Power supply build quality pictorial. part 2

[eccerr0r]

Re: Power supply build quality pictorial. part 2

Interesting. I was trying to trace some of the circuitry to verify the fact and thought it was still a flyback because there still only seems to be two nonisolated side windings and only one switching transistor (looking at a bunch of existing designs it looked like a forward design either needs two transistors (double ended, one winding plus the 3845 power winding) or two (single ended, two windings plus one to power the 3845) windings and a diode...)

I did not think of the output inductor, or compared the diode configuration, thought that those would be somewhat helpful on any design as it would help filtering no matter if it was forward or flyback.

[momaka]

Re: Power supply build quality pictorial. part 2

Powerman microatx HTPC custom PSU. 3845 switch controller (on bottom, not pictured)

Bad caps, bad caps...

Do you have a shot of the label for that PSU? A co-worker of mine bought several InWin small ITX cases that came with Powerman PSUs. I just wonder how similar his are to this. Powerman typically makes good PSUs with good designs, but bad caps.

The power estimates when I first got this PSU/case I remember calculating it to be just be on the edge of the power rating of the supply... I seem to recall it was rated at 235W but I don't remember exactly.

A Core 2 Duo PC with on-board video card will hardly go over 100 Watts (unless, of course, you have a lot of HDDs in there).

The strange thing is that it seems to have thermal sensing, I'm not sure if it's for shutdown or for the fan controller... might be for the latter only...

Probably fan controller only.

BTW - what tipped you off for this to be a forward converter... It sort of was confusing because all the UC384x supplies I've seen were invariably flyback.

You generally won't see flyback topology used for more than 100-150 Watts. The size of the output filtering caps required for flyback grows almost exponentially with heavier loads. You can get much higher current outputs with a lot less filtering with a continuous switching design, such as forward converter.

As PeteS in CA noted, the output toroidal inductor is a sign that you are not dealing with flyback (or any other discontinuous switching design, for that matter).

Interesting. I was trying to trace some of the circuitry to verify the fact and thought it was still a flyback because there still only seems to be two nonisolated side windings and only one switching transistor (looking at a bunch of existing designs it looked like a forward design either needs two transistors (double ended, one winding plus the 3845 power winding) or two (single ended, two windings plus one to power the 3845) windings and a diode...)

Well, the two most widely-used forward converter topologies are the single-transistor forward (i.e. what you have) and the double-transistor forward (what many high-power modern PSUs use, as a double-forward configuration greatly reduces the need for a large primary-side snubber).

The difference between single-transistor forward and flyback is pretty small, other than the gapped transformer and output inductor, as Pete explained.

[eccerr0r]

Re: Power supply build quality pictorial. part 2

Well, that's what's really confusing me. There does not appear to be enough windings or transistors to make it a forward design. It looks like there are only 2 unisolated side (4 pins) and one switching transistor which makes it look like a flyback design to my untrained eyes. One of those two unisolated windings is to power the 3845, not the auxillary winding that's typical in a forward converter. Okay citing wikipedia is a bad idea, but that "third winding" I cannot find (nor that D3 diode). The "second" winding I do see is that typical UC384x winding that's rectified and filtered, with that small cap that always dries out and kills UC384x PSUs... Speaking of which I should have replaced it while I had it out and swapped the output caps, oops. Grr. I guess I'll have to open it up again sometime.

I'll need to get a picture of the reverse/track side I suppose... And yeah I think this indeed is an Inwin low profile microatx HTPC case IIRC. I had another Inwin case with Powerman PSU and was pleased with the longevity of the PSU, but this one, caps have killed the joy...

[Behemot]

Re: Power supply build quality pictorial. part 2

I was to have a quick look at this chinese wonder, R-Senda SD-1600EPS. As usually, there is no word about this model anywhere on their site or even on the Internet so who knows what the heck this is. Also no 80 PLUS certification papers available from Ecova so it is likely bogus, though it does provide slightly over 80 % peak efficiency (close to bronze actually). Close to maximum power it was barely 80 %. The maximum output is about 1300 W than it shuts down. I wonder what is actually the rated power, 1800 W, 1600 W, 1600 W peak or what?

Notice the complete lack of input filtering, there is not even place for it, with the exception of the three Y ceramics (only two of them actually populated). The ripple on two rails I was looking at (+5 V, +12 V) was high over 200 mV Uses ancient CM6802.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

Well, it does say 1600W is the peak load. That fan better HOWL to get much more than 800W-1000W, probably. The -52 material cores of the smaller output inductors suggests a switch frequency higher than 50KHz. Their presence and the ~80% efficiency also suggest that it's regular PWM rather than quasi-resonant phase-modulated PWM. With the unfiltered AC running under the output and PFC/inverter heatsinks, I'm really skeptical it meets conducted noise. It looks like there is a fairly large inrush thermistor, with a relay that will short it not too long after power-on (that relay better be safety agency approved!). That saves power, but unless the relay is well chosen and the relay drive is really reliable, the contacts could weld closed, making for no inrush protection, or the relay could fail to close, resulting in a blown thermistor.

Depending on the cap brands and the quality of that fan, I suppose it could be a decent PSU with a realistic power capability in the 700W-1000W range.

[momaka]

Re: Power supply build quality pictorial. part 2

Depending on the cap brands and the quality of that fan, I suppose it could be a decent PSU with a realistic power capability in the 700W-1000W range.

I agree.

Perhaps try running a bunch of load tests with different loads on each rail to see how much current each rail can draw before the ripple gets out of spec.Then come up with a load test that combines the max power draw of each rail and see if the ripple is still in spec. If yes, then look at the maximum power draw and de-rate the PSU to 80% of that. So for example, if you can pull 1000 Watts out of it before the ripple on any of the rails exceeds the ripple specs, then call it a good 800 Watt PSU.

Also, while the PSU may not meet its labeled specs... if you got it for cheap and it performs on par or better than a name-brand PSU in that same price range, then you didn't get such a bad deal.

The only deal breaker really is the lack on input filtering. I see a bit of space between the primary and secondary heatsinks. Perhaps with some creativity, you may be able to add a line filter in there.

[momaka]

Jeantech JNP-500AP (PCB No:380 REV:B5 2005-07-12)

It's been a while since I posted anything in this thread.

Here is a Jeantech JNP-500AP power supply (PCB No:380 REV:B5 2005-07-12) that member Pentium 4 gave me a while back (thanks man!!)

As usual, we start with the outside shots:
https://www.badcaps.net/forum/attach...1&d=1498952434
https://www.badcaps.net/forum/attach...1&d=1498952434
It's nothing special on the outside, though the case is made of decent sturdy metal. Output wires are all 18 AWG. And as you can see, this PSU has a cubic butt-ton of Molex connectors for whatever reason (at least they are good quality Molex, and not the cheap PSU kind). For connectors, this is what we get:
20+4 pin main connector, 4-pin 12V CPU connector, 9 Molex, 2 floppy, 2 SATA, and a fan monitor. NO PCI-E 16x, unfortunately! :\

Moving on to the label:
https://www.badcaps.net/forum/attach...1&d=1498952434
That label looks familiar. I cannot remember on which PSUs I've seen it exactly, but it looks like Sirtec. Indeed if you search for the UL number online, E144195, you will find that it is a Sirtech power supply. And from what I gather, this is one of their “High Power” series. Funny, though – the label says “PEAK 500W”. So what is the continuous rating? No one knows, I suppose. But at least we get Active PFC and a ball bearing fan. Also the promise of two 12V rails. But let's remove the cover and find out what this PSU really hides inside.

https://www.badcaps.net/forum/attach...1&d=1498952434
https://www.badcaps.net/forum/attach...1&d=1498952434
Initial impressions: not bad at all. Lots of components and well-packed in there. However, you can tell this is an old design from 2005 – not just because the PCB says so, but also because of that add-on APFC board mounted on the primary heatsink. Clearly, this is from a time before APFC was standard in PSUs.

So let's have a look at that APFC board too:
https://www.badcaps.net/forum/attach...1&d=1498952434
Yup, this is definitely an old design “retrofitted” with an APFC board, as one can tell by the input capacitors, which are two 250V, 680 uF Teapo LXK caps (i.e. equivalent to a single 500V, 340 uF primary cap). I didn't measure how high the APFC boosts the voltage, but obviously it has quite a bit of headroom on the voltage. (If the APFC is tweaked to boost to 450V, those caps will hold approximately the same energy as a single 470 uF cap charged to 380V.)

Also, I forgot to mention – the APFC controller is a UCC3818N, and the 7812 linear regulator is most likely for its supply voltage.

Next, a picture of the secondary side:
https://www.badcaps.net/forum/attach...1&d=1498952434
Ugh, that tan crap glue is everywhere. At least it's mostly on the secondary side, though, and not much on the primary. When I see it on the primary side, that's when I get worried about it causing blow-outs. Other than the glue, it doesn't look bad. Lots of components and the output toroids are quite big with thick wire. Output caps are mostly Teapo SC, but they are big too, so hopefully won't be under too much stress. PWM controller is a SG6105d and that daughterboard contains a LM339 quad-comparator. Not sure what the LM339 does – protections, perhaps? It may also have one of its comparators used for fan control, so who knows. Speaking of fans they are two 80 mm Superreds: CHA8012CBS-A(E) (12 V, 0.17 A) and CHA8012BS-M(E) (12 V, 0.12 A):
https://www.badcaps.net/forum/attach...1&d=1498952434

And finally, the solder side:
https://www.badcaps.net/forum/attach...1&d=1498952434
Looks alright. Separation between traces on primary is okay too.

