Power supply build quality pictorial. part 2

[Pentium4]

Re: Power supply build quality pictorial. part 2

Yeah you're probably right. I'm impressed with how silent it is. Sleeve Bearing ADDA fans are fine after being oiled, but I would never trust one without being oiled.

[Wester547]

Re: Power supply build quality pictorial. part 2

Don't you have another version of that power supply that came out of a Gateway desktop? Is it pretty much the same as what the one you showed here too or is it different at all?

[c_hegge]

Re: Power supply build quality pictorial. part 2

I stuck my NMB 430W PSU on the tester (the one just mentioned here - https://www.badcaps.net/forum/showpo...postcount=1821). I didn't go through the full set of tests - I only had the time to do a best case scenario of 250W load.

The ripple suppression was excellent - about 25mV on the 12V and 10mV on the other two rails - but the efficiency was woeful - about 74% - on 230V. It was loud too - even with the wire fan grille modification. It seems to have a very aggressive fan curve, although that's kind of to be expected on server PSUs.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

I stuck my NMB 430W PSU on the tester

but the efficiency was woeful - about 74% - on 230V.

That's way better than I'd expect from a PSU from the beginning of the century/millenium...

Still not that much better than other PSUs that came a little bit later...

[c_hegge]

Re: Power supply build quality pictorial. part 2

^
I would argue that the efficiency is probably the least important aspect of a PSU's performance. It's the performance on the output side that really matters, and it does very well there (as I would expect form such an overbuilt unit).

[Behemot]

Re: Power supply build quality pictorial. part 2

IMO the efficiency is very important if it is able to maintain reasonable ripple suppression (around 50 mV/+12 V and up to 30 mV/other positive rails) and voltage control (+-3 %) what I take as a must. Than OFC it depends on what segment the PSU is, mainstream or high-end, lately quite often these wannabe high-end PSUs offer mainstream performance in this matter. Somebody has to pay the electricity bills after all

[c_hegge]

Re: Power supply build quality pictorial. part 2

IMO the efficiency is very important if it is able to maintain reasonable ripple suppression (around 50 mV/+12 V and up to 30 mV/other positive rails) and voltage control (+-3 %) what I take as a must.

Huh? The efficiency has nothing to do with the output performance. This NMB unit had pretty bad efficiency, but the output performance was excellent. As far as the power bill is concerned, it doesn't add that much. maybe just a few dollars per month - peanuts compared to devices like a refrigerator and oven.

[Behemot]

Re: Power supply build quality pictorial. part 2

Do I say it has anything in common? I said as long as it has OK output (which every at least mainstream PSU should have), than efficiency is important.

[stj]

Re: Power supply build quality pictorial. part 2

more eficiency = less heat = longer life.

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

more efficiency + bigger heatsinks = slower fan = longer fan life

[Wester547]

Re: Power supply build quality pictorial. part 2

And more efficiency + bigger heatsinks + slower fan + cleaner output + well designed power supply + well ventilated computer = long computer life.

Here's another Dell NPS-250KB (Newton Power) A 250W max output, version 02, manufactured on the 19th week of 2002. +5V is rated for 22A, +3.3V is rated for 18A, +12V is rated for 14A, -12V is rated for 1A, +5VSB is rated for 2A, and +5V and +3.3V combined is rated for 170W. This has very high quality heatsinks and very good ventilation with a nice and expansive honeycomb grille at the back. Unlike many of the Newtons within those years, this one, like mockingbird's, has all Japanese capacitors on the output and all the stock capacitors in the unit are pretty much Japanese excepting the tiny capacitors which are Taicon VT (and one 5mm 100uF 16V Taicon VR in the decoupling logic of the secondary). Primaries are 30x25 560uF 200V Nichicon LU. Bridge rectifier is GBU 8K by Panjit Semiconductor without a heatsink. +5VSB transistor is unfortunately a free standing IRFBC20 which makes it a two transistor +5VSB circuit with a 5mm 100uF 25V Taicon VT being the "critical" capacitor (I've never ever seen a failed Taicon VT, though, for what it's worth, so I think the Taicons in the unit can stay, and I will explain why later on in the post).

