Power supply build quality pictorial. part 2

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

New unit, Caps are the usual parts box from CWT, though w/o AVL...
Fan is a Yate Loon D80SH-12B -> 12V, 0,7A



I kinda like the unit, not too loud under load (though only ~20°C Room Temperature right now).

BUT, there is one thing that I really do not like: Semi-Fanless without Hysteresis...
And of course the usual 8pin Protection Chip by Sitronix...

[momaka]

Re: Power supply build quality pictorial. part 2

Ah, Y U NO include label/model info?!

That thing looks tiny - pretty sure that's not ATX.
And semi-fanless without Hysteresis? Does that mean fan goes On and Off like crazy when the PSU reaches that perfect temperature "sweet-spot" for the fan temp. trigger?

FYI, I did that once with a video card BIOS mod (silly noob mistake, obviously!) At a certain point, the video card fan would go near full-tilt and then settle down, then full tilt, and then back down - rather annoying.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Ah, Y U NO include label/model info?!

Lets say its a CWT SFX PSU - similar to Enermax Revolution SFX.

That thing looks tiny - pretty sure that's not ATX.

Correct, its a real SFX.

And semi-fanless without Hysteresis? Does that mean fan goes On and Off like crazy when the PSU reaches that perfect temperature "sweet-spot" for the fan temp. trigger?

No.
It means that that always happns, even at higher loads.

IIRC the Sirfa/Silverstone Implementation has a load part, where the fan goes always on at a certain load.

This unit dosn't have that. In normal Idle operation, the fan starts - for a second or two, then stops - for a second or two, repeat.
Even under high loads that happens a couple of times...

But it gets worse. When the fan starts/stops there is a motor noise thingy with a hissing (brrr - hui) -- wich makes it louder in semi fanless than in true normal mode...

What do I need to get rid of that? 2 Germanium diodes??

FYI, I did that once with a video card BIOS mod (silly noob mistake, obviously!) At a certain point, the video card fan would go near full-tilt and then settle down, then full tilt, and then back down - rather annoying.

That sounds like FX5800 ULTRA BIOS, that had two states:
Off and full, IIRC.

Those Fan controls on early nVidia Cards were just absolute garbage - and controlled by the driver by state (ie 2D mode or 3D Mode)...

[japlytic]

Re: Power supply build quality pictorial. part 2

VSN-011 Arcade Power Supply (+5V/7A, -5V/1A, -12V/1A)

All capacitors are QRT brand unless otherwise noted.
100uF 400V Chemi-Con primary side filter.
3300uF 10V x2 before +5V inductor with 3300uF 10V after it.
470uF 25V before -5V regulator (7905) with 10uF 16V after this regulator.
470uF 25V before -12V regulator (7912) with 10uF 25V after this regulator.
2SC3153 primary side switcher and BR36 input bridge rectifier.
FR304 rectifiers for negative rails with 15FWJ11 rectifier for +5V rail.

Creepage between primary and secondary sides seems a bit too close even though the secondary is meant to be grounded (ignore the crack on the board in the corner).

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

That Sharp PC613 seems to be an optocoupler. The 2SC3153 is a BJT designed for use in switching power supplies, and rather beefy (6A) for just 51W or 52W (as are the heatsinks). The BR36 is rated for 3A, overkill again. The lack of an output inductor suggests this is a self-oscillating flyback design. The BJT switch places the switch frequency in the 20KHz-30KHz range.

I couldn't see any date codes, but that layout looks like it was done with tape on vellum rather than CAD. So I'll guess it's early-mid 1980s vintage. It has no safety marks and minimal EMI filtering (there may have been filter components elsewhere in the system), which is consistent with the close primary-secondary spacing.

[momaka]

Re: Power supply build quality pictorial. part 2

This unit dosn't have that. In normal Idle operation, the fan starts - for a second or two, then stops - for a second or two, repeat.
Even under high loads that happens a couple of times...

