Excellent power model IT ATX 300w - shuts down after a split second

[kotel studios]

Yes, I only got 500mV one way.
Yes, I did replace the load resistors. This MM (UNIT) only has two places for probes. GND was in the GND pin off molex connectors.
I'll try again with an MM which has conections for amperage and voltage probes.

[kotel studios]

Okay new meter and it ampers on 3.3v rail got to 1.1A on MM but on adjustable psu it was under 500mA (480mA). I have set OCP and OVP for extra protection. Any ideas?

[прямо]

Okay new meter and it ampers on 3.3v rail got to 1.1A on MM but on adjustable psu it was under 500mA (480mA). I have set OCP and OVP for extra protection. Any ideas?

Doesn't make any sense, unless you don't know how to measure current with the multimeter.

[kotel studios]

All of my high quality ones just don't wanna measure, so ill just depend on current draw in adjustable psu.
3.3v - 480mA
5v immidently drops voltage to 3.8v and current is max setting (520mA)
I guess this points to 5v being bad? Although there isn't any shorts on it...

[stj]

there is a short now.
some shorts can be hidden behind mosfets or chips so they only show when the circuit is powered.

[kotel studios]

Then I am stumped cause for the life of me I cannot find it

[kotel studios]

Okay, so from the measurements on secondary side (up to the main transformator) I didn't find anything shorted on 5v. Haven't checked the PWM IC side fully, but it looks like the only culprit could be the 2003 PWM. Maybe the 5vsb damaged it when the rail had bad caps, albiet still letting it "power" on?

On another thought I should maybe continue with the testing on 12v rail. Maybe that ones screwed and makes the 5v rail look funky?

[kotel studios]

After some thinking I got an idea to remove the resistor which connects 5v to 2003 PWM IC and then back feed the 5v rail and the PWM IC and see if the short is there. Although I have no idea if this could damage other components, so I'll ask here if its okay to do what I wrote.

[momaka]

Okay new meter and it ampers on 3.3v rail got to 1.1A on MM but on adjustable psu it was under 500mA (480mA). I have set OCP and OVP for extra protection. Any ideas?

Yeah, agreed, that's not right. Either the bench meter or the "new" meter is not measuring right.
That said, the adjustable bench PSU meter does appear to be showing the correct current draw. Remember, the 3.3V rail has a 6.8 Ohm load resistor, so you should see approximately 485 mA of current draw... which you did!

If the new meter is a cheapo, rotate the dial several times. A lot of the cheapo ones have terrible contacts on the dial, and sometimes rotating it quickly multiple ways clears those up (at least for a little while.) If you still can't get both meters to display the same current from the 3.3V rail, just ditch the 2nd meter and connect the bench PSU straight to the Excellent ATX PSU. Also, set the current limit a little higher - 2-3 Amps, just so we can see if anything weird is going on or not.

5v immidently drops voltage to 3.8v and current is max setting (520mA)
I guess this points to 5v being bad? Although there isn't any shorts on it...

Well, this PSU has a weird circuit consisting of two diodes (D28 & D29) and a 0.39 Ohm 3W resistor (R73) all in series going from the 5V rail to the 3.3V rail. This is an additional "dummy load" that dumps power from the 5V rail to the 3.3V rail (why? - I don't know... but it's silly and I do have it removed on my working B300ATX). So two diodes drops @ more load = ~1.6V... so 5V - 1.6V = 3.4V left. Divide that (3.4V) by 0.39 and you get up to 8.72 Amps to flow to the 3.3V rail... but there is NO load on the 3.3V rail, other than its 6.8 Ohm load resistor (and thus, a maximum of 480 mA load on this rail.) But also remember that the 5V rail has a 15-Ohm resistor. 5V / 15 Ohms = 0.333 A = 333 mA. So the total current when you try to feed back 5V into the 5V rail will be 333 mA + 480 mA = 813 mA.

With that said, I will advise again that you set the bench PSU current limit a bit higher. 2 to 3 Amps should be better.

With that done, now you can try the voltage feedback test for each rail I mentioned in post #16 and then see if the Excellent ATX PSU can be turned On when jumper its PS_ON (green wire) to ground. And please report back what the PSU does when you do these tests for each rail. Use your 2nd meter to measure the voltage on the output rails, other than the one you're feeding back with the bench PSU. We want to see if the voltages go up to a higher value... or if even the Excellent ATX PSU stays on now.

there is a short now.
some shorts can be hidden behind mosfets or chips so they only show when the circuit is powered.

