Hey everybody -
Sorry I haven't been able to keep up with this thread and answer everybody's questions. I'm busy as heck with the "day job", plus various commercial headphone projects on the side, and have been travelling a lot, and will be travelling more...
Feel free to send me an email or PM if you have a specific question. I try to answer reasonably quickly, though it can take a few days.
A few comments:
The switching power supply has a hard time starting up into the cold filaments, which have a pretty low resistance. So it "hiccups" - turns on and then current limits, and turns off - several times at about a 1Hz rate, until the heaters are warm enough that they don't draw too much current for the supply. You can avoid this by increasing the second filter cap (C6 on my schematic) from 150uF to something like 470uF. Bigger here will not hurt anything - 1000uF or more is fine, no worries. What this does is make the bias on the FETs rise more slowly, causing the filaments to heat gradually, and the power supply is happier.
As for the output cap size - 150uF is OK for most headphones (even 32 ohms, IMO), 470uF is OK as well. Why did I use 150uF? 'cause I had them
I think there's no benefit in going as high as 1000uF. Don't assume that having flat frequency response to 20Hz is necessarily a
good thing; most headphones only produce distortion at that low of a frequency, and if the amp reproduces it you tend to increase IMD generated in the headphones. Just for kicks, try listening to test tones at these frequencies - what you hear is usually either nasty, or non-existant.
I usually like to set a -3dB point of 20Hz - 50Hz for the "chosen" headphones. YMMV, IMHO, and lots of other TLA's and FLA's apply here. Another downside to going big (no, I'm not talking about "male enhancement") is that you will get a bigger thump (get your mind out of the gutter) when you turn the power on. You can mitigate this somewhat by lowering the value of the output load resistors (R6 and R12), but you will be stealing current that could otherwise be delivered to the headphones.
I still haven't had time to try the Jameco linear supply, but it should not have the hiccup issue. Has anybody used it? Is there any hum? Many of the linear wall warts have pretty small filter caps. You can always add more capacitance (C1 in the schematic) if there is hum, even 10,000uF with a linear supply will probbaly work (but most switchers won't start up into a big capacitance like that).
You can't really substitute a 6J6 tube for the 19J6, since it requires 600mA to heat it up. I don't think 19J6's are too hard to find - there are lots of tube dealers out there. The beauty of this particular tube is that it has high enough Gm that you can get reasonable gain and distortion from it on 48V B+, and it's heater voltage and current just happen to make a good load for a class-A MOSFET source follower to drive 32 ohms.
If you want to try other tubes you need to find something with a filament voltage of (ideally) 10-25V and a current requirement of 100-200mA. 100mA into 32 ohm headphones is 3.2V peak, which is plenty loud. FWIW, for the engineering-inclined, you could design a similar amp using 12V car radio tubes (12AE6A, etc.) that need 12-14V at 150mA. You would bias the gates of the FETs at about 16-17V (Vgs of IRF510 is around 4V at this current). I picked the 19J6 partly because it is cheap and useless
I wouldn't be obsessive about the coupling caps - sure, polypropylene is a tad better than PE to most people, but any film cap will do. I'd be more concerned about making sure you have decent electrolytics. Black gates are not necessary - good low ESR caps (Nichicon UHE's, UPL's, or similar low-Z parts from Panasonic of Chemicon) are fine.
Did I read that right, that somebody gets to build this instead of taking an AP phyics final? That's waaaaay cool.
Well, that's all that comes to mind for now... have fun!
Pete
Oh the joys of Newark. At least I got free shipping.