But it gets worse. When the fan starts/stops there is a motor noise thingy with a hissing (brrr - hui) -- wich makes it louder in semi fanless than in true normal mode...

LOL, what a horrible design!

I'd just run the fan on 5V or 7V all the time, if that's good enough (probably would be).

OR....

What do I need to get rid of that? 2 Germanium diodes??

Any two diodes will do. Well, maybe preferably Schottky type for one of them to run the fan off of 5V all the time and then the other to connect to the controller, in case it does decide to spin faster. But in all honesty, if that stupid fan controller still cycles the fan On and Off even at high load, you might as well just permanently hook the fan to 5V or 7V and forget all the controller BS.

Sometimes shit designs are not worth figuring out and keeping - just remove and make it simple.

That sounds like FX5800 ULTRA BIOS, that had two states:
Off and full, IIRC.

Yes, the FX5700 and FX5800 were the first to do that, IIRC. But nVidia used that 2-speed design pretty much all the way up until the GeForce 8x00 series showed up (which started using a 4-pin PWM fan design). I have both GeForce 6x00 and 7x00 cards that only have two fan speeds, and it is indeed dependent on whether the card switches to "2D" or "3D" mode. (And sometimes, the clocks for 2D and 3D modes are the same and only the fan speed varies).

Despite the two-speed-only design, I wouldn't call it "garbage". At least the controller had hysterisis. And with the right heatsink mod, you can actually manage to get a cool and quiet card. I usually put Xbox 360 CPU heatsinks on these with a 60, 70, or 80 mm low-power fans. They run cool and because of the low-power fan, there's not much difference between 2D "slow" fan mode and 3D "full-speed" mode, as the fan speed doesn't change that much.

VSN-011 Arcade Power Supply (+5V/7A, -5V/1A, -12V/1A)...

Looks pretty well built for its power rating. Classic flyback design, as PeteS noted.

I couldn't see any date codes, but that layout looks like it was done with tape on vellum rather than CAD. So I'll guess it's early-mid 1980s vintage.

The vent-less radial caps also give that away, I think. Most early 70's gear, at least from my limited experience, seems to be done with axial caps and very few radials. In the 80's, radial electrolytics became more popular. But they were still vent-less for the most part. On that note, I got some old AT/XT PSUs off of Craigslist (looks like mid-80's stuff, too) and they appear to be built in a similar way. Going to post them when I sort through the pictures.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

LOL, what a horrible design!

Exactly.
And that's what for example the Enermax Revolution SFX and probably all based on that design do...

Its a bad Marketing implementation.
And I'm thinking about getting a scope because looking at the waveform would be interesting I believe.

Any two diodes will do. Well, maybe preferably Schottky type for one of them to run the fan off of 5V all the time and then the other to connect to the controller, in case it does decide to spin faster. But in all honesty, if that stupid fan controller still cycles the fan On and Off even at high load, you might as well just permanently hook the fan to 5V or 7V and forget all the controller BS.

Naa, its not permanently at high load, only for the first minute or two, until the heatsinks are saturated and actually heatsinking. Until then it starts/stops a couple of time And I already thought about using a Schottky Diode or rather a double one in a TO220 Case. A bit heatshrink on top and good to go. Problem is that its an SFX and its really tight.
But that is not the only issue I have in my mind...
Because of the way the fan sounds when starting and stopping, I think that the fan controller gives the fan a little push and then drops dramatically.
Though I need a scope to verify that...

Sometimes shit designs are not worth figuring out and keeping - just remove and make it simple.

If I had space, I'd use a radical method: Just install a Topower Fancontroller.
Problem is: I have no space, as you can see above. The only space I have is between the Primary and +5VSB transformer and that's about it...

And I don't want to cut into the fan cable. And the fan controller sits on the same board as the +12V Rectifier...

