If that source ends in a 10kohm log pot and you prototype the gain structure
to cranck it conservatively at 12 o'clock , you have 9 kohm input resistancea and that I know it needs at least 5 nF shunt to ground to keep it stable...1kohm is acceptable as input resistance but 1kohm log pots are hard to find.
When I see a post about a "headphone amplifier", I read "something to tinker with", which is probably true of most things on DIYA. If you just want a headphone amplifier, you can buy a complete product on Amazon for $25. So, I recommend you start with spice and try different things until you fall in love with one, and then build it. Options include:
1. "audio" op-amps have a limited output current, so you have to buffer them, but then the buffer is not current limited unless you complicate it, or just put a ~100 Ohm resistor in series so it's impossible to "short" the buffer. You can use a class-A buffer but then it wastes a lot of power which means it's not suitable for battery operation; it's not portable.
2. TI and others sell headphone amplifier chips. But this kind of spoils the whole DIY purpose in a couple ways. a) It's someone else's design and b) a custom chip is a support hazard; may disappear next year. https://www.ti.com/audio-ic/amplifiers/headphone-amplifiers/products.html
3. I suspect one of the many classic small stereo amplifier chips would make a fine headphone amplifier. Have a look at LM386, LM1877, etc. These are not usually super hi-fi, but you can fiddle with bias and feedback to "improve" them and still have plenty of output for headphones.
4. A discrete amplifier can easily be exactly what you want it to be, ie just the right power, efficiency, and fidelity. And you could make a great headphone amplifier with a box of assorted transistors. LV posted a circuit that ran on a single Lion battery, ie very portable. A headphone amplifier that requires a big heat sink is not very impressive, but if that's what you like, go for it.
1. "audio" op-amps have a limited output current, so you have to buffer them, but then the buffer is not current limited unless you complicate it, or just put a ~100 Ohm resistor in series so it's impossible to "short" the buffer. You can use a class-A buffer but then it wastes a lot of power which means it's not suitable for battery operation; it's not portable.
2. TI and others sell headphone amplifier chips. But this kind of spoils the whole DIY purpose in a couple ways. a) It's someone else's design and b) a custom chip is a support hazard; may disappear next year. https://www.ti.com/audio-ic/amplifiers/headphone-amplifiers/products.html
3. I suspect one of the many classic small stereo amplifier chips would make a fine headphone amplifier. Have a look at LM386, LM1877, etc. These are not usually super hi-fi, but you can fiddle with bias and feedback to "improve" them and still have plenty of output for headphones.
4. A discrete amplifier can easily be exactly what you want it to be, ie just the right power, efficiency, and fidelity. And you could make a great headphone amplifier with a box of assorted transistors. LV posted a circuit that ran on a single Lion battery, ie very portable. A headphone amplifier that requires a big heat sink is not very impressive, but if that's what you like, go for it.
@dreamth Good point, but I have had no troubles with 10K or 20K log pots. Remember, the maximum output impedance of a potentiometer is 1/4 its nominal value. Meaning a 10K pot would have at most a 2500 ohm output impedance. But a log pot at 12:00 (half way up) is likely to have an output impedance of about 900 ohms. Given a typical 100 source driving it, that's 1K ohm driving whatever is downstream (the amp). A 20K or 50K pot might indeed cause some problems to watch out for. 
If an amplifier is unstable with a high resistance POT on the input, it probably needs a ~100pF cap on the input, both for stability and RF filtering.
I have considered the possibility of "bridging" a headphone amplifier. But when I simulated one in LTC, I discovered the issue is much more complicated than expected.
I realized that the shared ground return of the two channels would have to be at 50% gain so that a mono signal on both channels would be 50%+50%. But turns out that potentially, the two channels may be 180 deg out of phase, so the inverted output would sum to zero, and given that each channel must sum the invert of the other channel on the common, that sum becomes 150% (clipping). So statistically, bridging could provide +3.5dB, but potentially be a net loss. So it's mostly a bad idea and a waste of time.
I realized that the shared ground return of the two channels would have to be at 50% gain so that a mono signal on both channels would be 50%+50%. But turns out that potentially, the two channels may be 180 deg out of phase, so the inverted output would sum to zero, and given that each channel must sum the invert of the other channel on the common, that sum becomes 150% (clipping). So statistically, bridging could provide +3.5dB, but potentially be a net loss. So it's mostly a bad idea and a waste of time.
Attachments
@Wik
The NJM4562 integrated circuit is a high-gain, wide-bandwidth, low noise, dual operational amplifier capable of driving 20V peak-to-peak into 600Ω loads. H/P specs can go as low as 16Ω. Emphasis is on driving low impedance load capability
Read-
https://nwavguy.blogspot.com/2011/08/op-amp-measurements.html
"*NJM4562 – How about a $3 LM4562 (see above) for $1? That’s pretty much what you get with the NJM4562. My measurements show the two to be very similar with the NJM version even needing less quiescent current. It has slightly more noise than the 2068 above but can better drive low impedance loads, with lower high frequency distortion, at higher gains. For gains < 4X into “easy” (non-reactive) loads of 2K or higher, the NJM2068 is the better part. If you’d rather have the Audi A3 over the very similar but much cheaper Volkswagen GTI, get the National part, otherwise save some money with the NJM4562.
HIGH CURRENT: For driving low impedance and/or significantly reactive loads, there are not many choices in dual DIP8 packages that work comfortably at +/- 12 Volts (24 volts total). The short list:
The NJM4562 integrated circuit is a high-gain, wide-bandwidth, low noise, dual operational amplifier capable of driving 20V peak-to-peak into 600Ω loads. H/P specs can go as low as 16Ω. Emphasis is on driving low impedance load capability
Read-
https://nwavguy.blogspot.com/2011/08/op-amp-measurements.html
"*NJM4562 – How about a $3 LM4562 (see above) for $1? That’s pretty much what you get with the NJM4562. My measurements show the two to be very similar with the NJM version even needing less quiescent current. It has slightly more noise than the 2068 above but can better drive low impedance loads, with lower high frequency distortion, at higher gains. For gains < 4X into “easy” (non-reactive) loads of 2K or higher, the NJM2068 is the better part. If you’d rather have the Audi A3 over the very similar but much cheaper Volkswagen GTI, get the National part, otherwise save some money with the NJM4562.
HIGH CURRENT: For driving low impedance and/or significantly reactive loads, there are not many choices in dual DIP8 packages that work comfortably at +/- 12 Volts (24 volts total). The short list:
- *NJM4556 – If you need lots of drive capability the NJM4556 can beat every dual DIP8 op amp I know of.This part is discussed more in the O2 Design Process article."
Last edited:
