Repairing Krell Amplifier

[ImpulsiveJames]

Long story short, I have a friend that got a broken Krell Amp, Evolution 403e, and is pretty much immediately going into protection mode. When using my thermal camera, it looks like a ton of mosfets (they looks like mosfets, I haven't pulled the thing apart enough to triple check) get up to 200f pretty darn quick, about 15 seconds or so.

I believe I narrowed the issue down to the capacitor board. It's essentially just an entire PCB filled with 100v 4700uf caps, 36 of them in total.

Time for the idiot check... I was under the impression if you test a cap with a multimeter, and it says it's shorted, it's garbage... When I test the caps (while in circuit and both correct polarity and reverse) they all show shorted, but I feel like when I do that for long enough, they eventually stop showing shorted.....

So I guess the question is, if I get enough charge into some of these, is it then reading correctly, or what? There is no way I'm charging these all the way up with a normal DMM and 10 mins of testing.

And yes, I'm aware that I have to remove them from the board to test accurately and yes, in circuit the short could be coming from anywhere, but I just unsoldered one of the caps and man...it sucked, I'm not looking forward to 3 hours of desoldering....

[cncjerry]

I have Krell amps, the larger balanced KSA100's and KSA50s, I think I have a 160B as well.  These amps draw so much current they will unsolder the connections if they aren't soldered properly.  A high resistance solder joint will melt right apart. My son was playing rap with really low bass on one and that happened. Once the main supply unsoldered and disconnected, it tried to draw all that current through the bias supply.  That lasted about a millisecond.  Took a while to repair it. 

Those caps are too large to test in parallel with a meter.  Use a power supply that has current limiting.  You will see high current at the start and it should tail off to nothing after the caps are charged.  Before you do this, put a bleeder resistor of some value across the connection, maybe 1000ohms 1w or whatever you have.  That will keep you from having to short them to draw them down to work on them.  Make sure the polarity is correct at all times.  If you have a bad cap, it should show up as swollen or leaking.   If you don't have a current limited power supply, try a battery charger for lipos or nimh.  They generally have current limiting for the charge rate, keep it below and amp.  If it continues tom draw a lot of current, assuming more than the bleeder, start isolating them in groups. Be careful of polarity.  Be careful of polarity.  That is three times now, ha! As you charge the bank, the charge rate should have a curve, high to low.

Most likely you have a blown output and driver.  Check the outputs a to b channel resistance.  Should find it in minutes.  95% or more stereo trips are blown outputs and/or drivers.  Save the PS for last.

Jerry

[ImpulsiveJames]

It's funny you mentioned the connections desoldering themselves....I found this last night while I was messing around...

Also, you rock! Thanks for all the info, I really appreciate it. I'll do some more research into testing caps like this. I do have a decent power supply, I'll see what I can do.... I think the bank already has two 10k bleeder resistors installed, but I'll double check

[edavid]

When using my thermal camera, it looks like a ton of mosfets (they looks like mosfets, I haven't pulled the thing apart enough to triple check) get up to 200f pretty darn quick, about 15 seconds or so.

Are you sure that is a problem?  Have you asked Krell about it?  From what I've read about this amp, it could be normal.

I believe I narrowed the issue down to the capacitor board.

What made you decide this?  How could a capacitor failure cause the amp to go into protection mode?  It seems very unlikely to me.

It's funny you mentioned the connections desoldering themselves....I found this last night while I was messing around.

Found what?  It's not obvious from the (blurry) photo.

[ImpulsiveJames]

From the photo, it shows a blob of solder that dripped down and touched another lead. The background is blurry, but I think the PCB is in focus

I believe the capacitor board has the issue because when I put my DMM in continuity mode, all inputs into the channel modules beep as if there is a short - which I assume is an issue, unless I'm just not use to testing larger capacitors/capacitor banks and am doing something wrong

The amp seems to be fairly 'smart' and able to detect when there is an issue. Either this solder drip, a shorted cap, or the temperature (or a combination) seems to be telling it to stop outputting and then tripping the breaker on the back (It almost seems as if the breaker is electronically controlled by the amp, but I haven't verified that).

I will consider calling Krell, but my experience with techsupport is either you and I know more than they do, or they tell you to get lost because that info is proprietary.... I'm just super sour with techsupport currently.

[edavid]

I believe the capacitor board has the issue because when I put my DMM in continuity mode, all inputs into the channel modules beep as if there is a short - which I assume is an issue, unless I'm just not use to testing larger capacitors/capacitor banks and am doing something wrong

Yes, you are doing something wrong.  You can't test large capacitors that way.  They will read as a short until they charge up.

[cncjerry]

Krell will want you to send it in.  I wonder if they are using the best solder.  I think binding posts might be better for some of those high current connections, or maybe silver solder.  On the KSA160B amp I referenced, the solder had dripped down onto the frame and the stranded wire had melted a little near the connection.  That amp is massive.  It heats up the room.  I use it for the low end of a biamped setup.  Test the caps as I suggested using a current limited supply.  Then check the outputs.  When you find the bad output, replace the driver too and if it is complementary, replace that as well.  Then you might have to adjust the bias on that channel.

Have fun.

Jerry

[ImpulsiveJames]

Yes, you are doing something wrong.  You can't test large capacitors that way.  They will read as a short until they charge up.

Ok, that's exactly what I needed to know.... But I got ahead of myself and pulled them all out

Are these leaking, or is this flux residue? (one on top looks perfect, the other 3 look like electrolyte might have leaked) I would say about 18 out of 36 caps have this residue

I should have time today to charge a few and see how it goes