So here is the detailed break-down of the PSU…

ICs
APFC controller: UCC3818N
APFC Vcc regulator: L7812C
PWM controller: SG6105d
Additional Protections (not sure which exactly): LM393
-5V rail regulator: L7905

Primary Side
*Receptacle with built-in filtering
* five 3.3 nF safety approved Y2 caps (including 2x in receptacle filter)
* four X2-class safety approved caps (0.1 uF in IEC receptacle and 0.22 uF on board + 2 others I couldn't read their value.
* three CM chokes (one in IEC receptacle is 0.3 mH)
* 6 or 8 Amp –rated bridge rectifier (can't read part info)
* one MOV between L&N, and one NTC thermistor (can't read info)
* input wiring rated for 600 V, 18 AWG
* APFC transistors: two 11N60
* APFC diode: BYC10600 diode
* primary input filter caps: 2x Teapo LXK, 250 V, 680 uF, 25 x 40 mm (dia. x h.)
* H-bridge transistors: two 2SC332_ NPN BJTs
* 2N60B MOSFET for 5VSB circuit (2-transistor design)
* size “39” or “40” main transformer, and size “19” for 5VSB and BJT drive transformers

Secondary Side

3.3V rail:
* mag-amp regulated
* rectifier: SBL4045PT
* filter caps: 2x Teapo SC, 6.3 V, 3300 uF, 10 x 25 mm
* PI coil: 5-turn, 6 mm core, between the two caps
* Min. Load Resistor: 7.5 Ohm, 3-Watt

5V rail:
* rectifier: two D____ in parallel (can't read their numbers, but they are in TO-3P cases)
* filter caps: 2x Teapo SC, 10 V, 3300 uF, 10 x 30 mm
* PI coil: 5-turn, 6 mm core, between the two caps
* Min. Load Resistor: 100 Ohm 1-Watt and two diodes + 1 Ohm resistor all in series to 3.3V rail

12V rail:
* rectifier: MBR30150P
* filter caps:
- 1x Teapo SEK, 16 V (?), 100 uF (?) , 5 x 11 mm
- 1x Teapo SC, 16 V, 2200 uF, 10 x 30 mm
* PI coil: 20-turn, 6 mm core, between the two caps
* Min. Load Resistor: 600 Ohm (?) 1 or 2-Watt

-12V rail:
* rectifier: two PR3002 diodes
* filter caps: Teapo SEK, 16 V, 470 uF, 8 x 13 mm
* PI coil: none
* Min. Load Resistor: none (output to -5V rail 7905 regulator)

-5V rail:
* rectifier: none – derived from a L7905 linear regulator from -12V rail
* filter caps: Teapo SEK, 50 V, 10 uF, 5 x 11 mm on L7905 output
* PI coil: none
* Min. Load Resistor: 910 Ohms

5VSB rail:
* rectifier: SB340 (?) Schottky diode
* filter caps:
- 1x Teapo SC, 16 V, 1000 uF, 10 x 20 mm
- 1x Teapo SC, 16 V, 470 uF, 8 x 15 mm
* PI coil: 5-10-turn, 5 mm core, between the two caps
* Min. Load Resistor: 47 Ohm, 2-Watt

Lastly, I want to mention that this PSU draws A LOT of power when OFF but plugged in – somewhere around 8-10 Watts, as measured by a P3 Kill-A-Watt meter. I thought the APFC might be fooling my meter. But alas, the PSU gets very warm when sitting OFF, so I would not be surprised if it actually is drawing that much power. Most of it seems to originate from the APFC section, as that side of the PSU is the hottest. I imagine the 2-transistor 5VSB is most likely responsible for drawing 3-4 Watts, and the rest is the APFC circuit. Worst part is, this PSU has Teapo caps everywhere, and Teapo does not like heat at all. So I have no clue how long it will last like that.

I have the PSU running in a PC that I use maybe for a few weeks or so every year when visiting some relatives. When not there, I keep the PC unplugged and stashed away, so perhaps that might help it live longer.

The other thing I have noticed is that Sirtech PSUs have somewhat poor voltage regulation, especially on boot-up. When powering on the PC, I hear an awful lot of squealing for 1-2 seconds before the PSU settles. The output voltages seem to wobble a bit during that time as well. And worst part is, I know it's not the caps, as I have another PSU built by Sirtech that I recapped, and it acts in a similar way (and even worse, when I filled in an “empty” cap spot).

So what are your thoughts on this Sirtec High Power PSU?

[Behemot]

Re: Power supply build quality pictorial. part 2

I doubt Sirtech has anything in common with Sirtec. It's some cheap chinese crap maker trying to fool somebody with similar name. The unit itself looks like some cheap copy of Chieftec either, the label is absolutely the same as Chieftec used for such old units.