True... but this PSU is simpler than a rock... so no MOSFETs or chips that can do this kind of thing.

Haven't checked the PWM IC side fully, but it looks like the only culprit could be the 2003 PWM. Maybe the 5vsb damaged it when the rail had bad caps, albiet still letting it "power" on?

Possible, but unlikely.
The "2003" PWM controller appears to take power through pin 1 through the 5VBS rail... which passes through R36 (a small 10-Ohm 1/8W) resistor, which would likely have showed signs of burning up if the 5VSB voltage was going too high.
I also suspect the totem-pole driver circuit for the BJTs likely also should be OK, because if it wasn't, you'd get pretty much no output voltages at all (all mV range tops) when you tried to turn on the PSU... and that's not the case - you get up to a volt or two on some of the rails, which suggests PWM IC is working, but protection is shutting it down.

Nevertheless, it doesn't hurt to check the BJT driver circuit either. It consists of transistors Q5 and Q6 (both 2sc945 or equivalent in TO-92 case), These are driven directly by the PWM IC through pins 7 & 8. Since the PWM IC driving pins 7 & 8 are open-collector outputs, they are pulled up with resistors R33 and R34, which themselves take power from the (secondary) aux. rail of the 5VSB traffo (provided by D9 and smoothed by electrolytic cap C15). With that said, pull out and check electrolytic cap C15 first. I believe it should be a 50V 47 uF, though I have forgotten what value it is and I can't see it on my unit without taking it out. Also check that R33 and R34 are good out of circuit (lift one leg and measure resistance... should be 1 KOhm resistors both.) And finally, check Q5 and Q6 by pulling them out of circuit and measuring hFE with a meter. Other components to check that are also part of the driver circuit: D10 and R24 (supply power from sec. aux. rail to the totem-pole circuit), D12 and D16 (protection diodes for Q5 and Q6, respectively), and the small electrolytic cap next to C15 (ground bypassing for totem-pole drivers Q5 and Q6), along with D19 (grounding diode for totem-pole circuit, found behind Q6.)
Let me know what you find for all of these components.

[kotel studios]

5v goes to 0.8v while injecting 3.3v, normal 0.2v.
Rest of the lines stat the same while injecting.
Oh an 5v now draws 785mA after setting max 1.5A.

I am starting to think I'm just huggng a dead horse and I should focus on that fsp300-60gtf or tagan tg380-u01 with shorted 5v rail due to my stupidity... But for now we'll see what can be done about this cause I am out of non-firehazard PSU's.

[momaka]

Oh an 5v now draws 785mA after setting max 1.5A.

Yeah, that's more like it.
Do these tests also for the -5V and -12V rail. However, I'm not sure if you will be able to use your bench PSU for this or not. In particular, if your bench PSU can't do negative voltage OR if it uses a grounded plug, then you can't use it. Or rather, you would have to be careful about it if you still want to use it to backfeed into the -5V or -12V rail. More precisely, you would have to plug it into an UNGROUNDED outlet so that its ground output is floating. Once that is done, then you connect ground of the bench PSU to either the -5V or -12V rail and its positive voltage output lead to ground of the Excellent ATX PSU. That way, when you set the bench PSU to output 5V, it will output -5V into the -5V rail of the ATX PSU.
I personally like to do this with 5V and 12V ungrounded adapters: simply plug adapter's (+) into GND of ATX PSU and GND of adapter into -5V or -12V of ATX PSU.
The repeat the test to see if the PSU powers up.

If nothing different happens when you try to backfeed these two, then there is certainly something else going on.

I am starting to think I'm just huggng a dead horse and I should focus on that fsp300-60gtf or tagan tg380-u01

Well, before you bounce back and forth, perhaps we really should finish concluding that this PSU is a dead horse with a few more tests. One of them that I like to do can definitely turn it into a dead horse if something goes wrong.

But before I suggest that test, please confirm the status of the components of the BJT driver circuit I mentioned above in my previous post.
Also, if you haven't already done so, replace the two small electrolytic caps on the primary side close to the BJTs (I think these are 1 uF or 2.2 uF or 4.7 uF... I don't remember exactly.)
All of these can cause this issue of the PSU to not want to turn On.

[kotel studios]

Q6 hFE is 292 while Q5 is . D19 is fine, same as the ones near it. All electrolythic caps replaced with good ones. Even with the primary side ones, but no go. Still dies after a second.

