Power supply build quality pictorial. part 2

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Here something for you to discuss.
Since attachements aren't enabled yet, pichosted...






Shot with me good old Nikon D5100 (and the kit glasses)...

[Behemot]

Re: Power supply build quality pictorial. part 2

Why again? Nothing new there. It's been reviewed several times over.

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

This thread needs its attached pics back!

[Per Hansson]

Re: Power supply build quality pictorial. part 2

This thread needs its attached pics back!

It's being worked on:
https://www.badcaps.net/forum/showthread.php?t=51582

[Compgeke]

Re: Power supply build quality pictorial. part 2

Here's something fun I found in the garage. If you're going to do it, might as well do it wrong. No idea when it was made, but it was sometime in the mid 70s - 80s most likely.

It's really as basic as you can get. Transformer, bridge rectifier (made with 4 individual diodes), capacitor and a 7812. As you can see in the pictures, despite being really basic it's really poorly built. Everything's just floating around, there's leads nearly touching the casing, the case isn't grounded, etc. I might remake it on some breadboard but then again I already have a decent 12V PSU.

[kaboom]

Re: Power supply build quality pictorial. part 2

Here's something fun I found in the garage. If you're going to do it, might as well do it wrong. No idea when it was made, but it was sometime in the mid 70s - 80s most likely.

1975 date code on the cap.

It's really as basic as you can get. Transformer, bridge rectifier (made with 4 individual diodes), capacitor and a 7812. As you can see in the pictures, despite being really basic it's really poorly built. Everything's just floating around, there's leads nearly touching the casing, the case isn't grounded, etc. I might remake it on some breadboard but then again I already have a decent 12V PSU.

That's not the problem; the fact that they're too small is what matters.

"What's wrong?" What isn't?


3A transformer with 1A diodes, no heatsinking for the very vintage 7812. Shock hazard wiring. No bypass cap from output to common.

Couldn't draw 200mA from it without overheating the regulator...

[momaka]

Re: Power supply build quality pictorial. part 2

Here's something fun I found in the garage.

Lol . Shouldn't this be in the gutless, bloated, and fried power supply hall of shame thread?

1975 date code on the cap.

Heh, I have a several Sprague caps from that era as well (most with datecode 76xxL) - all came from an oldschool HP bench power supply. Not sure if I should try reforming them or not. Who knows for how long they sat.

Speaking of which, I need to make a thread for that beast. Would be nice if I could fix it up and use it.

[mariushm]

Re: Power supply build quality pictorial. part 2

That transformer's not 3A ... more like 0.5A at best.

Anyway... 12.6v ac rms rectified is about 1.414 x 12.6 = ~ 17.7v dc peak and then there's 2x voltage drop on diodes so you're left with about 16v dc peak
and then you can figure out the max current based on the capacitor used (5000uF) because

C = current / ( 2 x frequency ac x (vdc peak - vdc min) )

so since the 7812 needs at least 13.5v to output 12v you can go backwards
0.005 Farads = current / ( 120 x (16-13.5) ) => current = 300 x 0.005 = 1.5 Amps

So you're looking at best 1.5 amps which is well within the 7812's capability.
At 1A the linear regulator will dissipate about (13.5v-12v) x 1a = 1.5 watts and considering the 7812 has a thermal resistance of about 65c/w (junction-air), then the regulator without a heatsink would be 65x1.5 = ~97c above ambient temperature .... really close to max 125c operating range.

In conclusion, like I said, it's unlikely this psu was even close to 1A ... I'd say more like 0.5A max.

[eccerr0r]

Re: Power supply build quality pictorial. part 2

Was staring at that picture, it looks like one of the transformers I physically own... If that was an Archer/Radio Shack 273-1511 transformer (printed on the other side), that really does have the heft to be a 3A transformer, secondary wiring looks like it's at least 22 gauge. Unfortunately the diodes and the 7812 can't handle it...

[mockingbird]

Re: Power supply build quality pictorial. part 2

Here's an OLD power supply I've had here sitting for a re-cap for quite a few years now, the HiPro HP-P3087F3. It's a very early 24-pin HiPro, predating the more common HP-D3057F3x.

When I found this, the fan was completely seized.



As you can see by the brown discoloration on the PCB, things got pretty toasty before it was put out of its misery. I wonder if it will even power on after the re-cap. Some rudimentary measurements show no shorts anywhere.



Looking at the labeled bag of the original caps, I see mainly OST RLP series and Teapo SC. Both the RLP and SC are bloated, unsurprisingly.

I removed the heatsinks along with the transistors so that I could apply some fresh silicone grease... I did a thorough job, removing all the old grease between the pad and heatsink and between the transistor and pad, re-applying a fresh coat for all of them.

