Solytech SL-8360EPS analysis + mods

[Pentium4]

Unfortunately I got rid of most of my power supplies a few years back so I hunted some on eBay. Of course I could have got some with already good caps but what's the fun in that when you can save unsuspecting victims on eBay from buying something that could cause them problems?

I'm also aware that Solytech is capable of making a decent unit, minus their caps. This one is on the 80 Plus website as being able to do its full wattage.



I bought two of these as a lot. One was made in July 2017 and one was made in September 2017. Interestingly, the only difference between the two was the primary cap.



The July 2017 had a 220uF 450V DeXon cap (no series) rated at 85C. I was surprised to pull it and see that it read 209uF and 0.22Ω ESR. (I decided to give Wurth caps a try)




The September one had a taller 220uF 450V ChengX LS cap rated at 105C. I found this very interesting because the DeXon 85C cap is bulged! As you can see there, and I am not able to press the disc down.




I was intrigued by this because I tested both of them with a paperclip and both turned on. Curious, I hooked them up one at a time to my Kill-A-Watt brick. I tested the one with the ChengX cap, it powered on and idled at a shocking read of 0W with a PF of 02.
Due to there being no minimum load resistors, the voltages were all out of whack.
12V = 12.83V
5V = 4.72V
3.3V = 3.38V
5VSB = 5.18V
-12V = -10.17V

I hooked up the one with the bulged DeXon cap, and this one read 10W draw with just a paperclip and a PF of 59. All of the rails read exactly the same as the healthy one except the 5VSB was 5.17V
So what's the deal with 10W more? Is it because the cap may be in "early" stage of failure? Is the leakage current increasing, causing the PSU to have to charge it more with the mains? I'm very curious about this since it didn't have an affect on the secondary at all (with no load at least).

Other shots:




The primary uses:
(For PFC):
x2 Toshiba TK10A60D (10A) FETs
Diodes Incorporated LTTH806RW (8A) power diode
(Switchers):
x2 Toshiba TK13A60D (13A)

Secondary:
3.3V: x1 Vishay 30V30CT (30A)
5V: x1 Vishay 30V45CT (30A)
12V: x2 Vishay 20V60CT (20A)

It looks like this PSU does actually have two 12V rails. (Pleas correct me if I'm wrong!) One of the rectifiers is just jumpered over to the other one where they both pass through the same winding on the big toroid. Then, they are jumpered to their own sections of 12V wiring but where they are jumpered from is the same pad. I'm used to seeing a giant shunt, instead it just uses two small jumpers (and a giant glob of solder lol)

In my next coming post I will show the recap and my intentions with this PSU!

Another note, what's up with this not having a current inrush limiter?? Could that have caused the main cap to bulge? Although it does have one MOV by the fuse. Should I add an NTC? I have one from a dead half bridge Solytech.

[lti]

Re: Solytech SL-8360EPS analysis + mods

I don't see any pictures, but this is an interesting one. I don't think I've ever seen a Solytech with PFC before.

The bulging cap probably is leaky, but you'd see it getting hot (and probably venting) at 10W. I haven't tried Wurth caps before, but considering the price, I would consider them for something non-critical.

Also, your title reminded me that I ran out of 22uF caps to replace CapXons. That's pretty much the only brand of caps that I still don't want to use.

[eccerr0r]

Re: Solytech SL-8360EPS analysis + mods

must have some sort of inrush limiter, the inrush limiter mainly affects your power switch from welding itself closed, also helps save your power lines from undue stress after a power outage.

Also most PSUs have lower efficiency (and crappy PF) at low drain as part of its rated power.

[Pentium4]

Re: Solytech SL-8360EPS analysis + mods

Ahhh, that was my bad, some all of the files were just barely too large. I'll reupload here:

[Pentium4]

Re: Solytech SL-8360EPS analysis + mods

I don't see any pictures, but this is an interesting one. I don't think I've ever seen a Solytech with PFC before.

I've seen a few but they are rarely actually branded Solytech. One was actually an Allied, a really nice one. I recapped it 10 years ago and it's still going! With the original YC PFC cap too.