What I don't get is why someone from trw says these are fire hazards. Sure, they ain't built like good quality fsp's or seasonics but these ones aren't cheaped out like some I saw.

[kotel studios]

I think I found the culprit. Injecting -12v makes the testing psu die (an working firehazard but for backfeeding -5/-12v rails its enough, didn't wanna risk killing my only good psu). Rechecked the backfeed connection, no issues.
Any ideas where to go next?

[kotel studios]

Q6 hFE is 292 while Q5 is . D19 is fine, same as the ones near it. All electrolythic caps replaced with good ones. Even with the primary side ones, but no go. Still dies after a second.

What I don't get is why someone from trw says these are fire hazards. Sure, they ain't built like good quality fsp's or seasonics but these ones aren't cheaped out like some I saw.

Q5 has 246 hFE. R34 and 33 are fine too.

Since everything looks to be ok in the -12v circuit I have an idea to remove the diodes connecting it to d14 and try to power the psu on. Not sure if that'll work tho...

[momaka]

I think I found the culprit. Injecting -12v makes the testing psu die (an working firehazard but for backfeeding -5/-12v rails its enough, didn't wanna risk killing my only good psu). Rechecked the backfeed connection, no issues.
Any ideas where to go next?

Interesting.
So you had another working ATX PSU which you took -12V from and fed it into the Excellent ATX PSU here, correct?
If so, before concluding anything else, check if that working ATX PSU can actually supply anything from the -12V rail without dying - i.e. put something like a 100-Ohm resistor between it's -12V output and ground to see if it can power that up. Also, I'm assuming that you're making the connections between the PSUs first and then you try to power up the working ATX PSU, correct? Reason I'm checking on this is because I want to rule out inrush current from caps as the reason for the PSU to trip.

Speaking of caps, when you replaced the cap for the -12V rail, did you connect it right / follow the polarity orientation on the board?
Just as a sanity check, for the caps on the -12V and -5V rails the POSITIVE (+) side of the caps connects to PSU GROUND. And the negative side (stipe) on each cap connects to the respective -12/-5V rail. If you get this wrong, that could certainly be why the PSU is tripping, as this would cause the electrolytic cap to be powered up backwards and draw lots of current. So please verify this is correct.

Since everything looks to be ok in the -12v circuit I have an idea to remove the diodes connecting it to d14 and try to power the psu on. Not sure if that'll work tho...

Perfect, this is exactly what I was going to suggest next if all of the above I mentioned checks out. It will work, but only if you back-feed the PSU with -12V afterwards. Otherwise the PWM controller will be looking for it, see that it's missing, and still shut down the PSU.

So, basically, remove diodes D20, D22, and D23 - all of these are part of the -12V rail rectifier circuit. Now try back-feeding the Excellent ATX PSU with -12V again, and see if the issue persists.
As far as I know, there is nothing else on the -12V rail that can load it down... so if your working ATX PSU still trips, there must be something else going on.

Also, worth mentioning is that you could use your bench PSU to provide -12V, but ONLY if your bench PSU is NOT GROUNDED on the AC plug.
To do that (feed -12V from your bench PSU into an ATX PSU), you set the bench PSU to output 12V, then connect its positive (+) to the ATX PSU's ground (black) wires. Next, take the negative (-) output from the bench PSU and connect it to the -12V rail (blue wire) of the ATX PSU. But again, for this to work, your bench PSU must not be grounded on the AC side. One way to know if your bench PSU is grounded or not is to look at the AC plug and see if the grounding pins connect to the (-) output of bench PSU. Since you're in Poland, I assume you standard Schuko type AC plugs, just like most places in Europe. On the Schuko plug, the two prongs that stick out are for the AC input - Line and Neutral. And the two metal pins that are recessed onto the outside perimeter of the Shuko plug are grounds.

What I don't get is why someone from trw says these are fire hazards. Sure, they ain't built like good quality fsp's or seasonics but these ones aren't cheaped out like some I saw.