The main switcher is an IRFPE40 (International Rectifier, single switch forward topology) - that 800V switcher has a maximum ON RESISTANCE rating of 2 ohms, so I can see why this unit is rated for 250W. The unit has four Infineon optocouplers. The main transformer is a size 35 (I did notice a transformer whine when the unit was running, though...). The +5VSB output diode is a SB340 (Panjit again) and the +5VSB output capacitors compose of a 2200uF 10V 10x28 Rubycon YXG before the ferrite coil and an 8x15 470uF 10V Nichicon PW after the ferrite coil. +3.3V is rectified by two STPS2045CTs. +3.3V uses the regular magamp circuit (both toroids use -52 material which is good) and is filtered by two 10x20 2200uF 6.3V Rubycon YXGs, one before the ferrite coil and one after. -12V is filtered by a single 8x20 470uF 25V Rubycon YXG, with L7912CV regulating it (linear reguation) and a free standing UG10DCT (General Semiconductor) rectifying it. +5V is rectified by a single MBR3045CT (General Semi again) and a single STPS3045CW connected to it (on the low side). +5V is filtered by another 10x20 2200uF 6.3V Rubycon YXG before and after the ferrite coil. +12V is rectified by a MBR20100CT (General Semi yet again, but this MBR20100CT has a maximum voltage drop of 0.75V @ 20A/125*C, going by the datasheet, which is nice because most MBR20100CTs have a drop of closer to 0.85V at those specs). +12V is filtered by a 10x31.5 1500uF 16V Nichicon PW and an 8x16 470uF 16V Rubycon YXG in parallel (the +12V rail also has a silver colored shunt for OCP). The unit has an AD0812HS-A70GL fan which doesn't spin very fast or move that much air but still spins like a new fan. There is a fan header for the ADDA fan so it can be replaced without getting the soldering iron, but, the fan header is quite a bit smaller than most, and the clip itself on the fan's wires also seems to be of a proprietary kind - a replacement, without ghetto modding skills, would more than likely have to come from another Dell Newton 250W. The primary side is supervised by DNA1001DL and the secondary side DNA1002E (on the second daughterboard, the first being on the primary along with the input filtering sister board - the input filtering is heavy and that sister board has two NTC thermistors) and LM324N (STMicro).

Here's why the Taicons stay: this unit runs incredibly cool, even with that Passive PFC blocking airflow to the secondary (which doesn't make the unit as heavy as you would think - the Passive PFCs in Hipros are much heavier). And this is in a room with an ambient temperature of 32*C: here in California, it's been very hot of late. I've checked the temperature of every capacitor in the unit after hours of 120W loads and after hours of standby and nothing in the unit gets remotely close to hot yet alone warm. Not the toroids, not the free standing transistors, not the transformers, not the free standing diodes, not the passive PFC, not the capacitors, not the heatsinks... no wonder LTEC capacitors last forever in these units. Of course it helps that none of the output capacitors are blocked by the secondary wires and they are nicely organized (not bunched up against each other), especially in units without a passive PFC. As you can tell from the shots, I'm not a photographer, but you can see how Newton's soldering is an impressive feat, and how the build quality is generally very good. Here is something I will say about Newton/Delta, though, in comparison to Hipro - the Hipros of this time period had much larger toroids for the +5V/+12V rails and had much heavier output filtering, so I would imagine, in tandem with a linear regulated +3.3V rail, that the ripple suppression and voltage regulation was far superior. As nice as this unit is, if I had to choose between this unit and a 250W Dell Hipro (recapped and with a fan swap or lubing if necessary), I'd go with the Hipro (despite Hipro's inferior soldering)...

Not that this is a bad unit by any means, still very overbuilt for what the label says. Sorry for the long post.

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

I would replace the 8x16 470uF 16V Rubycon YXG in 12V output filter with 2200uF 16V Panasonic FK / Nichicon PW / HE / Chemicon KY, whatever would fit there.

[momaka]

Re: Power supply build quality pictorial. part 2

Sorry for the long post.

I don't think you need to apologize. IMO, it's actually fun reading through a detailed post that breaks down the PSU down to the basic components.

+5VSB transistor is unfortunately a free standing IRFBC20 which makes it a two transistor +5VSB circuit with a 5mm 100uF 25V Taicon VT being the "critical" capacitor.