Yes, the FX5700 and FX5800 were the first to do that, IIRC. But nVidia used that 2-speed design pretty much all the way up until the GeForce 8x00 series showed up (which started using a 4-pin PWM fan design). I have both GeForce 6x00 and 7x00 cards that only have two fan speeds, and it is indeed dependent on whether the card switches to "2D" or "3D" mode. (And sometimes, the clocks for 2D and 3D modes are the same and only the fan speed varies).

And the funny thing is that its BIOS controlled (or assisted).
If you put the FX5800 ULTRA BIOS on the FX5800 it acts like an Ultra.

So it stops the fan when idling and revs it up under load.

The crap thing is that the stock Heatsink is garbage. And the FX5800 ULTRA one especially. I was able to get an FX5800 ULTRA Heatsink and could take a look at it. And god deserves the designer/product manager designing this heatsink to be hung by the balls....

a) the fan was tiny. And to make matters worse: Its not that loud. If you hold it in your hand, IIRC its on par with most smallish heatsink graphics card fans of the time (GF4 TI Stock Cooler is more annoying than this small thing)...
b) if the fan doesn't make much noise, where the heck does the noise come from? and that's resonance, and the Airflow, the crap way its designed and possibly vibrations...

But that's not the worst part. The worst part is that the surface area of the "Heatsink" is tiny. And people owning the card have found that the small Revoltec copper GPU Cooler is superior to the nVidia Design...
this one:
https://www.amazon.de/VGA-Kühler-f-G.../dp/B0002IA3AW

Despite the two-speed-only design, I wouldn't call it "garbage". At least the controller had hysterisis. And with the right heatsink mod, you can actually manage to get a cool and quiet card.

No, its garbage because there is no temperature control. And even with more modern cards (IIRC it was 400 or 500 series), they still didn't do a good job.
The AMD thing is foolproof. I controlled the fan of my 7970 in a bad way and then the fan starts spinning at 100% because the GPU was at around 94°C or so. nVidia doesn't seem to have this control and you can actually kill the GPU with low fan speed - wich they did themselves under some circumstances...

It might be the 2D/3D Clock bullshit and that the card thought ir ran in 2D with 2D Fanspeed but was running at full power in the World of Warcraft menus...

And with the digital fan control, either the card is too loud and could be quieter or it might actually overheat and you have no headroom at all.

TL;DR:
Its garbage because there is no temperature element to the regulation of the fan...

Going to post them when I sort through the pictures.

I really like to see that =) =)

[pdavid]

Re: Power supply build quality pictorial. part 2

Sector IBP-200-N AT power supply from '97-'98. Young Year OEM model IBP-42c REV 3.

Maybe it should go to the guttles topic due to all the corner cuttings and missing parts in construction.

Fan is using some clamp cutouts to be held in place.
No EMI filter, just two unrated ceramic caps.
Bifilar coil traces are bridged.
No inrush current limiter. Traces bridged.
No bridge rectifier just some diodes tangled together.
Genuine 105C NCC primary caps!
13007 primary swithers in To-220F capsulation bolted to the housing.
No dedicated heatsinks for the primary switchers nor the secondary diodes.
Transformer is EI type 28 size.
494 PWM controller.
D83-004 rectifies 5V bolted to housing.
5GG2C rectifies 12V bolted to housing.
Missing PI coil on -5 and -12V rail.
Some transistors for secondary protections and power good generation.
Lot of smd passives.
18AWG wires.

There is space for bigger main trafo and output components... even the EMI parts can be built in.
I'm going to keep this unit since the distributor was one of the infamous firms in the country offering super low end and high end parts both during the 90's.

[momaka]

Re: Power supply build quality pictorial. part 2

Sector IBP-200-N AT power supply from '97-'98. Young Year OEM model IBP-42c REV 3.

Maybe it should go to the guttles topic due to all the corner cuttings and missing parts in construction.