Funny you mention this. To be completely honest, I think there may be some truth to that. On my 1st unit that I bought back in 2005 and used for about 2 years (with lots of system hangs and restarts, as it was a very cheaped out L&C version), it finally had its floppy wires melt, along with part of the wires leading up to the Molex connector. I never understood why this happened, since the floppy drive I had it connected to was check and found to be OK... as was the CD drive the Molex connector was plugged into. So what made the wires short to each other and melt is still a mystery to me... and more importantly, why the PSU didn't shut down. Luckily, this happened just as I was powering up the computer, and I had the sides off for better cooling. So as soon as I heard the CD drive not start up and a weird hissing noise, I looked at the PC and already saw smoke starting to pour out of those wires. I reacted quickly and unplugged the PSU, so no more damage was done. Ever since then, I fixed up that PSU with better/more caps, but have not used it to power up anything, other than a DC motor or two once or twice.

With that said, my personal jury is still out whether these are fire hazard units or not. The 2nd unit I got recently has many caps bulged and I haven't even checked if it works yet... but I plan to fix it specifically so I can thoroughly test out its protections in order to satisfy this curiosity that I've had about this particular model of L&C. I suspect the "2003" PWM chip may be a bit "basic" in the protections department, and possibly isn't to catch overloads on the output, unless it's something very severe or a hard short-circuit... but we will see when I get mine recapped and tested (hopefully either this or next week if I get a chance.)

[kotel studios]

Filter caps are ok. Whats weird is that injecting -12v (33mA draw) as you said now makes the excellent power turn on for a sec and generate voltages, but still dies and fails to start after shorting pson. And yes, working psu can power on with 100Ω resistor.
This adjustable psu is grounded completely and I don't have any ungrouded mains outlets here (luckly).
Now I don't know where to go....

[momaka]

Filter caps are ok. Whats weird is that injecting -12v (33mA draw) as you said now makes the excellent power turn on for a sec and generate voltages, but still dies and fails to start after shorting pson.

Hmmm. we're making progress?? Maybe??
33 mA going to backfeeding the -12V sounds right.
Most of that current (almost 30 mA) comes from R64, which is a 510-Ohm resistor on my PSU going between -12V rail and ground. Next to it is R63 - also a 510-Ohm resistor - which is used for the negative rails sensing circuit (which then connects to pin #5, "PP" of the PWM IC through a few more components - namely R58 and small signal diode D24.)

I was looking at my PSU and various half-bridge PSU schematics, and came to a few interesting findings.

First, the "2003" PWM chip appears to actually be pin-pin compatible to ATX-2005 / ATX2005 / ATX-2005z chips (but NOT to be confused with AT-2005b, as that's a different PWM controller.) Well, not sure how that helps here exactly, but if one can be found online for purchase or if you have one from another PSU, at least you have more options now.
*EDIT* I found this old thread back from 2014 about these PWM chips:
https://www.badcaps.net/forum/troubl...2003-2005-2012
What I find interesting is that when I back-traced my PSU (with a 2003) chip, the pinout completely matches that of the one shown for the ATX2005_2005z.pdf datasheet there. But apparently, Deer's "2005"/"2012" chips are slightly different. Anyways, still worthy info, I suppose.

Next check resistor R31. It's a small resistor (found between the PWM controller and 3.3V rail wires) that connects pin #6 of the PWM IC and the 12V rail. Essentially, it allows the 12V rail to get detected/measured by the PWM controller to see its value. On my unit, this one is 47 Ohms. See what yours is and if it measures OK (preferably with one leg lifted.)

Then another backfeeding test... but this one a bit different.
-- Plug in the Excellent ATX PSU to AC power, but do NOT turn it on. Just back-feed 3.3V from your adjustable PSU into its 3.3V rail, and measure what voltage you get on pin #11 (VREF). Report back what you get. This test essentially checks the 3.3V rail shunt regulator input voltage is correct.

Lastly, something that I noticed on my unit: some of the component leads were left quite long and look dangerously close to shorting each other. When working on this unit, I do think that it would be possible for some of these leads and wires to get shorted together as the unit is moved around on the bench. So my suggestion is to check all of the component leads and make sure none are shorting to each other or to a nearby solder joint that they shouldn't be shorting to.

After some thinking I got an idea to remove the resistor which connects 5v to 2003 PWM IC and then back feed the 5v rail and the PWM IC and see if the short is there.

Looking back at some posts above, which resistors was this that you removed and did you place it back.

This adjustable psu is grounded completely and I don't have any ungrouded mains outlets here (luckly).

Ungrounded outlets are nice for cases like that. ... though may be a bit dangerous in some scenarios.
If your bench PSU's internal power supply is derived via a line-type (linear) AC transformer, then running it in an ungrounded AC outlet should be safe. On the other hand, if it uses a SMPS and has Y2-safety caps between Live and Ground as well as Neutral and Ground, that could get a bit "iffy"... in a sense that the PSU's ground-referenced parts will now float at 1/2 of your AC line voltage (230/2... or about 115 V AC!!!), which could potentially damage sensitive electronics... and even you might even feel it if you touched that with one hand and have any other part of your body touching something that is grounded. With "standard" capacity Y2 safety-rated caps (2-10 nF), though, the current should be pretty low (so not a shock hazard). However, I still do *NOT* recommend to run such PSUs ungrounded, unless you're totally aware of what you're doing.

..
..
And if none of this goes anywhere, I do have one final and possibly destructive (for the PSU) test... though maybe not if done with a dim bulb test ... but essentially, it involves directly driving the faulty (Excellent ATX) PSU's primary side with another working ATX PSU of the same topology (half-bridge with BJTs.) This way, if there is a faulty component, it will either heat up, burn out... or you should see which rail(s) (if any) are misbehaving. If none, then the issue narrows down to either PWM IC being bad or other signal/driving circuit issue and not a major power component issue. But I'll detail that in another post, if/when needed. Let's see the results of the above tests first.

[kotel studios]

And if none of this goes anywhere, I do have one final and possibly destructive (for the PSU) test... though maybe not if done with a dim bulb test ... but essentially, it involves directly driving the faulty (Excellent ATX) PSU's primary side with another working ATX PSU of the same topology (half-bridge with BJTs.) This way, if there is a faulty component, it will either heat up, burn out... or you should see which rail(s) (if any) are misbehaving. If none, then the issue narrows down to either PWM IC being bad or other signal/driving circuit issue and not a major power component issue. But I'll detail that in another post, if/when needed. Let's see the results of the above tests first.

I think an even better idea would be to isolate the rail sense thingies (apart from 5vsb, that one looks to be okay) on the PWM and inject the lines from a good PSU directly to the PWM (soldering wires). Then try to power on. I think this would also serve the same as your idea but not sure...

But before we go the "destructive" and "brute-force" way I'll do what you suggested and come back with the results.

[kotel studios]

Hmmm. we're making progress?? Maybe??
33 mA going to backfeeding the -12V sounds right.
Most of that current (almost 30 mA) comes from R64, which is a 510-Ohm resistor on my PSU going between -12V rail and ground. Next to it is R63 - also a 510-Ohm resistor - which is used for the negative rails sensing circuit (which then connects to pin #5, "PP" of the PWM IC through a few more components - namely R58 and small signal diode D24.)

I was looking at my PSU and various half-bridge PSU schematics, and came to a few interesting findings.

First, the "2003" PWM chip appears to actually be pin-pin compatible to ATX-2005 / ATX2005 / ATX-2005z chips (but NOT to be confused with AT-2005b, as that's a different PWM controller.) Well, not sure how that helps here exactly, but if one can be found online for purchase or if you have one from another PSU, at least you have more options now.
*EDIT* I found this old thread back from 2014 about these PWM chips:
https://www.badcaps.net/forum/troubl...2003-2005-2012
What I find interesting is that when I back-traced my PSU (with a 2003) chip, the pinout completely matches that of the one shown for the ATX2005_2005z.pdf datasheet there. But apparently, Deer's "2005"/"2012" chips are slightly different. Anyways, still worthy info, I suppose.

Next check resistor R31. It's a small resistor (found between the PWM controller and 3.3V rail wires) that connects pin #6 of the PWM IC and the 12V rail. Essentially, it allows the 12V rail to get detected/measured by the PWM controller to see its value. On my unit, this one is 47 Ohms. See what yours is and if it measures OK (preferably with one leg lifted.)

Then another backfeeding test... but this one a bit different.
-- Plug in the Excellent ATX PSU to AC power, but do NOT turn it on. Just back-feed 3.3V from your adjustable PSU into its 3.3V rail, and measure what voltage you get on pin #11 (VREF). Report back what you get. This test essentially checks the 3.3V rail shunt regulator input voltage is correct.

Lastly, something that I noticed on my unit: some of the component leads were left quite long and look dangerously close to shorting each other. When working on this unit, I do think that it would be possible for some of these leads and wires to get shorted together as the unit is moved around on the bench. So my suggestion is to check all of the component leads and make sure none are shorting to each other or to a nearby solder joint that they shouldn't be shorting to.


