Allied AL-D500EXP (SL-8500BTX) – blown primary side

[momaka]

Deer diary,

I have an Allied AL-D500EXP (SL-8500BTX) PSU with a blown fuse (T8AL), blown primary BJTs (pair of 2SC3320), and a blown 12V output rectifier (SBL4060P). What happened? – Idunno. But I suspect the PSU was loaded with more than it could handle. Given its built quality, which I made a detailed post about here (with pictures), this power supply is probably closer to being capable of no more than 300-350 Watts max. The system it was powering, IIRC, was an older and more power-hungry Intel i5 CPU bundled with a Radeon R9 290/X GPU, so the PSU was likely running right at its limits.

To fully troubleshoot a PSU like this, all of the associated parts that drive the primary BJTs (Qxx and Qyy) need to be checked. To do that, I had to remove the primary heatsink. While at it, I also removed the secondary HS to get to that blown 12V rectifier.

Here’s is Bamby’s mother in mid-surgery:
https://www.badcaps.net/forum/attach...1&d=1530222090
https://www.badcaps.net/forum/attach...1&d=1530222090
https://www.badcaps.net/forum/attach...1&d=1530222090
Given the small sizes of the components that drive those 2SC3320’s, I was almost sure I’d find more bad parts there. Miraculously, however, everything else appears okay so far. In other words, the doctors are saying she will probably make it alright.

Moving onto the secondary side, it turned out that the blown 12V rectifier also wasn’t fully blown – only one of its two internal diodes was showing short-circuit. The other had a normal diode reading. I haven’t checked it for leakage/proper operation yet. But if okay, I may just repurpose it as a 20 Amp diode for something else. Sorry, no good… err, working… parts go to waste with me. Even crap cap brands get a chance to “shine” again in cheap crap hardware.

Anyways, here is a picture of the components attached to the heatsinks:
https://www.badcaps.net/forum/attach...1&d=1530222090
https://www.badcaps.net/forum/attach...1&d=1530222090

Now, as mentioned in the built thread above, what had me puzzled the most about the PSU was this:

On that note, I don’t understand why deer has two 30 Amp rectifiers (STPS3045CT) in parallel for the 5V rail. This PSU should be meant for a modern 12V PC, so it really doesn’t need that kind of 5V capacity! Meanwhile, the 5V/12V output toroid has two thick wires for the 12V rail and only one for the 5V rail. What the #$%^, Allied?!?!

But I figured it out.
As I was troubleshooting the shorted 12V rectifier, I noticed that the 12V rail as well as the 12V rail windings on the main transformer read around 27 Ohms to ground (center-tap lead on transformer). Given that SMPS transformers don’t have too many turns, this stumped me a bit, and I thought maybe the 12V rail windings had gone bad too (which I’ve never seen happen before.)

However, that wasn’t the case at all. Turns out, this Allied PSU uses a design where the 12V rail has separate, isolated windings inside the transformer that have a center-tap which connects to the 5V rail rectified (DC) output. In other words, the 12V rail for this PSU is produced by superimposing two 7V rails on top of the 5V rail to get 12V. And that’s very likely why the 5V rail has two 30 Amp (STPS3045CWC) rectifiers in parallel – any current load that appears on the 12V rail also appears on the 5V rail. This means current pulled by the 12V rail passes through two sets of rectifiers: first the 5V rail and then the 12V rail’s own rectifier. Sounds weird and inefficient, right? – At least it does to me. As for the 27-Ohm reading I was getting - that turned out to be the 5V rail’s minimum load resistor, which basically made the 5V rail (and thus 12V rail’s transformer winding) appear as 27 Ohms to ground. Once I removed the secondary heatsink (and rectifiers), it was more easy to see this.

My only two guesses as to why the 12V rail is designed in such way is probably either to get better regulation between the 5V and 12V rails (as the 12V rail is loaded, so is the 5V rail) -OR- perhaps somehow this allows for less output windings in the main transformer (and thus saves copper… i.e. saves $$). I’m thinking it’s the former. But either way, again it looks like a rather inefficient design to me. The heat under the whole transformer, as well as the 5V and 12V rectifiers is a pretty good indicator that speaks for itself.