Please correct my terminology if I err, as I'm still sort of learning how PSUs work.

For the primary switcher, we have the STW9NK90Z:



On the secondary heatsink, we have the following transistors:

1) 2 x MBR2045CT. I assume these are for the dual 12V rails?

2) FYP1010DN (Not sure what this is for)

3) STP40NF03L (3.3V rail?)

4) STPS20H100C (Not sure what this is for)

5) STPS3045CW (5V Rail?)





On to the modular boards:



The bottom board sits in the primary section. On it we have the following:

1) TNY266P - for the 5VSB Standby

2) UC3843AN - SMPS Controller

3) LM393N

On the top board which sits in the secondary section we have the Weltrend WT7517 supervisor chip.

One thing I like about this power supply is that they've provided for 12.5mm capacitors in the secondary... But they've done this at the expense of pi filter coils and have instead used current shunts... I would need an oscilloscope to check if the better 12.5 caps I'm planning on putting in will make up for the lack of coils (there are only two).

[Wester547]

Re: Power supply build quality pictorial. part 2

That's a good power supply, mockingbird.

Not surprised to find the fan seized. ADDA sleeve bearing fans bear great notoriety for lacking the proper lubricant from the factory, but once oiled and maintenanced well, they're excellent and quiet fans. Nor am I surprised that the Teapo SCs and OST RLPs (especially the former) bloated when the fan seized long enough to cause that amount of discoloration to the PCB.

The two MBR2045CTs are actually for +5V rectification. The dual +12V rails are separated by way of the OCP shunts on the output. STPS3045CW is the +3.3V rectifier and STP40NF03L is the MOSFET that linear regulates the input (whatever that is) down to +3.3V. Being on the output of a forward converter, the FYP1010DN handles the pulses from the transformer and the STPS20H100CT is responsible for freewheeling. Yeah, it looks like Hipro used two capacitors in parallel for the +12V rail in lieu of a capacitor-input filter.

Here's another blast from the past. Yet another 200W NPS-200PB-73M, revision 01, made in China (Newton Power was one of Delta's Chinese subsidiaries, although the last NPS-200PB I posted was made in Thailand). manufactured in the 10th week of 1999. On the label, it reads 22A on +5V, 18A on +3.3V, a whopping 6A on +12V, 1.2A on +5VSB, 0.5A on -5V, 0.8A on -12V (0.8A combined), and 135W max for the combined +5V and +3.3V rating.

As one may guess, this PSU has the dreaded proprietary connectors from the old days of D(H?)ell. The capacitors are all Rubycon and Nippon Chemi-con on this one, SMH on the input and YXB and LXJ on the output, with small KMEs and SMEs (mostly KMEs) scattered about the unit. 470uF 200V 22x35 SMHs on the input, and very nice heatsinks as well. Bridge rectifier is a D3SB60 (4A, 600V). Main switcher is a Toshiba 2SK2746 (single forward converter), +5VSB switcher (yes, this unit is old enough to use the two transistor +5VSB circuit with an 100uF 25V 6x15 NCC SME being the "critical" capacitor) is a freestanding STD1NK60. The primary controller is DNA1001D. Main transformer is size 35. As far as input filtering goes, we have two MOVs, one NTC thermistor, three coils, two Y capacitors, two X capacitors, etc. It employs three optocouplers by General Semiconductor. As to the secondary side, +5VSB is rectified by a SB360 axial diode. +12V is rectified by a single STPR1020CT. +5V is rectified by MBR3045PT, +3.3V by STPS3045CW, and -5V is linearly stepped down from the -12V rail with a L7905CV on the secondary heatsink. The main toroid uses -52 material and the +3.3V toroid -26 material. The secondary side is supervised by DNA1002C. Output filtering:

+5VSB: 1000uF 10V 10x16 Rubycon YXB before and after the filtering coil.
-5V: 10uF 50V 5x11 Chemi-con KME after the regulator.
-12V: 1000uF 16V 10x30 Chemi-con LXJ.
+3.3V: 10x20 1500uF 10V Rubycon YXB before the filtering coil with 12.5x20 2200uF 10V Rubycon YXB after.
+5V: 10x20 1500uF 10V Rubycon YXB before and after the filtering coil.
+12V: 12.5x30 2200uF 16V Chemi-con LXJ and 8x11.5 330uF 16V Rubycon YXB in parallel.

All connectors are 18AWG excepting the floppy connector (22AWG). There's a daughterboard full of components on the secondary, although nothing on the underside. Inside I found an ASB0812HH Delta sleeve bearing rated for 3250RPM @ +12V @ 0.30A. Unlike the last NPS-200PB I had, this Delta fan actually had some lubricant left (although the other one was still working rather well regardless, and lubing it didn't change much). It spins very fast and moves copious amounts of air on the default controller, although that's still a far cry away from its full speed.

