

And note that experiment I have to point out, should tried a cap in series with resistor in different value across supply to simulate different ripple current, this is part of ESR characteriscs.

And bulging caps, SUUUURE, I had stuff that had tons of dried up capacitors that DID not bulge at all. That is why visual is not always reliable indictor of a diagnosis. Ratio of flat vs bulged capacitors is about 70:30 in CRT TV and other stuff where heat is not a factor but still bit of stress (high speed switching designs).

The reason some of them bulge and ones that did not bulge got dried out is simply byproduct of combination of bad electrolytic chemistry or abuse, heat does not matter as TVs had chassis at lowest part of cabinet and very open, thereby heat rises which is true with CRT and rear CRT projectors. in PCs PSU and mainboard are in very confined space and are blown on with warm air, MOSFET transistors/diodes adding heat and square wave into coils and both input and output capacitors taking the blunt abuse of these conditions.

Situation changes dramatically now that we are getting failed LCD/Plasma units is the cabinet are very confined and high stress due to high wattage demand on too few capacitors. Let alone those too often are low quality capacitors. Yet I see both non-bulged and bulged capacitors in these as well. This is what measuring tools are for these diagnostic work and I invest in these to speed up the checks.

Cheers, Wizard

[i4004]

Re: The whole Two Transistor Thing

wizard, in a case of 5vstby circuit if you have a bulging cap right there, what would you suspect first?
transistors and diodes? come on....

also, we don't need to guess...transistors and diodes can be pulled out and we'll see if they're good...

what i dislike about toasty's answer is that he attributes too much to the fact that
"even removing the capacitor did not cause 5vsb "runaway" to 12 volts."
i say that is not relevant at all.

relevant experiment was the one you described, good cap and a resistor in series with it...

i would say that cap takes most of the blame in most of such cases...
semis can fail too, true, but..well...show me the case where caps are completely OK and semis are messed and burned...
if you can't find that, you essentially agree with me...

oh yeah, just today i've found 5vstby design with transistors, but without lytics...gotta take some pix...

[Wizard]

Re: The whole Two Transistor Thing

wizard, in a case of 5vstby circuit if you have a bulging cap right there, what would you suspect first?
transistors and diodes? come on....

also, we don't need to guess...transistors and diodes can be pulled out and we'll see if they're good...

what i dislike about toasty's answer is that he attributes too much to the fact that
"even removing the capacitor did not cause 5vsb "runaway" to 12 volts."
i say that is not relevant at all.

relevant experiment was the one you described, good cap and a resistor in series with it...

i would say that cap takes most of the blame in most of such cases...
semis can fail too, true, but..well...show me the case where caps are completely OK and semis are messed and burned...
if you can't find that, you essentially agree with me...

oh yeah, just today i've found 5vstby design with transistors, but without lytics...gotta take some pix...

Which is good design.

Um, I didn't mean check the bulged capacitors, I just go ahead and replace these that bulged. I was just simply reminding that there are bad caps that doesn't show any external distress, these had to be checked.

Cheers, Wizard

[i4004]

Re: The whole Two Transistor Thing

yes...
i'm also of opinion that sometimes too much accent overall is put on caps alone without thinking about rest of the circuit...
BUT in this case we have clear signs on who is to blame...

we don't really see blown mobos from psus that had decent(not top caps, just decent...if that) caps in transistorized designs...

we see bulged caps and burned pcbs...and some of these take mobos with them...

[everell]

Re: The whole Two Transistor Thing

Here is a SeaSonic SS-250FS which had 5vsb measuring 12 volts when I got it. Same crappy two transistor design as the Bestec. I gutted the two transistor circuit and replaced it with the same DM311 circuit I put in a Bestec. No more problems with capacitor going HI esr and causing overvoltage.

[Per Hansson]

Re: The whole Two Transistor Thing

Could we please have some pictures showing how to identify the two transistor 5vsb vs the good design you describe?
Are there crappy 5vsb designs other than the 2 transistor model?