[Wester547]

Re: Power supply build quality pictorial. part 2

I doubt Sirtech has anything in common with Sirtec. It's some cheap chinese crap maker trying to fool somebody with similar name. The unit itself looks like some cheap copy of Chieftec either, the label is absolutely the same as Chieftec used for such old units.

I see a couple small G-Luxon SM capacitors on the primary side too. The usage of Teapo and G-Luxon capacitors is characteristic of Sirtec / Sirfa / High Power PSUs, as is the usage of a single tiny, ventless capacitor and another large capacitor for +12V output filtering.

Output caps are mostly Teapo SC, but they are big too, so hopefully won’t be under too much stress.

I don't see a single 12.5mm diameter Teapo SC on the output. Could be larger.

Worst part is, this PSU has Teapo caps everywhere, and Teapo does not like heat at all. So I have no clue how long it will last like that.

I think all Teapo SC (and of course lower impedance series) made after 2005 are replace-on-sight. Teapo was never a good brand but in my experience those made before 2006 are more likely to dry out silently under moderate thermal duress. Those made after seem to be notorious for bulging even whilst sitting in storage.

That PSU doesn't look too shabby sans the conductive glue and "suspicious" capacitors. I guess the PSU does not have any true "OVP" for +5VSB.

Here's an even older "retro trip"... a 96W Foxlink FTPS-0007 Rev: D power supply, a 120V only PSU for the original Xbox, revisions 1.2 - 1.5. +5V is rated for 13.2A, +3.3V 4.8A, +12V 1.2A, and +3VSB 0.045A. The power supply was manufactured in September 2003 (or the 39th week of 2003).

On the primary side, for the AC input line filtering, we have two coils, two Y capacitors, one NTC thermistor, two MOVs (one for the primary capacitor, one further up the input filter), and two X capacitors (one of which is a 0.047uF 250V Teapo XG-VS). The bridge rectifier is a GBU 4J by Panjit Semiconductor. The primary capacitor is a 680uF 200V Teapo LXK (25mmx30mm), and the primary side PWM controller appears to be a ST3845B chip on the primary side daughterboard. The topology is single switch forward. The main switcher is STP10NC50FP. Not sure what size the primary transformer is, but it looks to be equal to a size 28 transformer. The auxiliary transformer looks close to a size 16-19 transformer. The main standby switcher is a freestanding 2SK2865, and the startup capacitor is a 100uF 25V 6.3mmx11mm Teapo SEK. This PSU only has two optocouplers because +5VSB is linear regulated down to +3.3VSB inside the PSU itself rather than on the motherboard.

Before the freestanding TL431 regulator, we have a 6.3mmx11mm 100uF 25V Teapo SC and a 6.3mmx11mm Teapo SEK after (can't read the capacitance or voltage rating as that part of the sleeve is covered in glue). +5V rectification is done with a single MBR3045CT by General Semiconductor and +3.3V rectification with a single SB1040 by Panjit Semiconductor. +12V rectification is done by a single UG10DCT by General Semiconductor again. There appears to be another 5mmx11mm Teapo SEK on the decoupling side of the secondary logic, but I can't read its voltage or capacitance rating either as both are covered in glue. The secondary side controller is a TPS3510 on the secondary side daughterboard. +5V filtering is done with two 10mmx25mm 2200uF 10V Teapo SC in parallel. +3.3V filtering is done with one 10mmx25mm 3300uF 6.3V Teapo SC, and +12V filtering with one 10mmx20mm 1000uF 16V Teapo SC. All output wires are 18 gauge. The PSU has one 20-pin ATX connector and another molex connector.

A bit disconcerting is the fact that the +12V rail appears to have an 180Ω minimum load resistor with a 5% -/+ tolerance (I don't think I'm reading the color codes wrong). That's obviously too low for the +12V rail but there are no signs of discoloration on the underside or overside, and the ST310014ACE from the Xbox this PSU was powering only has 2,900 ON hours. So maybe not quite enough time to discolor the PCB from the underspec'd load resistor.

The PSU employs the conventional magamp circuit for +3.3V and +5V on the output, with an additional coil used to derive the +3.3V transformer pins from the +5V transformer pins. You may notice that both the primary and secondary heatsink are quite small and finless. This PSU is only rated for 96W and the original Xbox doesn't draw anything quite in excess of that, but even then the secondary heatsink reaches a "warm" surface temperature of 66°C (at 25°C room temperature) once the temperature has stabilized and the Xbox has been on for long enough. The equivalent Delta PSU interchangeably used in these older consoles (the DPSN-96BP) has a secondary heatsink with several fins and uses more overspec'd output rectifiers, so that heatsink barely reaches 41°C under the same conditions.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

The secondary side controller is a TPS3510 on the secondary side daughterboard.