That aside, I also checked the other rails’ output rectifiers – they all tested okay. Other important parts to check when dealing with a PSU with a blown primary side: the secondary-side base-drive transformer driver transistors, Q5 and Q6. These are a pair of 2SC1815 in TO-92 case. Because there are a few various diodes connected to them in the circuit, it’s best to pull them out and check them again. I did that and found they were okay (normal diode junction readings between B-E and B-C for NPN transistor, no reading between C-E). The diodes connected to them in the circuit were also okay.

So all in all, it looks like I might just have to replace the 12V rectifier, primary BJTs, and input fuse… and okay, maybe recap the PSU as well, once I get it working. (That is, if I want to use it in a PC like I intend to.) This, however, will happen when I get my next parts order. I sure could pull some parts out of crappier PSUs just for a test, but I am pretty confident that replacing the above mentioned parts will get the PSU working again… or maybe not? With my luck, I never know. I’ll update this thread when I get the repair done, though.

[kaboom]

Re: Allied AL-D500EXP (SL-8500BTX) – blown primary side

Don't forget to check both sets of base drive components for the switchers.
There's two small caps (I use 4.7/50) that back-bias those BJTs during the "off" time, ensuring they switch off quickly.
There will be two 4148s also, and a total of six resistors.
A pair of similar values: 2.2-10, 22-56, 1.2k-4.7k.

See diagram:
http://www.pavouk.org/hw/en_atxps.html

These are usually rebuildable- nice of Deer to stick with a half-bridge w/ mag amp on 3.3.
Sort of like the AA5 of computer SMPSes. Other than obvious cheep-outs, it has potential.

As long as the driver and aux transformers are OK, that is.

On what they can do to a computer in stock form, or the completely fake ratings, no comment.


They stack the winding for +12 out on top of +5 to improve regulation. +5 doesn't "float away" so much when that output effectively feeds into +12. Supplies w/o this would see +5 hit the +5.25 (or higher) setpoint and the PWM backs down to keep +5 in check. But this means loads on +12 w/o loads on +5, see low voltage on +12. That "stacked" output (partially?) compensates for this. As +5 tends to rise with increasing PWM duty cycle (to keep +12 correct), the "excess" on the +5 output is "sucked up" by the lowside of the winding feeding the +12 ckt.

Delta/Newton NPS160s (and others) do the same thing.

[PeteS in CA]

Re: Allied AL-D500EXP (SL-8500BTX) – blown primary side

I'm going to guess the 12V rectifier went and that blew back to the switch transistors. I've seen that stacking technique used, and it is for improved 12V regulation.

[momaka]

Re: Allied AL-D500EXP (SL-8500BTX) – blown primary side

Don't forget to check both sets of base drive components for the switchers.
There's two small caps (I use 4.7/50) that back-bias those BJTs during the "off" time, ensuring they switch off quickly.
There will be two 4148s also, and a total of six resistors.
A pair of similar values: 2.2-10, 22-56, 1.2k-4.7k.

Yup, already checked them prior to making the above post - all pass a simple multimeter check. I might pull the diodes out and re-test again on my breadboard with a source to see if they are leaky by any chance.

As for the two 4.7 uF electrolytic caps - haven't checked those, but I will be changing them anyways. Will check ESR after replacing them though, just for fun.

These are usually rebuildable- nice of Deer to stick with a half-bridge w/ mag amp on 3.3.
Sort of like the AA5 of computer SMPSes. Other than obvious cheep-outs, it has potential.

Exactly.

If it wasn't for the bad caps and overly-inflated labels, these will outlast just about every modern APFC PSU. No SMD and specialty parts to worry about either - everything is off-the-shelf through-hole.

As long as the driver and aux transformers are OK, that is.