The bulging cap probably is leaky, but you'd see it getting hot (and probably venting) at 10W.

What do you think was causing the significant increase in power?

Also, your title reminded me that I ran out of 22uF caps to replace CapXons. That's pretty much the only brand of caps that I still don't want to use.

Haha, nice. Yes I've seen lots of failed small CapXon caps. I actually ordered too many, I have about 80 Rubycon YXF 22uF 50V. They have 2021 date codes but I know I won't use them all in time. Do you want me to send you some? I will send them for free. Send me a PM!

must have some sort of inrush limiter, the inrush limiter mainly affects your power switch from welding itself closed, also helps save your power lines from undue stress after a power outage.

Do you think the MOV is being used for that purpose?

Also most PSUs have lower efficiency (and crappy PF) at low drain as part of its rated power.

Oh yeah, no doubt. I just thought it would be interesting to add

[lti]

Re: Solytech SL-8360EPS analysis + mods

I've seen a few but they are rarely actually branded Solytech. One was actually an Allied, a really nice one. I recapped it 10 years ago and it's still going! With the original YC PFC cap too.

The last Solytech power supplies I saw were the Cooler Master Elite series, which were still non-PFC half-bridge models in 2017 (with fully populated input and output filtering).

What do you think was causing the significant increase in power?

I've heard that those Kill-A-Watt meters are inaccurate at low power like that. It could just be that the failed cap affected the PFC circuit enough to start to register on the Kill-A-Watt without becoming a full 10W resistive load.

It's really nice that DeXon is a total CapXon knockoff.

Haha, nice. Yes I've seen lots of failed small CapXon caps. I actually ordered too many, I have about 80 Rubycon YXF 22uF 50V. They have 2021 date codes but I know I won't use them all in time. Do you want me to send you some? I will send them for free. Send me a PM!

I have way too many caps as well since I've really slowed down on the electronics stuff (at least outside of work). I still have some Panasonic caps from 2010. I don't think I will be taking the free caps right now, but I might in the future if you still have them. I don't think the device that still has CapXons is going to be used often enough to need recapped right away, either. Thanks for the offer, though. I wouldn't want to see them get thrown away.

[momaka]

Re: Solytech SL-8360EPS analysis + mods

Looks like Solytech has moved up the ladder in terms of build quality and design. I see this is a double-forward design and has pretty good rectifiers on the output, especially the 12V rail. For a 350W-rated PSU, this is quite nice. The output and APFC toroids also seem to use a more up-to-date 45 -type core. That being said, the output toroid does seem a bit on the small side. But it being a -45 core, it's probably just enough for about 300-350 Watts.

It looks like this PSU does actually have two 12V rails. (Pleas correct me if I'm wrong!) One of the rectifiers is just jumpered over to the other one where they both pass through the same winding on the big toroid. Then, they are jumpered to their own sections of 12V wiring but where they are jumpered from is the same pad. I'm used to seeing a giant shunt, instead it just uses two small jumpers (and a giant glob of solder lol)

Yeah, I think this PSU does indeed have two 12V rails. Well, it's a single large 12V rail split into two with "current shunts". Indeed I'm also used to seeing the thicker shunts instead of thin jumpers... but I guess if the thin jumpers present enough resistance at high current to cause a measurable drop in voltage, then they can be used just like regular current shunts. And I think I've seen this in some other PSUs as well, though I can't recall in which one right now.

I don't think I've ever seen a Solytech with PFC before.

My thoughts exactly.

But noooo!! Another manufacturer switching its PSUs to an APFC design. I like the old non-APFC PSUs better - just less things to worry about failing. The only good thing about APFC is if you live in a country with 230V AC and you get frequent brownouts. Since APFC is usually specced to work with down to 100V AC usually, the brownout must be very severe before the PSU drops out. But for 115/120V AC countries, I find APFC PSUs more fickle with brownouts an dropping out.

The bulging cap probably is leaky, but you'd see it getting hot (and probably venting) at 10W.

+1

Actually, even with no load, I don't see how the PSU can be using 0W when jumpered On. The transformer core, output toroid, and APFC toroid should all be wasting a small amount of power equivalent to a few watts each. So the PSU with the bulged DeXon cap is the one that seems to register properly on the KAW meter.