I believe that's a Supervisor IC, checking for UV/OV of the several outputs, providing a PG signal to the "MB", and a Fault signal to the P/S control circuit. The full-grown 14 pin version (e.g. the TPS3514) also checks for over-current on each output.

[momaka]

Re: Power supply build quality pictorial. part 2

I see a couple small G-Luxon SM capacitors on the primary side too. The usage of Teapo and G-Luxon capacitors is characteristic of Sirtec / Sirfa / High Power PSUs, as is the usage of a single tiny, ventless capacitor and another large capacitor for +12V output filtering.

^ This.

I've seen that same exact design in other Sirtec PSUs here over the years. Worse, some of them have that tiny cap shoved in a spot for a 8 mm or 10 mm cap. If you replaced it with an actual 8 mm or 10 mm cap with much larger capacity (470 uF and up), the PSU starts to oscillate and whine very badly. Been there, done that. I don't think I have posted pictures of that PSU yet.

Also, some Sirtec PSUs also use Jenpo caps on the 3.3V rail - a brand I have only seen in Sirtec.

I don't see a single 12.5mm diameter Teapo SC on the output. Could be larger.

Yeah, you're right. Those caps could be larger indeed. But for what it's worth, those ones in there are still decently large in terms of capacity (the 3300 uF caps for the 3.3V and 5V rails).

That PSU doesn't look too shabby sans the conductive glue and "suspicious" capacitors. I guess the PSU does not have any true "OVP" for +5VSB.

I doubt it - looks like a standard 2-transistor.

Here's an even older "retro trip"... a 96W Foxlink FTPS-0007 Rev: D power supply, a 120V only PSU for the original Xbox, revisions 1.2 - 1.5.
...
A bit disconcerting is the fact that the +12V rail appears to have an 180Ω minimum load resistor with a 5% -/+ tolerance (I don't think I'm reading the color codes wrong). That's obviously too low for the +12V rail...

Yes, the dissipation for that resistor will be a bit on the high-side at around 0.8 Watts. Probably runs around 60-70°C - hot enough to burn your finger, but not hot enough to darken wires and PCB - unlike cheap Deer / L&C PSUs that have their 12V load resistors dissipate around 1.44 Watts.

You may notice that both the primary and secondary heatsink are quite small and finless. This PSU is only rated for 96W and the original Xbox doesn't draw anything quite in excess of that, but even then the secondary heatsink reaches a "warm" surface temperature of 66°C (at 25°C room temperature) once the temperature has stabilized and the Xbox has been on for long enough.

66°C is not that bad for a heatsink inside a PSU, but obviously could be better, of course, like you noted about the equivalent Delta PSU . Now heatsinks running in excess of 80°C - that's pretty bad. And some cheap PSUs under load will go even much higher than that.

I believe that's a Supervisor IC

I believe so, too.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Also, some Sirtec PSUs also use Jenpo caps on the 3.3V rail - a brand I have only seen in Sirtec.

I've seen them in some newer TOpower P6 based units like be quiet Dark Power P6, also Tagan of course (they were the last to get rid of Topower and are no more)....

[Wester547]

Re: Power supply build quality pictorial. part 2

Also, some Sirtec PSUs also use Jenpo caps on the 3.3V rail - a brand I have only seen in Sirtec.

Jenpo = Jenpan Electronics = JPCON, one of the primary brands involved in the "stolen formula" motherboard capacitor plague. A very bad, avoid-at-all-costs brand.

66°C is not that bad for a heatsink inside a PSU, but obviously could be better, of course, like you noted about the equivalent Delta PSU . Now heatsinks running in excess of 80°C - that's pretty bad. And some cheap PSUs under load will go even much higher than that.

I was referencing the idle temperatures. The secondary heatsink in the Foxlink unit reaches 72°C under load. That's too hot IMO. The Delta's secondary heatsink is still below 50°C under load.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

The datasheet for the TPS3510/3511. The V(DD) pin of the IC is also the 12V UV/OV input.

[momaka]

Task TK-930TX 300 Watt power supply

While on the topic of Sirtec, here is another one: a Task TK-930TX 300 Watt power supply. Date code on its label says 0311 (November of 2003?) PCB inside says “02/12/26”. I don't know if that's a manufacturing date or PCB revision, but either way, it is quite old PSU.

Two pictures of the case shell:
https://www.badcaps.net/forum/attach...1&d=1499391275
https://www.badcaps.net/forum/attach...1&d=1499391275
Note: these pictures were taken after I did a few modifications to the PSU, like the round on the front (I drilled those for extra ventilation).