Yes, 100%.
I know it's an old *AT* PSU that probably would never see such high loads on its rails... but it's still a turd. If they at least mounted the primary and secondary-side silicon onto some minimal heatsinks, I'd say it's a passable AT PSU.

Fan is using some clamp cutouts to be held in place.
No EMI filter, just two unrated ceramic caps.
Bifilar coil traces are bridged.
No inrush current limiter. Traces bridged.
No bridge rectifier just some diodes tangled together.

Proves that old stuff doesn't always mean well-made, I suppose.
Also proves that crap manufacturers will always be crap.

Genuine 105C NCC primary caps!

LOL.
My guess would be that cheap crap caps weren't just as widespread back then, so PSU manufacturers often had to use Japanese ones. I have an old Leadman (the manufacturer of PowMax) with genuine NCC/United Chemicon caps too. Of course it's still a crap PSU otherwise.

I'm going to keep this unit since the distributor was one of the infamous firms in the country offering super low end and high end parts both during the 90's.

In that case, I hope you have some time to rebuilt it (i.e. add all the missing parts and definitely attach those rectifiers and BJTs onto some heatsinks - I just can't bear to look at them "naked" like that.)

I really like to see that =) =)

Still working on it, haha.

No, its garbage because there is no temperature control. And even with more modern cards (IIRC it was 400 or 500 series), they still didn't do a good job.
The AMD thing is foolproof. I controlled the fan of my 7970 in a bad way and then the fan starts spinning at 100% because the GPU was at around 94°C or so. nVidia doesn't seem to have this control and you can actually kill the GPU with low fan speed - wich they did themselves under some circumstances...

Temperature control is irrelevant if you have a well-designed and proper heatsink, in which case, a 2 to 3-speed controller will be just fine and can still be very quiet while keeping the GPU cool at all times.

Whenever I do my GPU cooler mods, I always make them such that the GPU stays under 60°C at 100% load in the highest possible room temperature I expect throughout the year (typically 30°C in the summer). So when winter comes, I get a proportional drop of 10-12°C drop in GPU temperatures as well.

New GPUs often simply are not matched with adequate heatsinks to keep them that cool, be it with or without a temperature-controlled fan.

And 94°C? LOL, I wouldn't call that foolproof at all. Consider your card on its way to the grave. It may not be that soon, but it will happen eventually.

These AMD GPUs are just known to become weak even just past 55°C. And anything above 65°C is a guaranteed death for them down the road. Not that nVidia is much more different.

[Y. Signal]

Re: Power supply build quality pictorial. part 2

Mean Well RT-125A

All the capacitors I can see are Rubycon and Chemi-con, except one small Jamicon near the outputs.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

New unit, Caps are the usual parts box from CWT, though w/o AVL...
Fan is a Yate Loon D80SH-12B -> 12V, 0,7A



I kinda like the unit, not too loud under load (though only ~20°C Room Temperature right now).

BUT, there is one thing that I really do not like: Semi-Fanless without Hysteresis...
And of course the usual 8pin Protection Chip by Sitronix...

Update about that Beauty:
I've made an extension Cable with a Schottky Diode from an old PSU in TO220 Package (MOSPEC 45V IIRC). Sadly didn't make any pictures, but you get what was done...

And the 3rd leg of the Schottky Diode I've soldered to the 3,3V Rail of the PSU (on the modular ATX Connector).
And its rather quiet, the fan rotates constantly...

