My PIMETA2 is broke -- please help!
Not meant to excuse myself of SNAFUs or responsibility -- nor do I claim any great DIY skill ;) … but …(along with several other “DIY” amps and mods), I’ve successfully built both a PPA2 and this now-“broke” Pimeta2 -- they all have worked pretty much from the get go… up to and including several interim tweaks and op-amp rolls. And, for well over a year, my Pimeta2 has survived near-everyday use (including gazillions of the ever-inconvenient 9v NiMH changes). Again, all w/o incident. Until now … my “indestructible” Pimeta2 has crashed, big time, and I have not been able to troubleshoot it for a week -- literally 20+ hrs of hair-pulling and still no luck.
How it all started (?? best guess!) and where stuff stands …
After rolling in a new op-amp, I turned the Pimeta2 on and heard a loud pop over my IEMs. Cause: I did not seat the opamp properly into the 8-DIP plug-in socket -- so I seated it correctly and turned on the amp. Sounds fine again but now LED2 is not lit (LED2 is one of two the class-A-biasing LEDs from Tangent’s schematic; I’ve always used both LED1 and LED2, and they are mounted on the front panel for easily visibility). Troubleshooting, I found that as I was inserting the opamp, I had broken one of the pins on Q2 (2N3904) [I use a SOIC-to-DIP BrownDog adptr and that’s hard to squeeze in w/o bending some of nearby components, like Q2.] Anyway, I replaced Q2, and the LED was restored … but now, I heard a loud pwr-on thump in the L ch only (which is now also dead); R-ch sounds fine and no pwr-on thump).
After a bit of checking, I noticed that the opamp (AD8620 or 8066) was getting VERY warm after about 5 sec. I moved the opamp to a breadboard and attempted “piecemeal” troubleshooting. The opamp draws 54mA (!!) when the V_out1 (Pin 1 for L-channel on dual-ch opamps) is normally connected to the L-ch ckt in front of it -- this includes the feedback loop, PREBUF, 100-ohm R, and the LMH6321 buffer.
If I disconnect the opamp’s V_out1 pin (L-ch), it draws a mere 8mA .If I swap output pins (dual opamp's pin 7 now feeds the L-ch ckt in front, I get the same 54mA draw -- this will make the opamp v. warm in a few secs). Feeding ONLY the R-ch ckt (L-ch open in front of opamp) with either V_out pin results in normal 8mA draw.
Note: opamp V_out (1 or 7, i.e., whichever is feeding L-ch ckt in front of it) has V = 2.8vdc. The one that feeds R-ch is normal (few mV)
I haven’t checked the LM6321 buffer’s current draw -- FWIW, it runs cool. And they do have built-in "OVERVOLTAGE PROTECTION" per National's datasheet. Just in case, I also replaced it (L-ch only). No diff (for all I know, this buffer could have blown again by whatever else is messing up the L-ch). I also replaced Q3L (2N3904, the L-ch op-amp class-A bias trans).
I checked the output of the buffer (the ‘phones O/P jack) -- and L-ch turns on with (and remains at) a whopping 2.6 volts DC offset (no wonder I heard that pop-thump)!! R-ch is normal (3-4mV). All this with NO audio/music signal at input -- i.e., running idle. Soundwise, R-ch sounds fine, while L-ch is "dead"/broke/whatever.
Shorts in the L-ch ckt ahead of the opamp output? None I could find with a decent DMM (Fluke 87). Alas, my o’scope is itself dead so I can’t do detailed troubleshooting.
Power suppl. -- which is TLE2426-based with 9v batt (default) -- seems to be fine -- so, all powered opamps/buffers are getting proper +/- 1/2 the supply volts.
- I've almost never use passive "protection" (on any amp ;) -- for the Pimeta2, that means C1 is bypassed (jumpered). BUT ... I do use output-current-limiting resistors, R11 (3k-ohm), for all buffers.
- Ground channel seems to be fine (this is a deduction based on the fact that R-ch seems to be good).
With the L/R-ch opamp(s) REMOVED, but with the buffers connected (in normal circuit), why do I get large DC offset at output for BOTH channels?
L-ch: 2.6V (as noted above); R-ch: 160mV.
If there is no input, the buffers should not output any voltage (other than a few mV idle, of course). Or am I missing something here?
My brain is mush and my head is bald from all the hair-pulling! Please help -- thanks!
[BTW: after that initial thump, which my IE-8s survived, I switched to a pair of junk cans -- they seem to be able to handle the 2.5V DC offset. This is how I’m able to test for music/signal and distortion]
EDIT: Problem solved (or isolated) somewhat ... but WHY?
Okay, I noted above that I replaced Q3L (the L-ch class-A bias trans.). Honest to dog I replaced it with another 2N3904 -- and I did. Yet when I removed it (just a few mins ago), all returned to behaving "normal" (at least WRT basic metrics: DC o/p offset is normal, quiet pwr-on/off, no pops or crackly volume, etc) ... except that topologically I have a 2N3904 Q3R for R-ch, and the rest of the class-A transistors (and all the related LEDs and resistors) all in situ (i.e., nothing else removed, bypassed or jumpered). WTF just happened? WTF is STILL happening: I installed another NEW 2N3904 and the problem returns again! (dunno if it's same batch as orig 2N3904 -- the one that worked in the Pimeta2 orig. built 2 yrs ago -- the one that I replaced a few days ago). I tried another putatively higher-quality NPN (Toshiba 2SC2240, and accounted for its pin diffs) ... but same problem when installed in the Q3L footprint. Yes ... I checked all traces -- with Q3L removed -- for shorts or opens. All's normal. At least as far as I can test. I haven't tweaked the bias trimpot because (I assume) Q3R is functioning normally???
Edited by alphaman - 3/8/12 at 9:39am