Power supply build quality pictorial. part 2

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Well, I'm bored and since here are some PSU geeks:
Could someone tell me the topology used in the following unint?

I still don't understand how that thing works (or how (in)efficient it could be)...

[Agent24]

Re: Power supply build quality pictorial. part 2

It's certainly different to what I've seen before. I don't know the topology just by looking at it but if you find the model numbers of the control IC(s) etc it shouldn't be hard to find out.

I do see a PIC microcontroller in there which is interesting, I wonder what they're using that for?

[stj]

Re: Power supply build quality pictorial. part 2

that probably runs at very high frequency.
those small transformers made from winding wire together and then winding the transformer are a pain to make.
it's done to avoid what's called "skin effect"

[c_hegge]

Re: Power supply build quality pictorial. part 2

I have a unit just like it in my spare PSU stash. It's an NMB. I can't remember much about the parts and design of it off the top of my head, though.

[Pentium4]

Re: Power supply build quality pictorial. part 2

Am I seeing date codes for 2003?! Wow that thing must have cost a fortune! And with 5600uF caps, that output must be clean

[stj]

Re: Power supply build quality pictorial. part 2

i'v seen build quality like that in SGI & SUN psu's

[Wester547]

Re: Power supply build quality pictorial. part 2

Bulk capacitance has more to do with voltage regulation (transients) than ripple suppression (which ESR and ripple rating determine). Well, bulk capacitance does play a factor in suppressing the low frequency ripple on the primary side (in the voltage doubler, and the degree of which passes straight through the output), but less so on the secondary (high frequency ripple). Yes, the skin effect isn't something you want, and higher switching frequencies foster it (causing the current flow to be concentrated toward the outer part of the wire, as stated before me). As for the topology, I see a tiny transformer on the primary side which I've seen in double forward driving the high side MOSFETs. So I think it might be a variation of double forward with Active PFC (unless it's a 230V only PSU) but I could be wrong. Needless to say, it's nice to see all those LXZs on the secondary, that PSU will last a very long time assuming nothing else fails.

[Behemot]

Re: Power supply build quality pictorial. part 2

It seems like having special transformer for each single positive rail.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Sadly it's a WTX PSU so I can't really test it :-/

[c_hegge]

Re: Power supply build quality pictorial. part 2

I might put mine on the load tester when I get a spare moment. I have re-pinned the ATX connector, so it's totally standard now. I also cut out the fan grille on my one and replaced it with a wire grille, as the punched out grille was quite restrictive.

[Behemot]

Re: Power supply build quality pictorial. part 2

Hardcore modders as myself usually cut the grille and put nothing in there…so than you can test your brand-new recapped Enermax Liberty with Samxons, pulling 300W from it, passively cooled (having fan obstructed by ATX 12 V cable). It worked fine for couple hours and from that time works several years for sure with the fan spinning normally. Last like a year year it runs 24/7. So I guess recap went OK

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

That reminds me of a mod someone in the Computerbase.de forum did. I saw it because he was trying to sell that PSU (fortunately no one did buy it).

Looked like this:

(c)EspE...

He removed the plastic foil from the fan, put the cover the other way around back onto the PSU and the fan ontop of that.

Well, I was speechless when I saw that...

I might put mine on the load tester when I get a spare moment. I have re-pinned the ATX connector, so it's totally standard now. I also cut out the fan grille on my one and replaced it with a wire grille, as the punched out grille was quite restrictive.

Thanks a lot.
I was thinking about doing the same, but a more hardcore approach...

I have a spare set of cables from a be quiet! E5-550W, witch failed. It's an FSP epsilon unit with 100% CapXon caps. And guess what.
Sadly I didn't see the thing before I scrapped that unit...
Well, you know those layouts, maybe...

So I could desolder the cables in the NMB unit and put sleeved cables from a modern PSU in it. Question is if it makes sense...

[Behemot]

Re: Power supply build quality pictorial. part 2

Well the Fusion 550 (basicaly OEM Seasonic 500 W) has fan from outside by design

[Pentium4]

Re: Power supply build quality pictorial. part 2

Found an Award PSU that looked suspiciously like a Hirpo, and it was! I did a fair amount of "mods" to it. Recapped it, upgraded the 470uF input caps with 560uF, replaced the 20A schottky on the 12V with identical 20A schottky's in parallel, soldered on two molex, a 6 pin PCIe, and a 4 pin to make the motherboard connector 24 pin, redid some bad joints. Single forward with a 10A FET on the primary. 5V uses a 40A schottky, and it uses a DC-DC conversion to make the 3.3V. The death-ADDA was dry so I oiled it. It runs very silent. I love how big the heatsinks are and the 12.5mm slots for the caps. I was really impressed with the voltage regulation on the 12V. I loaded it up to 175W almost purely 12V. While cold the 12V read 12.18V, after a half hour of running it read 12.17V, and the fan barely picked up pace.