[momaka]

Re: Lite-ON PS-6311-2d2 [Dell L305N-00]

Another plain PSU post. This one is a Dell L305-00 305 Watt PSU. OEM is Lite-ON, model PS-6311-2d2. It's for Dell's older BTX-style PCs like the venerable OptiPlex 755, Dimension E510/520 and etc. Actually, I think this one came out of a Dimension E5150 tower. Anyways, time for pictures:
https://www.badcaps.net/forum/attach...1&d=1541734913
https://www.badcaps.net/forum/attach...1&d=1541734913
https://www.badcaps.net/forum/attach...1&d=1541734913

Not much to this unit's cabling: 24-pin ATX, 4-pin 12V CPU power, 2x SATA power, 2x Molex drive power, 1x floppy, and 1x Dell proprietary connector (not used) - all 18-AWG wires, except for the floppy. This PSU claims dual 12V rails with total combined power of 264 Watts (22 Amps.) That's more than enough for a low/low-mid range PC with an HDD or two. And like many older PSUs, it also has a fairly strong 5V rail, rated for 22 Amps as well. Now let's see what's inside to back those numbers.

https://www.badcaps.net/forum/attach...1&d=1541734913
https://www.badcaps.net/forum/attach...1&d=1541734913
https://www.badcaps.net/forum/attach...1&d=1541734913

Meh, that's a pretty standard OEM unit there ^: single-transistor forward design with a 5VSB offline switch IC. In terms of build quality, this PSU has good input EMI/RFI filtering, decent size input caps (they are Elite brand, though), okay heatsinks, proper-sized transformers, and very decent output filtering. There are also indeed two 12V rails, as stated on the label. These come from a single 12V rail that is then split with current shunts. I presume the PS224 supervisor (on the secondary side) handles this duty. However, what's not so good is that the 12V rectifier before the rails split, is a *single* STPS20H100 in TO-220 package, rated for 20 Amps. With STF topology, there is NO WAY that part will output 20 Amps on the 12V rail… much less 22. Other OEM PSUs from this era such as Dell and HiPro that typically also use a 20A part on the 12V rail, only rate it for 14 to 16 Amps. So I'd say Lite-On is lying a bit here.

Of course, you can probably already see where the problem was with this unit. No? Have a look at the output caps more carefully again, then. One of them is bulged (an OST RLP). I got a bit suspicious of the others and proceeded to remove the rest of the secondary-side caps. I'm glad I did – close to half of them tested bad without showing any signs of bulging. All of the small “startup” caps on the primary tested bad too! Folks, this is why you don't ignore crap cap brands.

Aside from that, the PSU also has quite a bit of the tan glue that goes brown/black (and eventually conductive.) Nevertheless, I cleaned a big portion of it from the primary side, because I do plan on recapping this unit. While at it, I also removed the primary caps to clean some of the glue behind them. Good thing I did, because a lot of primary-side components between these caps and the heatsink were covered with the glue. There is also a 2.2 uF 450V Ltec TY electrolytic cap next to those large input caps. I got curious and investigated what it does. Believe it or not, this is part of the main PS primary-side snubber network. Now, I can't claim that I am a PSU expert, but I've opened quite a few PSUs and have never seen an electrolytic cap used on the primary-side snubber before (and certainly not with that capacity either). Since I removed it to clean some glue, I put it on my ESR meter like the rest. It read high-ESR (around 20-25 Ohms). Oh joy!

Next, the solder side: it looks okay, aside from a few long leads:
https://www.badcaps.net/forum/attach...1&d=1541734913

We also have some funny business going on with current shunt resistor, R227. It's installed on a ground trace, but is bypassed by the trace. What the heck, Lite-ON? The shunt does read 0.00 Ohms even on my component tester, though. So perhaps it's just there for extra current support (as the trace thins a bit in that location).

Last but not least, a fan label shot:
https://www.badcaps.net/forum/attach...1&d=1541734913
It's an ADDA model AD0812HS-A70GL, rated for 0.25 Amps at 12V. It spins well. Then again, I don't think it has seen much stress. Reason I say this is because the PSU has a temperature-dependent fan controller that runs the fan at only about 4 to 4.5V @ 20°C room temperature. I tried playing with a potentiometer hooked across the NTC thermistor on the heatsink, and I couldn't get the fan to budge past 6V. I think this PSU may require a fan controller mod too, but more on that in another thread. Here is the fan controller circuit, by the way:
https://www.badcaps.net/forum/attach...1&d=1541734913

That's all. I'll make a new thread for the recap with a list/diagram, in case anyone wants to recap one of these. Looks like a solid 305 Watt unit, so why not? I just don't quite trust the 12V rail power ratings.