Good point.
I forgot to check the BJT-drive output windings. The pass-through (primary feedback for secondary) is okay, though. Given that no other parts failed on the BJT drive (both primary and secondary side), I think the traffo should be okay too. If not, I got spares that may work.

On what they can do to a computer in stock form, or the completely fake ratings, no comment.

Yeah, these Deers can be a bit "rude and crude" at times.
I'm not sure what kind of protections the PWM COTY has and how effective they are, but I do know that a lot of the older half-bridge PSUs with DBL/TL494 or KA7500 controllers usually don't play nice with low line voltages / brown-out conditions - meaning, the output typically drops without UV protection kicking in. I only found this recently on a cheap PSU, but it seems to apply to better built ones too. And OV protection is usually set so high that these PSUs, again, never trip.

They stack the winding for +12 out on top of +5 to improve regulation. +5 doesn't "float away" so much when that output effectively feeds into +12. Supplies w/o this would see +5 hit the +5.25 (or higher) setpoint and the PWM backs down to keep +5 in check. But this means loads on +12 w/o loads on +5, see low voltage on +12. That "stacked" output (partially?) compensates for this. As +5 tends to rise with increasing PWM duty cycle (to keep +12 correct), the "excess" on the +5 output is "sucked up" by the lowside of the winding feeding the +12 ckt.

Makes sense. That's what I figured too after looking at it a bit more.

I guess the rectifier voltage drop also has a significant impact on the voltage regulation. With unstacked windings design, as the 12V rail is being loaded more than the 5V rail, the 12V rail rectifier voltage drop increases (due to higher current passing through it), whereas the 5V rectifier voltage drop is not so high due to lower load. Thus, the 12V rail "sags". But with a stacked design, any current that passes through the 12V rectifier also passes through the 5V rectifier (plus whatever current the 5V rail is loaded with, too). So now the 5V and 12V rectifiers have a much more similar voltage drop. And therefore, as the 12V rail voltage "sags" under load, so does the 5V rail's. Then the PSU can compensate for both by upping the duty cycle.

Delta/Newton NPS160s (and others) do the same thing.

Yeah, you're right. Now that you mentioned it, I think I've seen this in a few Bestec single-transistor topologies, as well as a few other PSUs.

I'm going to guess the 12V rectifier went and that blew back to the switch transistors.

But how can you tell? What if the primary blew, causing a big spike to appear on the secondary and thus shorting one side of the 12V rectifier.
... though given the heat under the secondary-side rectifiers, your guess is probably the correct scenario of what happened. But that means, perhaps the 12V rectifier was passing more than 40 Amps of current?? (i.e. more than 480 Watts?) It is supposedly a 40 Amp part, so it should hold up to that load (unless the HS was running way too hot or that rectifier is a reject part not capable of its full rating?)

[sam_sam_sam]

Re: Allied AL-D500EXP (SL-8500BTX) – blown primary side

I guess the rectifier voltage drop also has a significant impact on the voltage regulation. With unstacked windings design, as the 12V rail is being loaded more than the 5V rail, the 12V rail rectifier voltage drop increases (due to higher current passing through it), whereas the 5V rectifier voltage drop is not so high due to lower load. Thus, the 12V rail "sags". But with a stacked design, any current that passes through the 12V rectifier also passes through the 5V rectifier (plus whatever current the 5V rail is loaded with, too). So now the 5V and 12V rectifiers have a much more similar voltage drop. And therefore, as the 12V rail voltage "sags" under load, so does the 5V rail's. Then the PSU can compensate for both by upping the duty cycle

Very interesting and very informative

Thanks

[PeteS in CA]

Re: Allied AL-D500EXP (SL-8500BTX) – blown primary side

momaka, it's just a guess that makes the most sense to me. The rectifier shorts, possibly due to a voltage spike. When the rectifier shorted it shorted the winding and whichever transistor was on (or turned on after a dead time period) conducted a big current spike and shorted. When the other transistor turned on it was instant cross-conduction and that transistor shorted. If the input fuse didn't blow when the first transistor shorted, it did when the second transistor shorted.