Enermax Liberty ELT400AWT making clicking noise but still running

[Agent24]

A couple of days ago I turned my old Athlon 64 machine on and soon after power-on I heard a loud clicking noise.

It sounded very much like something knocking against fan blades, so I disconnected the case fan, put my finger on the CPU fan and a pencil in the PSU fan.

No, the noise still happened.

It seemed to be coming from the PSU itself. The PC was running 'perfectly' but as the PSU seemed to be the problem I immediately turned it off.


I disconnected the PSU and powered it up with a jumper. It still made the clicking just not as much and maybe quieter.

I found the higher the load, the faster and louder the clicking.

Seems very odd that the PSU still runs while making such an 'bad' noise!


I searched the forum and found a similar problem posted here: http://www.badcaps.net/forum/showthread.php?p=250827

I also got this response:

Originally Posted by ben7

Check the diode in the pfc circuit.

But I want to start a separate thread here to avoid confusion.


So, where is the PFC circuit and what kind of diode am I looking for?

[Behemot]

PFC is usually on the primary side, either on separate heatsink, or with switching transistors.

[ben7]

Originally Posted by Agent24

So, where is the PFC circuit and what kind of diode am I looking for?

It will be mounted on the primary heatsink, its just a single diode, not a bridge-rectifier. You can track the leads from the PFC choke, and you will find that diode, and a mosfet.

It will help if you temporarily remove the primary caps to get a better view behind them.

[c_hegge]

It looks like a transistor, but has two legs. In this picture, it's the part to the far right (the other two parts are the PFC transistors)

[Agent24]

Thanks

Will have a look and post photos

[ben7]

Originally Posted by c_hegge

It looks like a transistor, but has two legs. In this picture, it's the part to the far right (the other two parts are the PFC transistors)

That link is bad.

[c_hegge]

OK, Fixed. I attached it (it works here, though)

[ben7]

Also there should be a high power low ohm resistor for the current sense in the pfc circuit, you might want to check that. If its value has gone up it could fool the control circuit into thinking there is a short circuit.

[Behemot]

Wouldn't it think its shorted when the resistance would go down?

[ben7]

Originally Posted by Behemot

Wouldn't it think its shorted when the resistance would go down?

No, remember, ohms law!

It uses a resistor to measure the current; the circuitry measures the current by measuring the voltage across the resistor. In fact, it wouldn't think its shorted when its a very low resistance, then it is possible for the mosfets to blow!

In the control circuitry there is a pre-defined trip voltage. Say the power supply primary has a max power of 100w. With a mains voltage of 120vac, that would be rectified into DC, around 165v. The max load of 100w with a supplied voltage of 165v would draw about 0.6A. We use E=IR to find the resistance of the current sense
resistor that we need. The voltage we want to trigger the overload at is 1v. 1v = 0.6A * R, so, R = 1.666. As the current goes to about 0.6A, the voltage across the current sense resistor will be about 1v. Then the protection circuit will kick in.

But, say if the resistors resistance increased to 3 ohms.
Finding the new trip current:
1v = I * 3ohm
so, I = 0.333A
That means that at only 0.333A, there would be 1v across the resistor, making the overload protection trip prematurely! The protection circuitry would think that there was an overload, but the power draw of the psu was really only around 54 watts!

Edit: and some psu's with PFC will actually work without the pfc running, but of course it will have a bad power factor.

[Behemot]

Didn't know how the resistor is used in there before, thanks

[ben7]

Originally Posted by Behemot

Didn't know how the resistor is used in there before, thanks

You are welcome

[Agent24]

I got the thing open and figured out how to extract the board from the case...

I believe I've found the PFC diode, an I.R. HFA08TB60 - hidden behind the main capacitor, the PFC coil, and a little sub-board (The PFC control circuitry??) - see photo

Originally Posted by ben7

Also there should be a high power low ohm resistor for the current sense in the pfc circuit, you might want to check that. If its value has gone up it could fool the control circuit into thinking there is a short circuit.

There's a large (~1 watt size) resistor hidden under the primary-side heatsink just behind the HFA08TB60.

Coding seems to be Brown-Orange-Silver-Gold-Green which doesn't match up with any chart I know - what value do you suppose it should be?

EDIT: Totally missed another, larger resistor just below the PFC coil. Code is Orange-Blue-White-Gold-Green.


What should I be testing\checking the diode for?

[Behemot]

Short or partial-open. Use diode tester, what does it read in both directions? If I am nto mistaken, ordinary diode should read arround 300 ohm minimum.

[Agent24]

Going by the datasheet this thing is a bit more interesting than an ordinary diode... that said...

Diode check on multimeter says nothing in reverse bias and "448" in forward bias.
Resistance test says infinite in reverse and 530k Ohm in forward.

A standard 1N4004 shows "587" on the diode test and 618k Ohm on resistance. (forward bias)

I do believe though that some faulty semiconductors will appear fine on a tester but fail in-circuit because of the higher working voltages. How to test that I am not sure.


The resistors both measure very low values:

* Brown-Orange-Silver-Gold-Green reads as 0.12 Ohm
* Orange-Blue-White-Gold-Green reads as 0.03 Ohm

[Behemot]

It is more or less typical ultrafast to me…

The first resistor looks like non-standard 0,13 ohm. Silver seems to act like ×0,01.

The second is 36.9 ohms, ± .5%, according to calculator.

[ben7]

Originally Posted by Behemot

It is more or less typical ultrafast to me…

I think the same. It's probably good.

The BR-OR-SI-GD-GR resistor is supposed to be a 0.13 ohm 5%.
The OR-BL-WT-GD-GR resistor is supposed to be a 36.9 ohm 0.5%.!?

This will help with those odd 4 and 5 band resistors

[Agent24]

That second resistor (actually R14 on the board) is an oddly, very specific value... and shorted too?
Maybe it's supposed to be 0.0369 ohms 0.5%... no, that probably makes even less sense.

Even if that is the case.. where the heck do I get something that specific.. and WHY does it even need something that specific?

Maybe it's time to draw out a schematic and try to figure out what that resistor does...


Probably time to let you all in on a little secret... I actually have another Liberty 400 lying around... it was my friends and it blew a hole in a MOSFET. I didn't go any further than noticing that (and suspected the brown glue of death) but I guess I could have a look at its R14 resistor and see if it's the same.. of course it may well have had a similar problem, so that might be a waste of time.


Aside from the resistor, could there be anything else that's easy to check?

[Behemot]

The marking should be right, we used two independent calculators. If you won't find exactly same, you can always buy 33ohm plus 3,9ohm in series, you should easily get these two.

[ben7]

Try measuring the resistance of it the other way around.

Also, what is this resistor connected to?