Lastly, a component summary…

Primary Side

* 5x Y2-class caps (2x 1 nF, 2x 2.2 nF, 1x 0.47nF)
* 2x X2-class caps + 2x common-mode chokes
* 8 Amp 250V fuse, type T; 15sp 2R5 NTC thermistor, 1x MOV (+ to – pri. DC bus)
* GBU8k bridge rectifier (8 Amps, 800V)
* 2x Elite GM input caps: 200 V, 680 uF, 22 x 42 mm (ø x h)
* 1x FQA11N90 MOSFET for main PS (STF topology)
* Main PS snubber: RCD type (BYV26EGP diode, 2x 100-KOhm 2W parallel resistors, 2.2 uF 450V Ltec TY capacitor 10 x 13 mm)
* Startup + PWM Vcc cap: 1x Ltec LZP, 35V, 220 uF, 8 x 15 mm
* 35 mm –wide main transformer, 20 mm -wide 5VSB transformer

ICs
* UC3843bn primary-side PWM controller
* PS224 secondary-side supervisory
* TNY267PN offline switch IC for 5VSB generation
* four “817” optocouplers (2x for main PS + 2x for 5VSB)
* LM7912 regulator for -12V rail

Secondary Side
3.3 V rail (mag-amp regulated)
* D83-004 (TO-3P) Schottky rectifier (30A)
* 1x OST RLP, 6.3V, 2200 uF, 10 x 20 mm before PI coil (bulged)
* 1x OST RLP, 6.3V, 3300 uF, 10 x 25 mm after PI coil
* PI coil: 4 mm core, 6.5-turn, 14(?) AWG wire
* 4x 120-Ohm parallel SMD resistors (30 Ohms equiv.) for minimum load

5 V rail:
* 2x STPS2045CT (TO-220) Schottky rectifiers in parallel
* 2x OST RLP, 10V, 2200 uF, 10 x 25 mm with PI coil in between (1x slightly bulged, both bad)
* PI coil: 4 mm core, 6.5-turn, 14(?) AWG wire
* 4x 470-Ohm parallel SMD resistors (117.5 Ohms equiv.) for minimum load

12 V rail:
* STPS20H100CT (TO-220) Schottky rectifier
* 1x Ltec LZP, 16V, 680 uF, 8 x 20 mm before PI coil
* 1x Ltec LZP, 16V, 2200 uF, 12.5 x 25 mm after PI coil
* 2x Taicon HG, 16V, 120 uF, 6.3 x 12 mm, after each current shunt
* PI coil: 5 mm core, 15.5-turn, 16(?) AWG wire
* 75-Ohm, 2-Watt resistor for minimum load (1.92W of heat dissipation!!! )

-12 V rail:
* 1.5-2 Amp diode rectifier feeding into a 7912 regulator
* 1x Ltec LZP, 25V, 470 uF, 8 x 20 mm cap on 7912 reg. input
* 1x Ltec LZP, 50V, 47 uF, 6.3 x 12 mm cap on 7912 reg. output
* 1x 1-KOhm SMD resistor for minimum load

5 VSB rail:
* SB340 diode (as rectifier)
* 1x OST RLP, 10V, 1500 uF, 10 x 20 mm cap before PI coil
* 1x Ltec LZP, 10V, 470 uF, 8 x 13 mm cap after PI coil (bad – showing 400 uF and ~2 Ohms ESR)
* PI coil: 4 mm core, 11-turn (?), 20(?) AWG wire

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

If it weren't for the 20A 12V rectifier and the crap-caps that could be a credible 305W PSU. It's over 12 years old, though, so whatever it was used in, it gave very decent life (assuming it wasn't sitting in somebody's basement or attic for 10 years). The 3843, MOSFET switch, and -52 material output inductors suggest a switch frequency north of 50KHz.