[everell]

Re: The whole Two Transistor Thing

Here are a couple of pictures of this psu before modification. The 5vsb circuit has all components between the two heat sinks ( in the discolored area due to the heat and burning). The smaller of the two transformers is the 5vsb switching transformer. There are two transistors next to it, one small one and one larger one (TO220 case). Usually the larger transistor is screwed to the heat sink next to the main switching transistors, but not this one.

[i4004]

Re: The whole Two Transistor Thing

per, yes there can be 1 transistor design, but those should be pretty rare.
you recognize them by locating them near the +5vstby transformer. they're on primary side. as everell says, they usually burn the pcb when going bonkers.

when you can't find them, you'll usually find the ic there, and that's a good sign.
also, check the ic you see there find it's datasheet and you'll know if that's the one.

[Wizard]

Re: The whole Two Transistor Thing

Keep in mind with poorly designed 2 or 3 transistor design SMPS, they run slightly unstable if not right and heat up excessively either transistor charactization cause it to run away or blow up especially when diode get too hot and go non-linear fooling the SMPS into going wild. Bad cap at just right moment could cause runaway.

CTC203 and ATC113 3 transistor SMPS did not blow up because it was designed well due to zero voltage crossing design and run cool (rated 140W or so). Only time mostly blew up was when either flyback transformer failed or hot glue on a coil caused solder to break dirtying the pulses to the HOT's base that both faults causing HOT to fail and they always short circuit and if not quick enough, the screeching loudly 3-transistor SMPS eventually blow, popping all 3 transistors, burning out resistor, fuse last to go.

Just my random throughts.

Cheers, Wizard

[everell]

Re: The whole Two Transistor Thing

Hello.

My big questions are:

2- would it be wise to put a Zener Diode across such a 5V circuit? say maybe a 5.6v or 6.2v 5w one right after the filter capacitor(s)? Would this have a hope of blowing the fuse before the magic smoke gets out? Wouldn't there be some leakage during normal operation?

Keri

I wasn't too sure how to answer this question earlier, but now can. I have a Fortron FSP200-60ATV which I acquired with a dead 5vsb. See attachment of drawing showing that on the output side of the 5vsb transformer there is a diode rectifier, a zener, and a PI filter consisting of capacitor coil and capacitor. The pc board was badly burned under the diodes, smoke damage and burning of the diodes.

The main diode, marked SD840 is a 40 volt 8 amp Schottky diode. It was shorted. The zener was a 4735 rated for 6.2 volts and a maximum operating current of 146 mA. It was also shorted. Both capacitors were bulging.

When the 5vsb goes overvoltage, the zener will attempt to regulate the output voltage. However, if the two transistor circuit has the power capability to maintain the increased voltage, the zener just works harder increasing its current until it fails. If it shorts, as in this case, all of the power is carried in the diode between transformer and ground. So it also shorts.

Bottom line is that the 5vsb needs something to slow it down and lower that output voltage. A zener is just a bandaid, not a solution to a 5vsb circuit out of control.

[Wizard]

Re: The whole Two Transistor Thing

6.2V zener in that manner direct to ground and direct to the voltage of 5Vsb, it is a overvoltage protection only, when 5Vsb rises too high to 6.2V threhold, zener conducts to ground either the 5Vsb smps circuit shuts down (proper design) or motorboats or whining loudly (this what happened here with poor design) eventually burning up. Not regulation.

Zener is not good idea as regulating voltage directly via a resistor from unregulated voltage/higher voltage source. Also thermal unstable and resistor watt rating holds back on voltage quality on the load if too high the voltage DROPS especially with highly variable current loading.

Proper design regulator is eithe buck switching regulator (best, cool and efficient and does have protection if get overloaded) or a linear regulator (hot and limited voltage selections, excellent for isolating noise-sensitive point of use from noisy neighbours on same supply by number of regulators to each point of use to keep heat down.) or for point of use (just one chip, use a zener & resistor at around 20mA to control a main adjustable regulator or a transistor. Again not efficient and not overload protected at all. Even the LM78xx and LM317 stated thermal overload, they doesn't work at all, they simply fail with time if overloaded or shorted out on the output.

Cheers, Wizard