Astec AA16990 PSU failure

[momaka]

Re: Astec AA16990 PSU failure

There is just only ONE electrolytic cap on this board - 10uF, 25V, Nichicon. Replace?

Worth trying anyways. Save the old cap, though, in case it is still good. I see a few more small caps, though - one near that daughterboard on the secondary side and two more on the primary side.

Hmm, it looks like it will be difficult to get a new "AS431". Ok to replace it with a "TL431"?

Yup.

TL431, KA431, and AS431 are all the same thing, just made by different manufacturers. It's the "431" in the name that designates it's an adjustable shunt.

Same goes for your AS3842 PWM controller - it's an equivalent to a UC3842.

Another point I forgot to mention: are there any potentiometers or variable resistors anywhere in the PSU? Sometimes, those are used for fine-tuning output voltage regulation. Pots are a lot more susceptible to damage from moisture and dust, so if your PSU has one, check it.

[stj]

Re: Astec AA16990 PSU failure

no problem, everybody makes the 431 - but sticks different letters on them.

beaten to the post-button

[powermax]

Re: Astec AA16990 PSU failure

are there any potentiometers or variable resistors anywhere in the PSU?

I didn't see anything having a screw in this PSU.

[powermax]

Re: Astec AA16990 PSU failure

TL431, KA431, and AS431 are all the same thing, just made by different manufacturers. It's the "431" in the name that designates it's an adjustable shunt.

I just searched the web for both "TL431" and "KA431". Digikey and Moser offer this component in different packages. There are two TO-92 ICs having four-letters-suffix like "KA431AZTA" or "KA431LZTA". They look similar but I'm afraid there is some difference. I don't want to destroy my PSU by installing an incompatible IC. I'm completely lost...

[PeteS in CA]

Re: Astec AA16990 PSU failure

"TL431" is just TI's version of a very old (1970s design) part created by Signetics. Lots of companies make equivalents (Samsung's was a KA431, IIRC). They should be interchangeable.

The LM2903 is a dual comparator, while the LM339 is a quad comparator. The performance of the individual comparator sections of the two parts should be more or less identical. A triac, once fired, can conduct in either direction, making it suitable for AC (unlike an SCR, which conducts in one direction only). And once a triac has fired, it doesn't turn off until current is interrupted (basically, until the power supply is turned off).

[PeteS in CA]

Re: Astec AA16990 PSU failure

The "KA431AZTA" and "KA431LZTA" are compatible parts, now made by On Semiconductor (originally Samsung, then Fairchild bought the rights to the part from Samsung, and then Fairchild was bought by On Semi). The "AZTA" means 1% tolerance; the "LZTA" means .5% tolerance.

[powermax]

Re: Astec AA16990 PSU failure

I got an old Nippon Chemicon cap, 10uF, 25V, KMA Series (see attached image). Okay to replace it with this one?
http://www.mouser.de/ProductDetail/U...XjttuxTw%3d%3d

[momaka]

Re: Astec AA16990 PSU failure

It probably would... but what brand and series was the original cap?
The KMA you pointed to seems like a general purpose cap, and only has a rated endurance of 1000 hours.

I suggest going with 10 uF, 50V Chemicon KY, as they are usually readily available.

[powermax]

Re: Astec AA16990 PSU failure

but what brand and series was the original cap?

Like I said in my previous post, it's a Nippon Chemicon, "KMA" cap.

I suggest going with 10 uF, 50V Chemicon KY, as they are usually readily available.

Okay, there are two different ones:
1: http://www.mouser.de/ProductDetail/U...aeVizgO7F8s%3d
2: http://www.mouser.de/ProductDetail/U...167eVm9Bdhc%3d

The 2nd one has the "Life" of 5000 hours while the 1st one has no such indication. Which one would you recommend?

[momaka]

Re: Astec AA16990 PSU failure

Like I said in my previous post, it's a Nippon Chemicon, "KMA" cap.

Ah okay. So they had a general purpose 105°C-rated cap in there. Not a bad idea to change it then.

Okay, there are two different ones:
1: http://www.mouser.de/ProductDetail/U...aeVizgO7F8s%3d
2: http://www.mouser.de/ProductDetail/U...167eVm9Bdhc%3d

The 2nd one has the "Life" of 5000 hours while the 1st one has no such indication. Which one would you recommend?

They are the same cap. Only difference is how they are packed. The first one comes in "bulk" - i.e. individual caps in a bag. The second one is for caps to come in a paper "reel" - this is used where a machine automatically can take the cap from the reel and place it on the board... i.e. for production/industrial use.

So you can go with whichever one you prefer. I typically go with whatever is cheaper.

[powermax]

Re: Astec AA16990 PSU failure

I see a few more small caps, though - one near that daughterboard on the secondary side and two more on the primary side.

You're right! I've found four "KMA" caps in this PSU in total:

C207 (on the daughter board) - 10uF, 25V
C11 (near the daughter board) - 47uF, 25V
C46 (on the primary side) - 100uF, 10V
C30 (near the 12V output, see my schematic) - 10uF, 35V

I'm going to replace them all to eliminate this source of problems. C207, C46 and C30 can be easily replaced with the appropriate 50VDC, "KY" caps as you said because their diameters match and the length is not important.