That 75 ohm, 2W pre-load is nuts. Both for the power "wasted" and for running it so close to its rated power. Maybe consider a 100 ohm (or even 150) 2W part?

The bulging OST crapacitor is pocketed be tween two other caps and between two nice warm inductors. Great component selection!

[momaka]

Re: Power supply build quality pictorial. part 2

Well folks, I did a nearly full recap on this PSU. All of the pertaining info can be found in this thread:
https://www.badcaps.net/forum/showth...812#post861812

If it weren't for the 20A 12V rectifier and the crap-caps that could be a credible 305W PSU.

Agreed.
Given the system it was powering, though (a Dell Dimension E5150 with a "standard" 89W TDP Pentium 4 Prescott CPU), it was probably enough for what it is.

As for the crap caps... they're gone now, for the most part.

It's over 12 years old, though, so whatever it was used in, it gave very decent life (assuming it wasn't sitting in somebody's basement or attic for 10 years).

Well, I don't have a history of the PC it came out of, as it came from a repair shop that was scrapping this PC.

But given the dust accumulation both inside the computer it was in and the PSU itself, I can say it probably provided at least 5 years of service. Also, the motherboard in that PC had nearly every KZG cap bulged, and the board was almost exclusively done with KZG. So it's possible these helped the PSU caps fail sooner. Of course, it could be the other way around too. I guess when dealing with crap cap brands, one can never tell "whodunit first" .

Speaking of failed-with-no-visible-signs caps inside the PSU, here's an awesome collage of that (also shown in the recap thread above):
https://www.badcaps.net/forum/attach...1&d=1541973750

So now OST and Ltec are always on my remove-and-check (and likely recap) list when doing PSUs. I already had suspicions about the Ltec caps after recapping a 300W Delta PSU (where, again, I had a few "okay"-looking caps that turned out to be bad). CapXon is on that list too (except possibly for their KM series, since they always appear to bulge when they go bad).

That 75 ohm, 2W pre-load is nuts. Both for the power "wasted" and for running it so close to its rated power. Maybe consider a 100 ohm (or even 150) 2W part?

Allied/Deer/L&C did that on a lot of their PSUs. 100-Ohm 2W resistor on the 12V rail (1.44 Watts dissipation) will still run hot enough to discolor the PCB and output wires. You really want to go up to at least 270 Ohms to keep the dissipation at sane levels (around 0.5 Watts is okay). But I just removed that resistor altogether. Should be okay as long as the PSU is connected to a motherboard or there is some device to load down the 12V rail.

The bulging OST crapacitor is pocketed be tween two other caps and between two nice warm inductors. Great component selection!

And it probably doesn't get much airflow/ventilation from the fan.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

...
Allied/Deer/L&C did that on a lot of their PSUs. 100-Ohm 2W resistor on the 12V rail (1.44 Watts dissipation) will still run hot enough to discolor the PCB and output wires. You really want to go up to at least 270 Ohms to keep the dissipation at sane levels (around 0.5 Watts is okay). But I just removed that resistor altogether. Should be okay as long as the PSU is connected to a motherboard or there is some device to load down the 12V rail.
...

Back in ancient times Boschert's standard practice was to mount resistors 1/2 watt and greater .1" or .2" off the PCB. But as long as there is some reasonable constant load on the 12V (= not just HDDs) what you did should be OK.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

New unit, Caps are the usual parts box from CWT, though w/o AVL...
Fan is a Yate Loon D80SH-12B -> 12V, 0,7A



I kinda like the unit, not too loud under load (though only ~20°C Room Temperature right now).

BUT, there is one thing that I really do not like: Semi-Fanless without Hysteresis...
And of course the usual 8pin Protection Chip by Sitronix...