Also, your title reminded me that I ran out of 22uF caps to replace CapXons. That's pretty much the only brand of caps that I still don't want to use.

As much as I dislike CapXon myself, there are many other brands that I would consider way worse. At least the KM series seem somewhat stable if not used under much stress in a low-frequency application (i.e. audio gear and etc.) A few years back, I recapped the main board of a worthless 15" LCD with a grab-bag of recycled caps from my bin, including CapXons. But hey, anything's better than small Canicon caps that consistently read over 150 KOhms ESR

must have some sort of inrush limiter, the inrush limiter mainly affects your power switch from welding itself closed, also helps save your power lines from undue stress after a power outage.

But indeed there doesn't seem to be one.

Perhaps the manufacturer is relying on the inductance of the APFC coil to somewhat limit the initial current spike enough that an NTC is not needed?

I also have a PSU that doesn't have an NTC, and it's a non-PFC PSU too - a Thermaltake TR2 430W (XP-550np). I use that PSU in my GPU test PC, which I power cycle with the switch on the back almost every time I get it out of the closet and test a new (to me ) GPU. The switch has easily seen over 1000 cycles now. Hasn't welded or loosened yet. The primary caps are 2x 200V 680 uF units... which is substantial capacity when running with 120V AC (which is how I've used this unit its entire life.)

Also most PSUs have lower efficiency (and crappy PF) at low drain as part of its rated power.

And APFC units tend to have worse efficiency overall than non-APFC PSUs.

However, since this Solytech uses double-forward config, it's going to be a lot more efficient than their old half-bridge design any day. With H-bridge, a lot of power is wasted in the drive circuit for the BJTs on the primary... and even more power when those BJTs are under high load. Under-specced Schottk rectifiers or ultra-fast rectifiers also waste a lot of power at higher loads... hence the switch to synchronous rectifiers with MOSFETs.

Do you think the MOV is being used for that purpose?

No, the MOV (I'm assuming that blue disc-shaped component by the fuse?) is just for clamping voltage spikes on the AC input, such as abnormally high AC voltage due to a lightning strike and etc.

I've heard that those Kill-A-Watt meters are inaccurate at low power like that.

I'd say they are somewhat accurate, but just not too hard to fool.
Yes, they register reactive power fairly accurately (both inductive and capacitative loads.)
However, it's when they start dealing with non-linear loads that readings might become inaccurate. I still can't pinpoint what exactly causes this and how to reproduce the effects... but when I place a heating element in series with an ATX PSU (on the AC input), some PSUs show really really high reactive and real power. IIRC, one PSU I tested with about 100W of load on the output was showing to draw about 250 VA reactive power and 200 Watts of real power - number that just didn't add up at all, at least for the real power. My series heating element was adding only about 30 Watts to the total. So how the KAW came up with 200W of load is a mystery to me.

It's really nice that DeXon is a total CapXon knockoff.

I noticed that too!
Sad, isn't it?

[lti]

Re: Solytech SL-8360EPS analysis + mods

But noooo!! Another manufacturer switching its PSUs to an APFC design. I like the old non-APFC PSUs better - just less things to worry about failing.

Back in 2017 when these were made, Solytech was still making non-PFC half-bridge models with the usual exaggerated power ratings (the "500W" Cooler Master Elite bundled with a cheap and nasty case that somehow cost more than a Corsair 200R). Today, the only half-bridge models are probably the really cheap and nasty $10 ones (except now the gutless wonders cost $50).

As much as I dislike CapXon myself, there are many other brands that I would consider way worse. At least the KM series seem somewhat stable if not used under much stress in a low-frequency application (i.e. audio gear and etc.) A few years back, I recapped the main board of a worthless 15" LCD with a grab-bag of recycled caps from my bin, including CapXons. But hey, anything's better than small Canicon caps that consistently read over 150 KOhms ESR

Yeah, they're only audio caps. They're also from 2021, and we can only hope that CapXon has improved, considering the stuff I see them in.

Strangely, I've never needed 22uF caps in all these years.