[Pars]

Hi,

I am in the process of recasing a MINT I built about 4 months ago, and which has always worked fine. I wanted to convert it to wall power for use at work, and had picked up an Elpac WM113, +/- 12, +5V wallwart (was way cheap on Ebay).

I am putting it into a Hammond, and had te room, so I built a board with 4x1000uf Muse caps, and a pair of 1N4002 diodes on it. I removed the TLE on the amp board, the switch (jumpered it), and the crowbar diode. It is wired like:

I have a DPDT switch in the +/- 12V line before the diodes. I double checked everything, both ohming it out, and voltages to make certain that it was all wired up correctly (it is all connectorized so it is easy to check here and there).

The problem: the first time I powered it up, offset was as before, 0.000V per channel (limit of my one DMM). However, the second time I powered it up, the right channel went to +10.2 V, and has stayed there. Nothing appears to be getting hot, no magic smoke, etc. Left channel works fine (verified with phones jumpered in on one channel... I didn't have any around that I wanted to toast on the right channel).

So, it would appear that I smoked either the BUF634 or the right channel of the opamp (AD8620), or both? Any way to tell which before I order parts? Any glaring deficiencies in the additional caps circuitry, or the implementation in general?

Thanks,

Chris

 

[Excalibur]

If you have any other dual opamps with the same standard pin-outs as the AD8620, you can try one of those to see if the AD8620 is causing the problem.
Or, remove the Buf634's and measure the offset at the AD8620's. That will give you some information to go on.

 

[tangent]

That could be a dead chip, but it could also be a broken feedback loop. You might ohm out the signal path relative to ground and make sure everything along the right channel path matches that of the left. Do this with the power off.

If that fails, it's probably easier to solder tack a jumper across the suspect buffer than to remove it from the board. Don't do this with headphones, since I think it'll do strange things to the amp's gain. Just measure offset and a test tone going through the amp to see if you get reasonable performance or not.

 

[Pars]

Quote:

Originally Posted by tangent That could be a dead chip, but it could also be a broken feedback loop. You might ohm out the signal path relative to ground and make sure everything along the right channel path matches that of the left. Do this with the power off.

Thanks. I'll check this, though because the amp had worked flawlessly before I started in on this, I am not expecting to find anything. Running a signal (music) in and looking with a scope, I noticed differences on the opamp input in level (hardly any for bad channel). However, it appears from ohming it out that there is a bad mismatch in tracking on the particular Pana pot. Need to do some more checking.

Quote:

Originally Posted by tangent If that fails, it's probably easier to solder tack a jumper across the suspect buffer than to remove it from the board. Don't do this with headphones, since I think it'll do strange things to the amp's gain. Just measure offset and a test tone going through the amp to see if you get reasonable performance or not.

So I could just jumper from input to output across the buffer, and perhaps determine if the opamp itself is working? I thought this might be the case, but not being versed in multiloop design, didn't know what might occur. I had the BUF634 datasheet already, so will reference it.

Thanks for all the input, and will follow up with what I find. I had recalled reading stuff about opamps latching to one rail or the other, but had assumed this is a temporary situation (i.e., until powered off or something).

 

[Pars]

Well, as Tangent suggested, I checked out the feedback loop on the bad channel compared to the good channel, and they appear to be the same, ohming every resistor out to ground (both ends). I also didn't mention that I do have the CRDs in for biasing, and they both ohmed out the same also (don't know what the symptom for a blown CRD would be, guess if it shorted that would show up easily).

I then tack soldered a jumper across the buf634 on the bad channel (pins 3 and 6, although it would have been easier to just jumper R7 to the already jumpered R8 position).

No luck... offset still at + 10.2Vdc. So I guess this means it is a blown channel in the opamp. I am going to order one or two buf634s, and probably a couple of cheap opamps (OPA2134/OPA2604) to test with instead of putting another AD8620 at risk.

On the AD8620s, anyone hear differences between the A and B versions? The A version is ~$21 versus ~13 for the B version.

Thanks,

Chris

 

[tangent]

The B version of the chip has better DC specs. Since audio is AC and the A version's DC specs are adequate w.r.t. output offset and such, the differences don't matter.

 

[Pars]

Ooops, my bad... it is the B version on this one that is expensive.

Well, it was definitely the opamp, tho I have no idea why it went as everything was connected properly, etc. The only thing I can think of is I have jumper sockets on the protoboard that I have the additional filter caps mounted in so I can take them out of the circuit to see if I can hear a difference with them in or not. I originally fired the amp up with them disconnected, and everything was fine. I then jumpered them in, and retested the voltages before plugging the amp board back in. I may have not discharged them before plugging the amp board in... can't remember the exact sequence of events.

I have an OPA2134 in now, and have the amp put together... offset looks fine (1mv on 1 channel, 0-1mv on the other). I haven't listened to it yet (getting late for old farts like me... hehe). I will run the OPA2134 for awhile, and maybe try out the OPA2227 and OPA2604 before putting a new AD8620 back in if all is well.

Thanks for the help, and will post pics when it is done!

Chris

 

[tangent]

Quote:

I have no idea why it went as everything was connected properly, etc

You may have staticked the chip while you were fiddling with things.