I saw a unit like that several months ago at Micro Center for $5. Really wish I would have gotten it now. The one that I was looking at was rated for 30A on the 5V rail, IIRC (but only 8 or 10A or so on the 12V). I still have an old Athlon XP system that could use a quality 5V-heavy design. The only thing that stopped me from buying it was the 2-transistor 5VSB circuit. Otherwise, it was a very similar unit like yours - all Japanese caps, huge heatsinks, nice honeycomb stamps. No PPFC, but that's okay. They no longer sell these old refurb units, though.

I wonder why yours has four optocouplers. On single transistor forward, you usually need only three. Perhaps there are more protections. Or I wonder if this adds short circuit or OPP protections to the 2-transistor 5VSB circuit. The HiPro HP-P2507F3P have four as well, but IIRC, two were for the 5VSB circuit (which is generated with a TOPSWITCH IC).

I guess the 5VSB must be very efficient if nothing gets warm. The older Deers I have make the primary heatsink quite warm. Now that I have a Kill-A-Watt meter, I guess it would be interesting to see what the no-load and full load efficiencies are of various PSUs I own.

[Wester547]

Re: Power supply build quality pictorial. part 2

I would replace the 8x16 470uF 16V Rubycon YXG in 12V output filter with 2200uF 16V Panasonic FK / Nichicon PW / HE / Chemicon KY, whatever would fit there.

I'm not aware of anything besides Samxon capacitors being available in 10mm as far as 2200uF 16V capacitors go. The PCB is unfortunately only silkscreened for 10mm capacitors on the secondary. Since they are in parallel, it's more voltage regulation that would worry me for the lack of bulk capacitance, but the unit seems to do fine and is very stable in all regards.

I don't think the fourth optocoupler helps the two-transistor +5VSB circuit's lack of OVP. I believe it is there as a crowbar protection circuit of sorts, but, something has to tell the PSU to shut down so the PSU can "latch" - with no OVP for +5VSB that won't happen. I always thought the Hipro's fourth optocoupler was there as a crowbar protection circuit of sorts as well since I noticed an ST Triac without a heatsink in those Hipros, in the secondary. But, I never took the time to observe where the traces were going between those areas on the underside of the PCB. Those Hipros are probably better off without PPFC since the huge PPFC blocks so much airflow to the primary heatsink in those units. PPFC also increases the stress on the main switcher by reducing the voltage and increasing the current. Yes, those Hipros have +5VSB generated by either a Tiny Switch (TNY27x) or a TOPSWITCH (TOP200YA) which has a 10uF 50V capacitor driving it. Those TOP200YAs have a heatsink so I imagine they get warm.

[c_hegge]

Re: Power supply build quality pictorial. part 2

I'm not aware of anything besides Samxon capacitors being available in 10mm as far as 2200uF 16V capacitors go

I use Panny FK 2200uF 16V 10mm caps all the time in repairs.

[Wester547]

Re: Power supply build quality pictorial. part 2

I use Panny FK 2200uF 16V 10mm caps all the time in repairs.

Whoops, my mistake. I see now that the datasheet lists 2200uF 16V Panasonic FK in 10x30. I guess that would be the only choice.

[stj]

Re: Power supply build quality pictorial. part 2

i try to only use FK & FR.
infact i try to only use panasonic in general - they are cheaper than rubycon

[Behemot]

Re: Power supply build quality pictorial. part 2

I have 3300uF/16V D10 Samxon RS caps. Chemi-Con KZN D10x40mm are being discussed ATM.

[Pentium4]

Re: Power supply build quality pictorial. part 2

And more efficiency + bigger heatsinks + slower fan + cleaner output + well designed power supply + well ventilated computer = long computer life.

Here's another Dell NPS-250KB (Newton Power)

WOW! That is a nice PSU. Very overbuilt indeed. I don't see that thing ever failing! No wonder it doesn't get hot, look at those heatsinks!

[RJARRRPCGP]

Re: Power supply build quality pictorial. part 2

considering the motherboard in there doesn't use the 12V 4 pin plug.

That's a rare breed! Intels were using the +12 V connector before AMD did, even with socket 423.

OTOH, Asus socket 462 (AMD) motherboards always used the +5 V rail.

If it's using a lot of watts, it may cook one of the 20-pin ATX power connector pins.