You people have missed something interesting.
The +5VSB don't use the Bridge Rectifier, has its own Diodes for Rectification...

[pdavid]

Re: Power supply build quality pictorial. part 2

IMB PC compatible power supply from the late 80s Hungary.
75W total power. Currently not functional. I dissasambled it and several parts proved to be faulty.
The construction and parts used is very reminiscnet of the era.
There are some "western" parts since analogs were too difficult for cccp to produce so it had to be licensed or bought if possible...

Some 220V line emi filter parts:
Siemens 2x47nF capacitor in some sealed and rounded barrel type enclosure.
Bifilar choke of unknown parameters.
5Ohm NTC for current limiting.

Primary:
Discrete 2Amp diodes forming a graetz bridge.
2x350V 100uF input buffer capacitors.
BU326A primary transistors in To-3 package.
UC3524AN pwm controller.

Aux:
On a seperate board a small transformer and some diodes, resistors, capacitors.
Also on another board the feedback circuit consists of a small transformer and some passives set up around the main pwm. On the other side some of the secondary output capacitors and the negative voltage rail regulators are located.

Main transformer as an EE core 35mm type.

Secondary:
5V: Rectified by BYS28-45 in To-247 package.
12V: Rectified by BYS254-45 in to-220 package and fed to two 7812 regulators.
The positive voltage outputs have seperate current stabilising coils.
-5V: From a separte transformer tap a diode rectifies then feeds an 7905 regulator.
-12V: Same tap connected to another diode that feeds an 7912 regulator.
KT206/400 thyristor is used as an overcurrent/short circuit protection/crowbar.

There is a full aluminium Sunon fan running on 230V AC.
Electrolitic capacitors are MM/Videoton brand of various types. Most of them are still holding up. Some of the small polimer ones are shorted. Ceramics are ok. The 5V and 12V snubber circuit resistors are a bit cooked.


I'm planning to fix this unit and and see how it performs

[momaka]

Re: Power supply build quality pictorial. part 2

Wow, we've come a long way with switching power supplies since then.

Be careful before applying full voltage to any of the original electrolytic caps. They all likely need to be removed and reformed separately. Otherwise, I'm pretty sure they are still okay, as old caps have thicker aluminum foil plates, and thus tend to hold up okay even after a long time. Electrolyte probably isn't in the best condition, though, so they might be a little "leaky".

I'm also surprised only "350V" caps were used for the input. That's obviously cutting it a bit close, as a 230V AC line comes out to about 325V DC.

Anyways, really cool PSU. Keep us posted how it goes with the repair.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

Wow! That almost looks like it was hand built by a hobbyist! I started working with switching power supplies in 1980, and I'll just say that what was produced by Boschert and Astec around that time looked much nicer. Boschert's XL75 (75W) was a discontinuous flyback design. The 3524 type PWM was usually used for half bridge or full bridge designs.

Silicon General's SG3524 and TI's TL494 were probably the first PWM ICs on the market, starting in the mid 1970s. The panel displays in the first generation F/A-18 Hornet used military versions of the 3524.

[Sparkey55]

Re: Power supply build quality pictorial. part 2

Wow, we've come a long way with switching power supplies since then.

Be careful before applying full voltage to any of the original electrolytic caps. They all likely need to be removed and reformed separately. Otherwise, I'm pretty sure they are still okay, as old caps have thicker aluminum foil plates, and thus tend to hold up okay even after a long time. Electrolyte probably isn't in the best condition, though, so they might be a little "leaky".

I'm also surprised only "350V" caps were used for the input. That's obviously cutting it a bit close, as a 230V AC line comes out to about 325V DC.

Anyways, really cool PSU. Keep us posted how it goes with the repair.

Back in the Old Days it was common to see 175 VDC rated power supply caps that ran from the 117-120 VAC lines.