In the case of C11, the "KY" equivalent is unfortunately too long (11mm vs 7mm). Is there another replacement possibility for this one?

[momaka]

Re: Astec AA16990 PSU failure

In the case of C11, the "KY" equivalent is unfortunately too long (11mm vs 7mm). Is there another replacement possibility for this one?

KZE series are sometimes smaller for the equivalent capacitance. But all I could find for KZE was a 56 uF, 25V for 7 mm height. 56 uF is just a step-up from 47 uF. And typically, a step-up in capacitance very rarely causes issues. So you may be able to use it if nothing else is possible. Here is the link for the 56 uF KZE with 7 mm height:
http://www.mouser.com/ProductDetail/...EbjMs2LB624%3d
Mouser part #: 661-EKZE250ELL560MF0

That said, I still suggest to go with a 47 uF cap, if possible.

Why is the height of C11 a problem?

I see that C11 is next to the daughterboard. If there are components above it, you can just tilt the new (bigger) cap a little bit to avoid that. So that way, an 11 mm cap should fit no problem.

[powermax]

Re: Astec AA16990 PSU failure

I replaced four small caps (C46, C11, C207, C30) and "AS431B" with "TL431AI".

Good news: the fan spins, nothing has exploded so far.

Bad news: the rails are still low (+5Vrail = 4.34V, +12Vrail = +7.20V, -12V= -4.4V). The voltages fluctuate quite much.

I'm testing the PSU without load. I've read somewhere that these ol' PSUs don't operate correctly without a proper load. Is it true?

[powermax]

Re: Astec AA16990 PSU failure

When I disconnect the fan, the -12V rail shows "-11,4V". With the fan connected, it has only -4,4V...

[Per Hansson]

Re: Astec AA16990 PSU failure

Have you checked the batteries in your meter?
Because I would not expect the computer to work with voltages so much out of whack...

When I wrote the above I did so believing you where measuring these voltages with the computer connected as a load.
If you are saying you have done all this testing with no load at all then yea, that explains things.
The PSU needs a load to get the voltages stable, this is true for most cheap/older power supplies.
Some "solve" this by having a "minimum load" resistor for each rail.
But by "solve" that is a pretty crappy solution.
Because it's only really needed when you don't have a load.
So in effect it just produces allot of heat inside the powersupply making the caps fail much sooner than they would have otherwise.

Now please don't try to power your computer with the PSU just because I have said this.
You need to verify that the voltages actually are stable.
So try to find some large resistors to test with.
Lamps from a car can also be used but be careful because the inrush current is extreme and could actually damage the PSU.

[powermax]

Re: Astec AA16990 PSU failure

So try to find some large resistors to test with.
Lamps from a car can also be used but be careful because the inrush current is extreme and could actually damage the PSU.

Yes, such a "testing" circuitry built using incandescent bulbs has been already mentioned several times in this forum. Unfortunately, I couldn't find anything concrete.

Do you have any tips or helpful tricks on how to build such a testing circuit for my PSU?

[Per Hansson]

Re: Astec AA16990 PSU failure

No that's only when repairing a PSU (The light bulb, wired in series with the PSU will light up only if the PSU is shorted, thus preventing whatever is shorted from exploding)

Just put some resistors as load directly on the output cables where the mainboard would normally go...

[powermax]

Re: Astec AA16990 PSU failure

No that's only when repairing a PSU (The light bulb, wired in series with the PSU will light up only if the PSU is shorted, thus preventing whatever is shorted from exploding)

Ok, thanks! BTW, This post does explain in details how the mentioned bulb test works.

Just put some resistors as load directly on the output cables where the mainboard would normally go...

Okay, is it correct that I'd need to attach a 1 Ohms/25W resistor to load the 5V with 5A and a 6 Ohms/24W resistor to load the 12V rails with 2A?

[PeteS in CA]

Re: Astec AA16990 PSU failure

...
Okay, is it correct that I'd need to attach a 1 Ohms/25W resistor to load the 5V with 5A and a 6 Ohms/24W resistor to load the 12V rails with 2A?

Your calculations are correct, but 25W resistors at 100% rated power would get extremely hot. 50W resistors - panel mount types mounted on a heatsink or tubular (this style) - would be a better choice.

Or maybe:

* Two 2 ohm, 25W resistors in parallel for the 5V;

* Two 12 ohm, 25W resistors in parallel for the 12V.

[powermax]

Re: Astec AA16990 PSU failure

Your calculations are correct, but 25W resistors at 100% rated power would get extremely hot.

Yes, I learned to pick at least the doubled wattage because I'm not going to "fry an egg" with the load resistor

Or maybe:

* Two 2 ohm, 25W resistors in parallel for the 5V;

* Two 12 ohm, 25W resistors in parallel for the 12V.

Good advice, thanks a lot! The question is that the 5V rail of my Astec PSU has the max. power of 11.6A. Should I test this PSU with a load close to its maximum, say 10A? If yes, what combination can I use to meet the required resistance of 0.45 ohm, 55W?