As far as the case shell quality goes, this PSU uses rather cheap steel. About the same as a $15 gutless wonder. It is not as light as a gutless wonder, though. Nonetheless, there are other signs of cheapness, like the output wires, which are all rated 20 AWG, 300V (but at least they have a UL number on them). Speaking of output wires, this power supply has a 20-pin ATX connector, a 4-pin 12V CPU connector, 5x Molex connectors, 2x Floppy, 1x Auxiliary connector (AT-style), and a 3-pin fan connector (for monitoring the PSU's fan speed).

Moving onto the label:
https://www.badcaps.net/forum/attach...1&d=1499391275
Although there are no combined power ratings to be found anywhere, at least the individual current ratings on each rail appear reasonable. UL number on the label is E144195 – same as on that Jeantech JNP-500AP power supply I posted above. Despite looking like a cheap PSU, the UL and other safety marks do not appear to be forged.

Next, a solder-side shot:
https://www.badcaps.net/forum/attach...1&d=1499391275
The soldering looks pretty decent. Judging by the separation of the traces on the primary side, this PSU probably does indeed have valid certifications.

And now let's see the guts:
https://www.badcaps.net/forum/attach...1&d=1499391275
https://www.badcaps.net/forum/attach...1&d=1499391275
https://www.badcaps.net/forum/attach...1&d=1499391275
I'm getting mixed emotions here. :\ It doesn't look bad, but for 300 Watts, it could be a lot better. The primary heatsink is nice and thick, but lacks fins. Meanwhile, the secondary heatsink has fins but is not that thick. Main transformer appears to be size “35”, but I would say it's a bit on the skinny side. 5VSB is a 2-transistor design *without* a “critical” cap on the primary side, but it has a free-standing IGBT as the switch device. 3.3V rail toroid is nice and big, but the 5V/12V/-12V rail toroid could be bigger. And finally, two of the three input RFI/EMI filter caps are regular metal film instead of X2-class safety-rated. Speaking of caps, all caps inside this PSU are Teapo, except for the two Jenpo on the 3.3V rail (one of which is bulged and leaked badly). At least the PSU has a decent fan: an 80 mm Protechnic Magic MGT8012MS, rated for 0.15A @ 12V.

So let's have us a detailed breakdown:

ICs
PWM controller: SG6105d
-5V rail regulator: L7905

Primary Side
* has male IEC receptacle for monitor cable (remember those? )
* three 1.0 nF safety approved Y2 caps
* one X2-class safety approved cap (0.22 uF, HQX brand, if I remember correctly) + two metal film caps (0.1 and 0.22 uF, again going off top of my head here)
* two CM chokes (one soldered on IEC receptacle)
* PBL405 (4 Amp) bridge rectifier
* two MOVs in parallel with the bulk input filter caps + one SCK057 NTC thermistor
* input wiring: 600 V, 18 AWG
* NO voltage selector switch (PSU has jumper wire for 115V operation)
* primary input filter caps: 2x Teapo LXK, 200 V, 470 uF, 22 x 35 mm (dia. x h.)
* H-bridge transistors: two 13009 NPN BJTs in TO-220 cases
* primary switch: 12N60 IGBT in TO-220 case (rated for 12A continuous at 110°C and 27A at room temp… lol Sirtec, isn't that a bit overkill? ), 2-transistor design
* size “35” main transformer, and tall size “19” for 5VSB and BJT drive transformers

Secondary Side

3.3V rail:
* mag-amp regulated
* rectifier: SBL3040PT in TO-247 case
* filter caps: 2x Jenpo CE-LE, 6.3 V, 2200 uF, 10 x 30 mm (one bulging
* PI coil: 5-turn, 6 mm core, between the two caps
* Min. Load Resistor: 6.8 Ohm (?), 2 Watts.

5V rail:
* rectifier: SBL3040PT in TO-247 case
* filter caps: 2x Teapo SC, 10 V, 3300 uF, 10 x 30 mm
* PI coil: 5-turn, 6 mm core, between the two caps
* Min. Load Resistor: 100 Ohm 1-Watt and two diodes + 1 Ohm resistor all in series to 3.3V rail

12V rail:
* rectifier: UF1606 (?) in TO-220 case
* filter caps:
- 1x Jenpo CE-LE, 16 V, 100 uF , 5 x 11 mm
- 1x Teapo SH, 16 V, 1000 uF, 10 x 20 mm
* PI coil: 28-turn, 6-7 mm core, between the two caps
* Min. Load Resistor: don't remember… 270 Ohm (?), 1 or 2-Watt

-12V rail:
* rectifier: two FR302 (3 Amp) diodes
* filter caps: Teapo SC, 25 V, 470 uF, 8 x 20 mm
* PI coil: more than 6 turns (?), 6 mm core before cap
* Min. Load Resistor: don't remember (output to -5V rail 7905 regulator)

-5V rail:
* rectifier: none – derived from a L7905 linear regulator from -12V rail
* filter caps: Jenpo CE-WH, 25 V, 100 uF, 6.3 x 11 mm on L7905 output
* PI coil: none
* Min. Load Resistor: ??