[momaka]

Re: Power supply build quality pictorial. part 2

As you can see by the brown discoloration on the PCB, things got pretty toasty before it was put out of its misery. I wonder if it will even power on after the re-cap.

It should. It's a HiPro!

One thing I like about this power supply is that they've provided for 12.5mm capacitors in the secondary... But they've done this at the expense of pi filter coils and have instead used current shunts... I would need an oscilloscope to check if the better 12.5 caps I'm planning on putting in will make up for the lack of coils (there are only two).

No, they have everything right on there, I think.
You only see one big toroid, because the 3.3V rail is linearly generated (hence no 3.3V output toroid and no 3.3V saturation coil near the main transformer). That big toroid does all of the rails.

One of the PI coils is for the 5V rail, which is fine and as on any other PSU. The other is likely for the 3.3V rail (though it could also be for the 12V rail, I'm not sure). Either way, I see spots for 7 or 8 output capacitors. With most of them being 12.5 mm, you can fit 3300 - 4700 uF caps easily for each rail, which should be plenty to keep the ripple very low.

Here's another blast from the past. Yet another 200W NPS-200PB-73M, revision 01, made in China

Nice PSU!
I have a Lite-On from that same Dell era with the proprietary ATX connector as well. It has all Japanese capacitors, just like yours. Still works like a champ. I don't remember if I posted pictures of it here or not, so let me see that. I also have another "blast from the past" coming up as well. Pentium 4 has actually been wanting to see if for a while now, so I will try to post it sometime soon.

[mockingbird]

Re: Power supply build quality pictorial. part 2

The two MBR2045CTs are actually for +5V rectification. The dual +12V rails are separated by way of the OCP shunts on the output. STPS3045CW is the +3.3V rectifier and STP40NF03L is the MOSFET that linear regulates the input (whatever that is) down to +3.3V. Being on the output of a forward converter, the FYP1010DN handles the pulses from the transformer and the STPS20H100CT is responsible for freewheeling. Yeah, it looks like Hipro used two capacitors in parallel for the +12V rail in lieu of a capacitor-input filter.

Thanks for clearing that up and helping me to learn.

One of the PI coils is for the 5V rail, which is fine and as on any other PSU. The other is likely for the 3.3V rail (though it could also be for the 12V rail, I'm not sure). Either way, I see spots for 7 or 8 output capacitors. With most of them being 12.5 mm, you can fit 3300 - 4700 uF caps easily for each rail, which should be plenty to keep the ripple very low.

There are two unpopulated spots on the very edge of the board in the center for two more 10mm caps. I traced the top one to the 12V rail and the bottom one to the 3.3. So I'll be adding caps there...

By the way, I got my parts for the capacitor re-former, so I have to get started on that at some point.

edit: One odd little detail I wanted to add. The fan's positive wire went to the positive lead hole for the unpopulated 10mm capacitor in the 12V rail. It's odd because there is actually a place on the board marked for the fan, and the negative lead goes there, so I'll move the positive lead over there, and it's in parallel with a small electrolytic. So I wonder why they didn't do it that way to begin with.

There's a thermistor that's attached with a bracket and screw to the heatsink, I wonder if that's only for protection or also for fan speed. But I doubt that, because there's no discrete fan controller board here like I see in FSP units, and I didn't recall seeing anything regarding a fan controller in the Weltrend.

[Wester547]

Re: Power supply build quality pictorial. part 2

Nice PSU!
I have a Lite-On from that same Dell era with the proprietary ATX connector as well. It has all Japanese capacitors, just like yours. Still works like a champ. I don't remember if I posted pictures of it here or not, so let me see that. I also have another "blast from the past" coming up as well. Pentium 4 has actually been wanting to see if for a while now, so I will try to post it sometime soon

Thanks. I look forward to what you have to post.

And that Hipro definitely has a fan controller. The thermistor is for exactly that. Hipro's ADDA fans (and/or Cheng Home Electronic/Superred fans) wouldn't be so silent if there was no controller and they were directly wired to +12V (those two 100V rectifiers are for the +12V rail, to be clear). Lti found that when that thermistor wasn't touching the heatsink in Hipro PSUs, the fan would no longer spin even when well lubricated. The thermistor could also exist for over temperature protection.

[kaboom]

Re: Power supply build quality pictorial. part 2

Mockingbird, you've got a fan controller on that board.

To the right of the fuse and AC connector in your pic, there are two metal oxide resistors and a Q601. The 300-series parts in this area are for +5VSB feedback and compensation; the 600s control the fan, and possibly sends thermistor signals to a comparator (that LM393) for overtemp shutdown, if this unit has it.


