>
my 634's were HOT!!! ... I can only touch them for 4 seconds.
> BUF634 datasheet says to use a resistor when feeding pin 1 (BW), we're feeding it -15VDC.
The
datasheet page 2 shows two extreme connections. BW pin to V-, 15mA-20mA.
You have 29.7V and possibly 20mA, 600 milliWatts. I remember when a plastic DIP-8 was barely good for 300mW. It appears this PDIP is 100 deg C/W, so you are near 60 deg C rise. At 25 deg C ambient, 85 deg C, or 185 in Pennsylvania degrees. This will brown meat, scald skin, and touch-time may well be 4 seconds.
I like < 50 deg C rise but I'm old-fashioned. The '634 -is- safe at this heat.
I was gonna suggest wiring for lower bandwidth.... but that darn 6141 is also wide-band. Both are inside a NFB loop. One or the other
must be slower.... if they are similar speeds, you have 2/3rds of a phase-shift oscillator and Murphy will find enough other "minor" phase-shifts to get a real ring going. Yes, I do think many people are using the TO-220.
In Meier's own words: "The LM6171 is a
diva. If everything fits she sounds gorgeous but if you don't treat her well she will provide you with a lot of
headaches. ... know exactly what you're doing. The LM6171 is very fast and therefore prone to oscillation..."
(emphasis mine)
Compare your implementation with Meier's:
http://www.meier-audio.homepage.t-online.de/bilder/img_0006small480.jpg
Yes, you can make perfboard
almost as neat and compact as his PCB. Although I do suspect that the 6171 on perfboard will be
very difficult to tame, may need Bob Pease's experience and intuition.
Look, I'm re-designing this thing, which isn't my job, and it's late.
>
I cleaned up the many 15v leads
OK, what about signal?
What about ground-path? (No, I am NOT going to side-side scroll over-large fuzzy images to trace it out.)
>
killed me laying this out was feeding the IC power rails
On perfboard: #24 solid insulated wire. When you get to a chip pin, peel 1/4" of insulation, wrap or hook the wire on the socket pin-stub, solder, and run on to the next chip. Between chips, keep your + and - wires together, perhaps twisted. Dot with hot-glue to keep from flapping. There's no reason to have more than one power-entry for this whole board. There's no reason to whine for PCBs.... my other HP amp is half perfboard and half P2P (HOT transistors and 10W resistors), I have a several-chip mike mixer on perfboard.
Notice I laid my '072s in a line. The power rails run down the middle, between the pins. That leaves the "outside" of the chips for the more complicated networks.
>
Sounds like a ground loop to me
How can you tell, from "sound", if a hum is caused by a loop?
And there isn't any trick to grounding this, 'cuz it is largely differential. Your power supply common MUST go to headphone jack common: that's where the load current happens. Then to '634 rail-caps, the other side of that output current. Then you have the volume pot ground, the diff-amp 10K reference, and the 3K3 resistor reference: these should come together at one point (which is why my pot and input chip are just two inches apart). The XLR jack ground and 680pFd ground will probably go to case... these are NOT signal grounds, they are garbage grounds. And given the diff-input, I'd tie signal ground to case at the headphone jack (partly cuz those plastic jacks break, and my metal jacks force grounding).
>
I was feeding one of the 15M resistors ...with +15VDC and not -15VDC.
FORGET those dang 15 meggers. I think they will do just as much "good" to plus, minus, or on your dog's collar... they do NOT make any big difference to anything. If they make a minuscule difference... well, you have bigger problems still. I'm not saying take them out... just forget them.
Shame you have only 2 Watts. There is a much simpler good-sounding (and cheap!) headphone amp, but for studio work it wants much more than 77mA.