5VSB rail:
* rectifier: SB240 (?) (2 Amp) Schottky diode
* filter caps:
- 1x Teapo SC, 16 V, 470 uF, 8 x 15 mm (?)
- 1x Teapo SC, 25 V, 330 uF, 8 x 15 mm (?)
* PI coil: 12-turn, 5 mm core, between the two caps
* Min. Load Resistor: 47 Ohm, 2-Watt

I also did some 5VSB tests for this PSU a long time ago, but I don't know where I put them. I did a recap of the PSU after that, so the tests I have now are with the recapped, modified PSU (which only has a 100 Ohm load resistor on the 5VSB rail)

5VSB tests:
No load: 1.8 Watts, 5.5 VA
5 Watt load: 9.5 Watts, 18.5 VA (a little over 50% efficient)

Speaking of my mods on this PSU, here they are:
https://www.badcaps.net/forum/attach...1&d=1499391275
https://www.badcaps.net/forum/attach...1&d=1499391275
Basically, I removed the CM choke on the IEC receptacle, because the wires looked a bit dodgy, and I didn't want to risk seeing them get torn and cause a short-circuit. The X2-class safety cap also got removed and replaced the 0.22 uF metal film cap. Meanwhile, the 0.22 uF metal film cap replaced the 0.1 uF metal film cap.

Other notable improvements:
• added a heatsink on the 12N60 IGBT. It's not that big, but definitely bigger than before.
• Recapped 3.3V rail with one Panasonic FM 6.3V, 2200 uF cap and a 10V, 3300 uF Teapo SC from the 5V rail. Then, the 5V rail got a Panasonic FM 6.3V, 2200 uF cap too. On the 5VSB rail, I replaced the 16V 470 uF Teapo SC cap with a Nichicon HC 10 x 13 mm cap (from a Radeon 9700 video card - that's all I had at the time). The 330 uF cap got booted for a Chemicon KY 6.3V, 1000 uF. And finally, the 12V rail got a 16V, 1200 uF Teapo SM cap in place of the 100 uF Jenpo. Clearly, I thought this was a worthwhile PSU back when I recapped it…more on that in a bit.
• Drilled round ventilation holes on the case's front.
• Closed top vents with plastic sheet to improve flow
• Changed minimum load resistors: 5VSB received a 100 Ohm, 1 Watt resistor; -12V and 12V rails got 1 KOhms each; 5V rail got 100 Ohms, 1 Watt; 3.3V rail I don't remember, but it might have been whatever was on the 12V rail before.
• Finally, I opened up the fins on the secondary-side heatsink for better cooling.

So was all of this worth it? The answer is: NOPE.
After testing the PSU in several different computers (both 5V and 12V –heavy ones), I found that the voltage oscillated very badly. In fact, before replacing the 12V rail 100 uF Jenpo cap with the 1200 uF Teapo SM, I first tried a 1200 uF Panasonic FM. On a 12V-heavy PC, I couldn't even get the PSU to boot the PC – the PSU would squeal and the voltages would vary all over the place. That's when I tried the Teapo SM cap. The PSU seemed a bit more stable and would boot the PC, but it still squealed a lot. I then tried back the original Jenpo, and things improved only marginally. That's when I had enough and ditched the PSU in my spare PSU stash for messing with it another day. It's been sitting like that for a few years now. And TBH, I have acquired better PSUs, so who knows when I will get to this one. I trust even Deer better than this POS – at least Deer will regulate and operate normally when it is rebuilt properly.

So the conclusion is: Sirtec sucks at voltage regulation. Granted the Panasonic FM caps I used may have been too much in terms of low-ESR to cause this, as well as the minimum load resistor swapping… but still, I've never had a PSU from any other brand misbehave like this when modded. Heck, I've put Nichicon HN and HZ in a Deer that had general purpose CapXon KM caps… and guess what? – It still worked fine and stable.

[Wester547]

Re: Task TK-930TX 300 Watt power supply

While on the topic of Sirtec, here is another one: a Task TK-930TX 300 Watt power supply. Date code on its label says 0311 (November of 2003?) PCB inside says “02/12/26”. I don’t know if that’s a manufacturing date or PCB revision, but either way, it is quite old PSU.

0311 = week 11, 2003. I see "0308" (week 8, 2003) on the main and feedback transformer. The unit was probably made in March 2003, with that December 26th, 2002 date being the date the PCB was made. At least the primary heatsink isn't finless altogether.

At least the PSU has a decent fan: an 80 mm Protechnic Magic MGT8012MS, rated for 0.15A @ 12V.

Hey, I have that exact Magic fan model. Currently using it as an intake fan (wired directly to +5V, to cool off the hard drive) in one of my older PCs.