C612 goes from one of the +12 outputs to the collector of the fan's pass transistor (Q601). This explains why the controller still worked when the fan got its +12 from the unpopulated cap spot. One of those resistors shunts the transistor, keeping the fan running at a "minimum speed." Sometimes, in other PSUs, there's a small electrolytic elsewhere in the (base?) ckt for that transistor, to kickstart the fan at high speed for a second or two when +12 is first applied- overcomes sticky grease or oil.

If you tied the black wire from the fan to any common/GND, it'll run full-speed.

[Wester547]

Re: Power supply build quality pictorial. part 2

Mockingbird, you've got a fan controller on that board.

To the right of the fuse and AC connector in your pic, there are two metal oxide resistors and a Q601. The 300-series parts in this area are for +5VSB feedback and compensation; the 600s control the fan, and possibly sends thermistor signals to a comparator (that LM393) for overtemp shutdown, if this unit has it.


C612 goes from one of the +12 outputs to the collector of the fan's pass transistor (Q601). This explains why the controller still worked when the fan got its +12 from the unpopulated cap spot. One of those resistors shunts the transistor, keeping the fan running at a "minimum speed." Sometimes, in other PSUs, there's a small electrolytic elsewhere in the (base?) ckt for that transistor, to kickstart the fan at high speed for a second or two when +12 is first applied- overcomes sticky grease or oil.

I've seen two Hipros (one 250W, the other 300W) with fan controllers that work exactly like that. The fan spins at full speed for one second and then quickly ramps down. The 300W has its fan spinning a bit faster, although both have the same model (the common AD0812HS-A70GL). I've seen some 92mm double ball bearing Minebea fans built for Dell work the same way with their NTC thermistor soldered in series with the fan (the fan roars for one second upon powering it up and then quickly slows down... the really high speed fans can sound scary when this happens).

[kaboom]

Re: Power supply build quality pictorial. part 2

Some of the made-for-Dell proprietary ones have an extra wire that controls fan speed; even a semi-custom Newton from a weird Gateway does this. What was originally the -5V lead had been repurposed for the MB control the PS fan. This lead was terminated on the MB by a pullup to +5; an open collector transistor driver pulled this down (or let it float) according to system temperature. Or, the BIOS controlled it, via opamps or other devices. That transistor wasn't driven til the BIOS started, so the ctl lead was driven to +5 via the pullup= full fan speed.

One had to be careful using these with other boards. If a MB loaded that (formerly) -5V pin, the higher pullup resistor in the PSU would no longer supply current to the internal fan driver, leading to a permanently-slow speed fan. I think the fan could be (almost?) stopped by shorting that lead to GND!
Best was to remove the pin from the body and insulate it, or use to manually force a slightly higher speed.

These were unusual. I still have this one, but have the fan-ctl lead tied to +5 thru a resistor (don't remember the value) outside the case. By making it external, it's easer to control the fan w/o taking the board out to change that resistor.

IIRC, they were certain NPS-160 units. They had a white wire in the same position as -5V on the ATX plug, but that voltage was not present on the label and the fan-ctl function was not called out.

[ben7]

Re: Power supply build quality pictorial. part 2

Very nice power supply, Wester547!


Cheers,
-Ben

[Wester547]

Re: Power supply build quality pictorial. part 2

Very nice power supply, Wester547!


Cheers,
-Ben

Lol thank you sir!

IIRC, they were certain NPS-160 units. They had a white wire in the same position as -5V on the ATX plug, but that voltage was not present on the label and the fan-ctl function was not called out.

That does sound a tad daunting, but actually sounds neat. I wish more PSUs gave the option to control the PSU fan via the motherboard, but based on what you said, it's probably not always the best idea for those who don't know what they're doing.

[lti]

Re: Power supply build quality pictorial. part 2

This is the power supply that came with a cheap "weather station" (measures indoor and outdoor temperature and humidity). Searching for the model number brings up a 5V 1A USB charger, so I expected something like the second AC adapter tested here with a more honest label. Instead, it looks like it really was a USB power supply with the output voltage set 0.5V lower and a cheaper switching transistor (Q1) and smaller output filter cap.

On the input, there is a 5D471K MOV, a bridge rectifier, and a CLC pi filter with Jwco 2.2uF 400V caps. Q1 is an MJE13001 (the PCB is designed for a TO-126 package transistor, most likely an MJE13003), and Q2 is an S9014. The AC input wires are 22AWG and UL listed, but the voltage rating was cut off. There appears to be good isolation between the primary and secondary. On the secondary side, there is a 1A rectifier diode, a Huayu 220uF 16V filter cap. Q3 is an SE431 shunt regulator (TL431 equivalent).