Looking back at some posts above, which resistors was this that you removed and did you place it back.

R58, D24, R31 are good. Both resistors show whats written on them (R58 written 520Ω, reads 540Ω in circuit, R31 written 41Ω, reads 46Ω in circuit).
When backfeeding 3.3v on pin 11 of PWM, 48mV jumps to 59mV for a sec and then back to 48mV.
Double checked the pins for suspicious/shorted and found nothing.
I think my lc b300atx had the 2005 PWM IC? Not sure, gotta find the leftovers and check. Maybe it wasn't killed by 5vsb rail.

[momaka]

When backfeeding 3.3v on pin 11 of PWM, 48mV jumps to 59mV for a sec and then back to 48mV.
Double checked the pins for suspicious/shorted and found nothing.

Hmmm.. that's not right.
Are you sure you measured pin 11 (maybe you're getting it confusing with another one when you flip the PCB)? If you are sure you measured pin 11, let's simplify this test a little bit and re-do it. I already did it on my PSU just to verify, and the results were as I expected them to be. So here's how to proceed:

1) Instead of having PSU plugged in the AC, just feed 5V into the 5VSB rail (probably best to use another ATX PSU, because you will be injecting another voltage concurrently, soon.)
2) Verify that pin #1 of the PWM IC is getting 5V - the same voltage that you read on your 5VSB line (please specify in your response what it is too, as this might skew the results a little.)
3) Inject 3.3V into the 3.3V rail and verify that you can read 3.3V on any of the orange wires when doing so.
4) Check the voltage of pin #11 (VREF) of the PWM IC and report back what it is. It shouldn't be in the mV range. It should be in the Volts range.

I think my lc b300atx had the 2005 PWM IC? Not sure, gotta find the leftovers and check. Maybe it wasn't killed by 5vsb rail.

Maybe.
At the very least, the primary BJTs out of that one might come in handy for testing the Tagan PSU... but that's another story.

Both resistors show whats written on them (R58 written 520Ω, reads 540Ω in circuit, R31 written 41Ω, reads 46Ω in circuit).

Just FYI: 520 Ohms and 41 Ohms are not standard values.
On my PSU, R58 is [blue, red, brown, gold].... 620 Ohms (a standard resistor value according to E24 standard codes).
And R31 I have as [yellow, purple, black, gold]... or 47 Ohms... which your reading of 47 Ohms confirms it's probably fine.

Another testing tip: resistor usually almost always show lower resistance in circuit... or at most, the same value that is specified by its color code (and maybe plus the added tolerance... e.g. a 100 Ohm resistor can read as high as 105 Ohms if it's 5% tolerance type.) Now, if a resistor does test higher resistance in circuit than what is specified by its color code (and tolerance spec), then it probably is bad or marginal. But again, if you read lower resistance, that's usually OK.

I think an even better idea would be to isolate the rail sense thingies (apart from 5vsb, that one looks to be okay) on the PWM and inject the lines from a good PSU directly to the PWM (soldering wires). Then try to power on. I think this would also serve the same as your idea but not sure...

Well, no, this won't work, because once you insert "good" voltages into the PWM IC, it won't bother telling the primary side to run (much... if at all). So this won't test neither the primary side power components, nor the secondary side power components.

On the other hand, that "destructive" test I suggested is different. By using the PWM driving signals of another PSU (and injecting them directly at the two totem-pole pins of the middle transformer), the Excellent PSU's primary side BJTs will be instructed to run according to the other PSU's signals. This means that even if there is a fault on the secondary side of the Excellent ATX PSU's, it will not see it and not care, because there's essentially no feedback. The usefulness of this approach is that you will see whether the primary side is capable of switching and staying running... and if the secondary side power rectifiers will produce any voltages. The voltage rails might not be at their proper values (i.e. 5V might be 4V or even 3V or 7V... and likewise with the 12V)... but they should all at least be proportional to each other - e.g. if 5V is 20% higher (6V), then you can expect the 12V rail to also be 20% higher (14.4V). Only the 3.3V may regulate properly, due to being an independent regulation (mag-amp) circuit... unless its supply line (5V rail unrectified) drops too low - then it may not be able to regulate at 3.3V. But either way, all of these rails should output something / some voltage and be stable at it.

But let's leave the "destructive" test for now. Even with the dim bulb device, there's a lot that can be misconfigured to go wrong.