I trust even Deer better than this POS – at least Deer will regulate and operate normally when it is rebuilt properly.

The PSU would probably operate well enough if the exact specs of the original capacitors were matched. The PSU probably isn't worth all that effort, though. It looks more like a 250W PSU to me.

So the conclusion is: Sirtec sucks at voltage regulation. Granted the Panasonic FM caps I used may have been too much in terms of low-ESR to cause this, as well as the minimum load resistor swapping… but still, I’ve never had a PSU from any other brand misbehave like this when modded. Heck, I’ve put Nichicon HN and HZ in a Deer that had general purpose CapXon KM caps… and guess what? – It still worked fine and stable.

If memory serves, similar results came of your recapping experiences with a Bestec.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

Wester called it on the date code. Dates silkscreened on PCBs are fab dates rather than revision dates, in my experience.

Sounds like the Panasonic FM caps are way "too good" for that design. Given the age of the PSU, Nuchicon PJ would be the lowest impedance series I'd suggest, and maybe PM or PS series would be even better matches. What would worry me about a marginal company like that is that the compensation might have been designed for even older parts, and newer/better parts used without checking the compensation.

As you said, there's much to like about the design (the 5V and 3.3V rectifiers, the MJE13009-type bipolar transistors), but the heatsinks probably make 250W-275W more realistic, and AWG #20 output wires are pretty wimpy (UL1007 300V wire seems to be pretty typical for PC PSUs like that one).

[Agent24]

Re: Power supply build quality pictorial. part 2

Are you sure the PSU isn't just faulty?

A squealing noise can't be normal, especially if it happens with higher ESR capacitors too.

[momaka]

Re: Task TK-930TX 300 Watt power supply

0311 = week 11, 2003. I see "0308" (week 8, 2003) on the main and feedback transformer. The unit was probably made in March 2003, with that December 26th, 2002 date being the date the PCB was made. At least the primary heatsink isn't finless altogether.

Cool. Thanks for the info!

The PSU would probably operate well enough if the exact specs of the original capacitors were matched. The PSU probably isn't worth all that effort, though. It looks more like a 250W PSU to me.

Well, I do have some spare Jackon GP caps. After all, wouldn't crap caps fit just fine in a crap design?

And yes, I second the notion that this PSU is rated closer to 250 Watts max. Actually, there is a writing next to the fuse that says "160~250 Watts"

If memory serves, similar results came of your recapping experiences with a Bestec.

Yup. And also a 300 Watt HiPro and a CWT-made MPT-301 PSU (fake Macron).

What would worry me about a marginal company like that is that the compensation might have been designed for even older parts, and newer/better parts used without checking the compensation.

That's very likely a possibility too.

Like I said, I already have some GP 105C-rated caps. Will put those in there at some point and see how the PSU runs. If it runs okay, I might grab a few good-quality GP caps from Mouser/Digikey/TC. If not, then I might have me a fun experiment on compensation - just mess stuff up and see what happens. (ElectroBOOM, anyone? )

As you said, there's much to like about the design (the 5V and 3.3V rectifiers, the MJE13009-type bipolar transistors), but the heatsinks probably make 250W-275W more realistic, and AWG #20 output wires are pretty wimpy (UL1007 300V wire seems to be pretty typical for PC PSUs like that one).

Well, quite frankly, even 250 Watts is more than enough for me - as long as it's 250 Watts of stable and clean power. I have no problem pushing full 12V rail capacity on any of my HiPro PSUs, and they still stay in spec in terms of voltage.

Are you sure the PSU isn't just faulty?

A squealing noise can't be normal, especially if it happens with higher ESR capacitors too.

I haven't tried using higher ESR (GP) capacitors yet.

I don't think it is faulty, because the Jeantech I posted about above also squeals for a second or so when I press the power button on the PC (though it could also be the APFC in that PSU that causes that).

The again, you could be right too, as I have not checked the ESR of the really small caps around the controller. With most of the caps being Teapo and a few Jenpo in between, there's bound to be some bad small caps. I didn't have an ESR meter back when I recapped this PSU.

[Agent24]

Re: Task TK-930TX 300 Watt power supply

I don't think it is faulty, because the Jeantech I posted about above also squeals for a second or so when I press the power button on the PC (though it could also be the APFC in that PSU that causes that).

The again, you could be right too, as I have not checked the ESR of the really small caps around the controller. With most of the caps being Teapo and a few Jenpo in between, there's bound to be some bad small caps. I didn't have an ESR meter back when I recapped this PSU.

I can imagine how funny noises at startup could be normal for some designs, as the power supply comes up to normal operating frequency and stability etc, but to squeal continuously does not seem right to me.

The only normally audible "PSU" I know of is a CRT TV's flyback.

I wonder if even it has a somewhat leaky semiconductor, maybe one of the rectifiers?