Nothing, reallly - it's morphed into the beast with three mosfets I suppose I could do something with a dozen or so J112s that would make noises like a headphone amp, but I decided to go my own way, as usual. I'm not real big on copying other people's circuits, even Papa's.... I wish I could say the same of other people I could think of.
I'm working on a headphone amp that's shaping up to be the beast with 13 jfets. All of them are J112, a jfet that's indescribably cheaper than the elusive 2SK/LSK170. I'm getting what looks like a sine wave our of the beast, but with a THD of about 3%, meaning I have a fair bit of sim-tweak work ahead of me, as I don't know right now what will be the magic bullet that will make the creation straighten up and fly right.
Looks like R15 has 13.5 volts across it, so there's plenty of headroom to replace it by a current source and get maybe 75x ( +37 dB) more open loop gain, to reduce distortion further. In the spirit of All-MOSFET all the time, a single depletion mode NMOS plus degeneration resistor would give you a nice current source. The (Microchip DN2530) might be one possibility. Or another feedback current source with 2 devices, like the output stage.
Just for grins, I tried using a perfect current source on top of the input depletion mode mosfet. I had to put a 33k resistor in parallel with it to keep things from getting crazy. However, the THD did go down to ~1.5 X 10^-2%. I have at home a small stock of an Infineon depletion mode mosfet that may be just the thing, along with that parallel resistor.
Since the circuit has a generous 10.5 volts of headroom available, you could build a cascode current source using two DZT5401's -- which are in SOT-223 packages -- and get even higher gain / lower distortion, plus simulation .MODEL accuracy. DigiKey has 19,000 of them on the shelf at $0.34 per piece (qty=10).
I ended up using my favorite depletion mode mosfet, the BSS159, on top of the input BSS159 and tweaking the value of the resistor in parallel with with the supplemental current source to 22k. I'm getting 1.49 X 10^-2% THD, and the odd order harmonics are progressively lower as compared to the 2nd. I like the BSS159 a bit better as a supplemental current source, as its capacitance is dramatically lower than the other depletion mode fet I was considering. It may run a little warm in that supplemental position, but I can live with that. I also have the model for that fet in PSpice, so running the sim was a breeze..
That may reveal the exact amount of open loop gain increase, which is responsible for reduced distortion as mentioned in post #433
I chipped away at the parallel J112 headphone amp design, and got the distortion down to ~0.2%, mostly 2nd harmonic. However. I'm not really looking forward to building the silly thing, as I would have to match a dozen J112s.... Jameco has the J112s for a pretty cheap price. I bought about 1500 of them when they were really cheap - they've since upped the price. Digi-Key's price is not too shabby if you're willing to buy 1000 pieces.
Last edited:
Another possibility is the TIS75 - I have 5000 pieces of those - a lucky buy from BG Micrro when they were still around. It's tough to beat a 10 cent jfet. They are original TI parts, and BG Micro had them hanging around in stock for ages. I finally asked them to quote 5k pieces, and they gave me a price I couldn't refuse....
Last edited: