Power supply build quality pictorial. part 2

[tom66]

So, we've looked at switching power supplies, now let's look at a lab bench linear power supply.

I've owned this thing for about 18 months, but it was second hand, so no idea how old it really is.

It is a 0-35V 0-3A constant-voltage/constant-current power supply, manufactured by Tagasaki of Japan. The model number is GM035-3. And it only cost me £30. I can find very little information on the internet about this model, so I suspect it's mid 1980s, especially given the Chemicon caps in it are mostly SM series, which was obsoleted a while back.

The engineering in this thing is fantastic. The transformer is huge - especially for a 105W max output! There are three Toshiba power transistors in parallel on a massive heatsink (with OR-ing resistors), controlled by one main board. Current sensing is high side, not some low side stuff which might ignore earth shorts. (The output can be fully floated, which has been useful several times.) And a very clever way of reducing the power dissipated in the transistors: two SCRs select between three taps on the power transformer, reducing the input voltage to the transistors at lower output voltages. The control logic is surprisingly simple, comprising a few op-amps and a TL431 reference. The potentiometers on the front are only single turn, but I haven't needed precision from it yet, especially given the gauges are only useful as a general guide.

I'm impressed at how low noise it is (below the noise floor of my scope in most cases, even approaching full load), how reliable it has been (no problems, even after shorting the output @ 35V so many times), and how well built it is.

I've run a universal motor off this drawing 2A at 35V. That's the maximum output load I've been able to demand from it so far, and it coped very well. My only wish now would be for a higher output voltage, up to 60V (SELV limit, but still painful!), for testing some high voltage input dc-dc converters. And blowing up cheap Chinese PSU fans . (Sometimes, 35V isn't enough!)

[momaka]

Wow, that is a beautiful PSU!
My L&Cs are truly ashamed.

[Behemot]

ACBel from Lenovo A55 desktop, 225 W total max.. Typical ACBel, somewhat decent, some small Chemi-Cons, Rubycons, one Chemi-Con on output, rest is L'Tec (one already bulged). Input caps Elite.

[goodpsusearch]

TMP-ANS LC-A350ATX

The good thing about this power supply is...... that it is not going to power a computer anymore

Not worth of anything!

It reminds me of those power supplies:
http://www.badcaps.net/forum/showthr...ier#post114292
http://www.badcaps.net/forum/showthr...ier#post199701

[ratdude747]

Originally Posted by goodpsusearch

TMP-ANS LC-A350ATX

The good thing about this power supply is...... that it is not going to power a computer anymore

Not worth of anything!

It reminds me of those power supplies:
http://www.badcaps.net/forum/showthr...ier#post114292
http://www.badcaps.net/forum/showthr...ier#post199701

Sledgehammer Bait.

I see nothing redeeming about that PSU, other that that at least the PCB is full size and that there is no lead weights inside.

This PSU belongs in the hall of shame...

[PeteS in CA]

Come now, rats! There's not a lot of stuff, you know, beefy components to block airflow ... assuming the fan turns. "LC" = Lousy Crap, right?

[Agent24]

From Chinese philosophy, less is more?

More smoke?

[momaka]

Originally Posted by goodpsusearch

TMP-ANS LC-A350ATX

Looks very similar to the LC-B300ATX I posted in post #491. It's funny they rated yours for 12A on the 12V rail, yet it only has the 3A "diodes-on-a-bracket" for a rectifier .
What brand of fan was on yours? Powelogic, Rulian Science, or something else? If Powerlogic, keep it. They are actually pretty decent sleeve bearing fans. Properly oiled at the facory so they don't seize up like other cheaper sleeve bearing fans.

[goodpsusearch]

The classic sleeve bearing globe fan!

Just like this:

[Behemot]

Yeah exactly the same one, branded Premier (however only 300 W), I just removed from one computa. I would dare to say, this one became classic now:-D

[rogfanther]

I raise you an even crappier PSU :


http://www.badcaps.net/forum/showpos...10&postcount=1

[ ]

[Behemot]

I think that one is only some misunderstanding from the guy designing the PSU's label. I believe, it is not like CT-2045S means 450 W peak, but CT-2045S means 200 W peak

[rogfanther]

At least it would be a more honest labeling. I contemplated rebuilding it as a learning exercise, but after changing rectifiers, heatsinks and caps, it still would not have places for filtering coils. When I finally build a load tester, maybe I will measure ripple in it just for science advancement or for the nice fireworks....

Too bad they still sell these things for 450W and people try to use it in their quadcore machines....

[c_hegge]

That coletek won't even do 200W with the ripple in spec. It's a very low end CWT unit. compare http://translate.google.com.au/trans...26prmd%3Dimvns

[pdavid]

Since we started to throw out a bunch of old e-waste that piled up in the years I managed to save some AT psus recently. From Compaq and some unknown desktops mostly. These are all working models.

Here is an Enhance unit from circa 90-91. Half bridge, usual AT abilities and connectors. Has an interesting setup of diodes for the 12V rail. Nitsuka primary and Luxon secondary capacitors Voltages are fine, the fan runs at full speed all times.

[pdavid]

Liteon AT psu for Compaq. Components are brutally overspeced for the labeled wattage. It has a single 6 Amper mosfet in the primary. I wonder about the efficency. Caps are NCC except a 1uf 450V Jamicon in the primary. This psu won't start without a load, first I though it's faulty. There is no Power Good wire. The fan is temperature controlled, a thermistor is tied to a small capacitor near the output filters. 40A for 5V and 15A schottky on the 12V rail... okay. Would somebody lighten me up on this topology?

[momaka]

Originally Posted by pdavid

Here is an Enhance unit from circa 90-91. Half bridge, usual AT abilities and connectors. Has an interesting setup of diodes for the 12V rail.

And yet there are still cheap power supplies out there that use the 2 diodes on a bracket for the 12v rectifier. At least this one has 4. Still better than a no-name cheapo PSU.
The traces on the underside are funny on that Enhance. Reminds me of old-school 8-bit 2D games .

Originally Posted by pdavid

Liteon AT psu for Compaq. Components are brutally overspeced for the labeled wattage.

That's pretty typical for an old Lite-On PSU. The Lite-On PSU in an old Dell Dimension XPS R400 I have is build like that too.
Topology is single-transisor forward converter. I think it's a little newer than the ancient half-bridge, but still very old. Efficiency is probably the typical 75%<.

[rogfanther]

Originally Posted by c_hegge

That coletek won't even do 200W with the ripple in spec. It's a very low end CWT unit. compare http://translate.google.com.au/trans...26prmd%3Dimvns

I have some boards of that model in the junkbox. At least it has some bigger heatsinks. But that is an example of the "normal" psus found in computers around here, and in every store in the streets. And when asked, sellers will swear it is "450W", and many will keep talking about the "450W nominal" as if the computers really used it.

[PeteS in CA]

That's pretty typical for an old Lite-On PSU. The Lite-On PSU in an old Dell Dimension XPS R400 I have is build like that too.
Topology is single-transisor forward converter.

Yep, single-switch forward. The PWM is a 3843, current mode, single-ended. The dual ended (for half- and full-bridge designs) PWMs in that family are the 3846 and 3847. The label, construction and parts choices suggest that the P/S really could do the maximum current of all O/Ps at the same time, i.e. 200W. I'm going to guess that the 7905 and 7912 for the -5V and -12V O/Ps are rated for .5A, but are limited in this application but not being mounted on a heatsink. The date code on the inductor suggests that it was built in April or May of 1997. I think UCCs LXF series was their best low impedance series at that time. All in all, the kind of solid design I would expect of Lite On from that era. Don't know if they are still good.

[Th3_uN1Qu3]

Powerlink LPJ2 "400W".

4-diode treatment with lame 4A diodes. Non safety approved Y caps. No other input filtering to speak of.

Primary: D13007s as main switcher, C5027S on standby. EI-33 label on the transformer... i ain't buyin' that. It's an ER-28 actually. 2x 470u Canicon caps, but judging by the can size real capacity is lower.
Secondary: 20A schottky on +5v, 10A schottky on 3.3v, and 12A fast recovery on 12v. 12v diode is stuck directly to the heatsink without a silpad. Small output inductors (coupled inductor is on a T90 core). No pi filters. 2x 1000u 10v Fuhjyyu TM on +5v. 1x 1000u Fuhjyyu TM on +5vsb and 3.3v, both bloated. 1x 2200u 16v Canicon on +12v, and 1x 470u 16v Canicon on -12v. Weltrend controller.

Can you believe this thing was trusted to power a C2D E6400 and 7600GS?