Doesn't appear there is a whole lot of interest in this little project
, but i am slowly making headway
.
I ordered the wrong output resistors, they were the correct value, but the lead size on the resistors are too big to fit in the holes on the board
so I had to solder them on top of the holes, a pain but only temporary till i get the correct ones.
Once that was done, I finished putting it together and tried to set the buffer bias, this turned into my first problem to debug as the right channel was basically shorting across the 2.2 ohm output resistors. I was able to find a bad solder joint on that channels biasing circuit and was able to sneak the soldering iron in and fix it.
Next issue that came up was really bad op-amp oscillation on power up. If I touch the board on the right spot(s) they will settle down and the DC offset voltage at the output is very low (~6-12mv), and the amp sounds very very good, at least until the op-amp(s) start oscillating again and I have to fiddle with the board to get it to stop again.
So I took a look at the board layout, the schems, the op-amp data sheet, and the parts recommendation guide and came up with a few reasons this could be happening;
I don't have an ideal board layout for the op-amps, its not terrible, but not ideal. to make the board small enough I couldn't get the op-amp input resistors, or the decoupling caps (C5) as close as I would have liked to the requisite pins on the op-amp. They still quite close, and using the original PPAv2 board as a guide, total trace length is not that much different. However, I should have paid more attention to the filled V-ground plain on the board, it may be too close to the input resistors on the op-amp side. I also note that R8 and C5 are designed to form a low pass filter for the op-amp power, I got tunnel vision on this point, I read the parts selection guide where it stated that .1uf was the smallest to do any good, so I designed it for that sized cap because I could get a wima polypropylene to fit nicely, but I glazed over the fact that doing so would raise the frequency of the intended filter by a factor of 10 if I left R8 at 10 ohms. So now, if I want to fix it I have to either go with a lesser quality polyester or (egad) ceramic cap, or replace R8 with a 100 ohm resistor. Not sure I like either of these, but i think changing the cap might be the lesser of the two evils.
My choice of op-amps was not ideal either (tunnel vision again), I wanted to make this amp as close to my original PPAv2 as possible, so i spent $$$ on OPA627/37 units, however, these amps minimum voltage level is around +/-4.5v, so absolute minimum voltage at the power rails is 9v. With the rail isolation circuit it seems to be that minimum battery voltage is closer to 9.5 volts. Nominal voltage on a 3 cell li-po pack is 11.1, however minimum voltage is anywhere from ~7.5 to 9 volts, so I cannot take full advantage of this battery with these op-amps. The oscillation problem gets worse as voltage drops, it is still there even with a fully charged (12.6-12.9v) battery pack, but the amp gets more and more unstable as the pack drains.
So, I think changing the op-amps is the first order of business, I will try the AD8610 to start with as that is the recommended amp, and I think i might even have some. I also want to try some others that are more forgiving of lower voltages, but that will be down the road a bit. I am hoping that the oscillation issues go away with a better suited op-amp, and as it seems to be more stable at higher voltages I'm hoping this will be the case. After seeing it assembled, There appears to be room enough on the backside of the op-amp board to add some chip style decoupling caps, not sure this would help but worth a try on a future board spin (if there is one).
This morning I started a road trip for work, so I thought I would do a long term test. As per normal, as soon as I turned it on it was oscillating like mad, If I touch the bottom of the board at the input and output connectors (that seems to be the best place) it stops and I started listening. I was driving all day so I was able to get a good idea for how it sounded, thankfully the op-amp oscillation only popped up a hand-full of times and was usually easily fixed buy just touching the board near the op-amp connection or at the input/output connectors. I was using my Iphone 5 for the input, and my M50's on the output side of things.
The good news is it sounded great, it seemed to have all the things that I loved about my full sized PPA, great power and separation, extremely great detail everywhere, and oh yea, lots of power
.
I was also pleased to note that even though i had the buffers biased to 20ma each, the board never got very warm. I have been worried since i started this project that with such a small layout and everything being so close, that heat would be a big issue, however now I am fairly certain that it wont be.
I was also able to get 6 hours of play time through it, although the last 45 min or so I had to trim the volume down a tad (just a tad) to keep the op-amps happy, then it started clipping and oscillating and generally being a PITA, so I shut it off and measured 11.01 volts on the battery pack, hmmm, that should still be enough,....but guess not.
I am open to ideas or suggestions, I am still quite a bit of an amateur at these things, if anyone has any thoughts other than mine on these op-amps. I will be happy to post a jpeg of the circuit board layout, as well as provide the Express PCB files if anyone is interested in looking at them. The schematic is basically identical to the PPAv2, the only thing I did was add connectors to mount the op-amps and most of the buffer circuit to daughter cards, and I changed pin outs where needed to use SOT 23 components.
The resistors values are all the same as the original PPAv2 schem, I used 1000uf for the power rail caps (10x100), .1uf and .01uf wima PP for C2, 100uf (2x50) for C4 and .1uf for C5.
Oh, and for Q3 i used 512-PN4391 (50ma), my thought was these would be good for just about any op-amp, but i might have been better off with 4392's in that spot.
joe