[Wester547]

Re: Power supply build quality pictorial. part 2

Some joints were bad? Did they actually cause the PSU to be unstable or did they just "look" bad (which would be characteristic of Hipro as back then they did average soldering)? +5V uses a 40A schottky? So does that mean the other one is 30A (+3.3V, I know the FET itself that regulates +5V down to +3.3V is a 40A ST part)? Can you read the label on the main switcher behind that daughterboard on the primary? Sorry for all the questions. I really like how large those ferrite coils are, much larger than the ones in the 200/250W Hipros from Dell (but those units are still tanks after a recap and if necessary either a fan swap or lubing). That toroid is kind of small for 300W, though (the toroids in those 250W Hipros are much larger). Also, the fan barely picked up pace because you added another 20A schottky on the +12V rail. Note how the thermistor is mounted to the secondary heatsink so the fan revs up as the temperature of the secondary heatsink augments. That unit uses linear regulation to get +3.3V so the +3.3V voltage regulation and ripple suppression should be very very good. Also, is the fan AD0812HS-A70GL or AD0812MS-A70GL?

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

Very good job and photos!

What is the marking on the fet?

[Pentium4]

Re: Power supply build quality pictorial. part 2

Some joints were bad? Did they actually cause the PSU to be unstable or did they just "look" bad (which would be characteristic of Hipro as back then they did average soldering)? +5V uses a 40A schottky? So does that mean the other one is 30A (+3.3V, I know the FET itself that regulates +5V down to +3.3V is a 40A ST part)? Can you read the label on the main switcher behind that daughterboard on the primary? Sorry for all the questions. I really like how large those ferrite coils are, much larger than the ones in the 200/250W Hipros from Dell (but those units are still tanks after a recap and if necessary either a fan swap or lubing). That toroid is kind of small for 300W, though (the toroids in those 250W Hipros are much larger). Also, the fan barely picked up pace because you added another 20A schottky on the +12V rail. Note how the thermistor is mounted to the secondary heatsink so the fan revs up as the temperature of the secondary heatsink augments. That unit uses linear regulation to get +3.3V so the +3.3V voltage regulation and ripple suppression should be very very good. Also, is the fan AD0812HS-A70GL or AD0812MS-A70GL?

They looked bad, as in extremely "blobby" which is unlike Hipro. And surprisingly, there was a slightly lifted pad next to the grounding leg of the primary heatsink. Yeah the 3.3V one was 30A. The toroid does look small, I bet it could do 300W but gets hot. Yes, I'm hoping that the extra rectifier helped the efficiency out, it's just using two MBR20100CT's. Plus, that heatsink is so tall and thick!

Yeah I think the ripple will be in good shape, the caps I used should be quite a bit better than the Teapo SC, SS, and SEK used. I like how Hipro has such a nice feedback loop design so they aren't picky with caps.

Shoot, I don't remember the fan model number but I'll try and remember to check tomorrow!

Very good job and photos!

What is the marking on the fet?

Thank you! It's a Fairchild FQA9N90C, so I was wrong, it's rated for 9A.

Edit: Here's a pic of that little daughter-board:

[Wester547]

Re: Power supply build quality pictorial. part 2

So the fan spins rather slowly? Did you stick a flashlight onto it and were the blades visible or did you just not feel much airflow from it, or is it just because it's extremely quiet?

[Pentium4]

Re: Power supply build quality pictorial. part 2

It spins slow but moves quite a bit of air. I'd guess it spins around 800RPM at low load and 1000RPM under that load. The fan is a AD0812HS-A70GL

[Wester547]

Re: Power supply build quality pictorial. part 2

That would mean that the blades are spinning very slowly - 800RPM - 1000RPM is too slow for an 80mm fan to be moving much air at all. I'd guess it's closer to 1200RPM to 1400RPM (the blades still may look somewhat "slow" at that point but it'd be hard to see them without a flashlight - at 800RPM - 1000RPM you wouldn't even need a flashlight to see the blades spinning, assuming the room has good light). I find that ADDA fans actually don't move as much air as Delta or Sunon fans do at the same speed, though (but those brands are quite a bit noisier).