Power supply build quality pictorial. part 2

[RukyCon]

Re: Power supply build quality pictorial. part 2

Looks a bit bare for a 500w unit, particularly the transformer, heatsinks, and primary capacitor all seem a bit small. IMO, It might be fine for up to 400w, but probably not too much more.

Speaking of Segotep, I have/had a Segotep 750w unit that I had bought for a computer I was building for my older sister, and before installing it, I took the lid off a took a couple of photos, and it overall didn't look like a bad PSU, other than the capacitor choice, which was a mix of Chengx with a few SC caps mixed in. And while I probably should've recapped it before giving it to my sister, I didn't because I didn't want the PSU to blow up the moment I recapped it, especially since it was still under warranty (it did not have any warranty stickers on it to stop me from opening it up, which is why I did). So odds are, it will probably be back here in a few years when I inevitably have to recap it.

[Dan81]

Re: Power supply build quality pictorial. part 2

Segotep is almost like Intertech - the more cash you pay, the better the unit. The less you pay, the worse it looks inside. And trust me, I've laid eyes on some units that were more atrocious than leotron's unit.

As for OEM, couldn't find either designs... I highly doubt leotron's unit is made by Seasonic (it looks too barren to me) while yours apparently traces back to a brand named "Colorful".

[eccerr0r]

Re: Power supply build quality pictorial. part 2

Well, not the first one to present a FDPS-100N-R5 but I picked this up from a hamfest. It was dead as a doornail on first powerup. Apparently quite abused and found a broken resistor, then found it did not hold up voltage. Pretty much instantly blamed C9, a 47uF 35V capacitor, and after replacing that it fired right up.

Not sure if this psu was meant to really be open frame, was there a case for it? Not sure how the rectifier is normally cooled, that thing should get quite warm if this psu really can get to 8 to 10A... how many watts should I be able to get out of this thing?

At least it was easy to fix. Interesting PSU, wonder how this (TL494) compares to the UC3842 based SMPS...

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

Looking at that SEGOTEP "500W" P/S, the input lytic, TO-220 switch devices, heatsinks, and output inductors (-26 material, suggesting transistors and a switch frequency under 50KHz, probably 70%-75% efficiency) look more suited to 235W-250W continuous.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

The TL494, like the SG3524, is voltage mode and 1970s technology. The UC384# family are current mode and 1980s technology. Voltage mode means there is just one feedback loop, sensing the output voltage. Current mode means there is also a feedback loop for the main transformer primary current, which allows limiting the current for each current pulse. Having just the one feedback loop, TL494 type PWMs have somewhat simpler compensation. OTOH, the maximum practical switch frequency for 384# family PWMs is higher, which means MOSFETs can be used to fuller advantage.

[eccerr0r]

Re: Power supply build quality pictorial. part 2

Interesting, the current/voltage mode never made much sense as current is instantly converted to voltage with that current sense resistor...which is used as a voltage into the control circuitry! But indeed there are two sense inputs on the UC384x.

The UC384x PSUs I see, invariably the controller chip is on the hot/line side so it can monitor switch current. However I always wondered about how much power is being burned supplying the UC384x, though usually less than 2 watts or so. The TL494s seem to always be on the cold/isolated side of the PSU though it has two error amps (and two outputs).

Since the UC384x only has one output, it can only drive one MOSFET without external devices. The TL494 has two outputs and I thought pretty much all of the ones I've seen were BJT switches in push pull...probably losing power in base drive.

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

The UC3846 and UC3847 are part of the "family" and can drive half or full bridge inverters.

The typical power scheme for 384# PWMs uses a fairly high value resistor connected to the rectified AC that charges a capacitor that is paralleled with a zener to provide start-up power. There is a winding on the transformer that then provides Vcc for continued operation.

IIRC, in TL494-based half bridge designs, the bridge self-oscillates for a few cycles, which provides Vcc to the TL494, which then controls the inverter. One thing to watch for in these designs is that after turning off AC power, that self-oscillation can start again while the input lytics' voltage decays, resulting in output voltage a second or two after AC was switched off (BTDTGTTS).

[eccerr0r]

Re: Power supply build quality pictorial. part 2

That was kind of a surprise to me when I was trying to debug this last TL494 PSU, after seeing so many ATX PSUs where the 2-transistor or a viper or the such provides power for the TL494 which this circuit does not (since it is not ATX and uses a mechanical switch to power up/down). Thus was expecting that same line power charger as the 384x uses but it couldn't have been done this way since that would break isolation... until it dawned on me that there were too many components there to simply pass the transformer coupling to the switch transistors...

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

The self-oscillating start-up was new to me until 1997 or 1998 when I saw the restart after power-off thingy happen with a Mitac power supply I was testing while working at Sun Microsystems.

My experience with voltage mode PWMs has been mostly with 3524 family ICs rather than TL494 types. TL494 types have been around since the 70s, and 384# family types since the 1980s. They're good examples of devices that are just too effective in performance and cost to become obsolete after a short time.

[eccerr0r]

Re: Power supply build quality pictorial. part 2

And nowadays people are thinking TL494? Too many pins!

[RukyCon]

Re: Power supply build quality pictorial. part 2

I'm surprised that companies still use TL494 or equivalents as the backbone of some more modern power supplies, when it seems like it would be cheaper to use something like 384x series parts (unless I'm wrong).

And just for example. Last year, I took apart a dead water bottle cooler I found in the trash (possibly made around mid-2010s, outward design seemed pretty modern), and the main PS board (12v, possibly 100w max) relied on a KA7500 as the main controller.

Oh, and the reason it was dead was because the only secondary capacitor had failed (can't remember the series, but it was an Aishi 2200uf 16v).

[eccerr0r]

Re: Power supply build quality pictorial. part 2

btw that TL494 PSU above, measured 72W at the wall, 13.2V@4A output to a head light bulb=52W = 73% efficient... not so great but meh.

KA7500/TL494s are everywhere... just don't see any 78S40s ...

[PeteS in CA]

Re: Power supply build quality pictorial. part 2

TL494 & clones are cheap, well known, and the design engineering effort is comparatively minimal. The downside is that TL494 & transistors half bridge designs are relatively low efficiency (70%-75% was good, ca. 1980, but that was >40 years ago). Really good efficiency nowadays means using phase modulation, quasi-resonant inverters, and synchronous rectifier MOSFETs.

[leotron]

Re: Power supply build quality pictorial. part 2

Since the segotep doesn't have anything below the board my dream on building power ful PSU is become a reality

[Dan81]

Re: Power supply build quality pictorial. part 2

Looking at that SEGOTEP "500W" P/S, the input lytic, TO-220 switch devices, heatsinks, and output inductors (-26 material, suggesting transistors and a switch frequency under 50KHz, probably 70%-75% efficiency) look more suited to 235W-250W continuous.

I'd be careful pulling more than 180W max from that.

[Dan81]

Re: Power supply build quality pictorial. part 2

Raidmax RX-700AC. Not 700W by any means, but my guess would rather be somewherenin the 400-500 range. APFC design, Teapo caps all around, pretty good heatsinks.

Considering I had to choose between this and a Gigabyte P650B, I'm rather happy I went with the Raidmax. They really have improved over the years. Not much, but it is quite a lot when you consider they were pushing absolute safety hazard Sun Pro units (I still have the casing from one, currently houses a Spire Jewel inside.) - maybe Corsair's roast taught them something?

And to end this post on a good note - is it me, or did they copy a FSP design? It kinda reminds me of the 60GLN series.

[RukyCon]

Re: Power supply build quality pictorial. part 2

The design looks very similar to a Coolmax 600w PSU i picked up a while ago and have been meaning to post here for just as long, even though i have everything i want to say about it written down, the only reason i haven't put a post together yet is because i wanted to remove a couple of parts to get a look at their part#.

[Dan81]

Re: Power supply build quality pictorial. part 2

It's also sold under the KRPW/Kuroutoshikou brand in Japan I think? I remember Tech YES City getting one and straight away I noticed it was the same Andyson platform as my RX-700AC.

They're good PSUs honestly - this lil' bugger managed to power a R9 280x, Phenom X6 1055T, 5x 1TB HDDs and a DVD-ROM without breaking a sweat! I've set it aside for a Athlon 64 x2 4600+ 939 build w/ a 8800GTX.

Maybe it's me, but it seems XHY and Andyson really are capable of making good units. The XHY-based unit I have is a 750W Intertech FP-750W w/ a 14cm fan, and even that one is a powerhouse, despite the PCB itself being rated for 400W - I'm fairly sure that's just the design they used, as the sillicon inside does look enough for pulling upwards of 550W to 650 in the actual configuration. Only the caps are a bit sus on it - BH/Nicon. It was lightly used though - the unit I pulled it from didn't have much inside, aside from a MSI P55-CD53 and a i5 750.

[pdavid]

Re: Power supply build quality pictorial. part 2

I've got a decently build AT PSU:

SP2-425F EMACS.
Has an automatic input voltage selection circuit.
Very decent EMI input filtering.
Input buffer capacitors are Rubycon USR 680uF.
Half bridge switching two 2SC2625 NPNs.
Center tapped 44 size main transformer.
Some of the output capacitors are swollen, all PCE-TUR .
TL494 PWM for primary switching.
LM339 in the secondary for protection and voltage supervision.
Lots and lenghty peripheral conectors. AWG 18 wiring.

I haven't a clue who is the OEM. 1999 date code on the pcb.

[Dan81]

Re: Power supply build quality pictorial. part 2

No-name special

Why is it special? The insides shall tell you everything.




No case swap involved, that is really how I found it, in a rather nice MSI K8T Neo-FIS2R + NX6600GT AGP custom build.

FWIW, according to the fuse chart, this thing is realistically just 50W short of its rated power. Not bad, unlike most china specials I've had to work with.