You don't need to. The PC power supply has +12, -12v, and ground. The molex has +12, +5 and ground. You can tap the -12v from the board connector.
I wouldn't use that -12V, since it can supply a couple of 10's of milliamps only and is left out completely in modern PCs.
I've abandoned the ground channel for two reasons:
1) Op amps don't seem to share the load, as in the ground channel doesn't contribute to output power.
2) Power supply noise was a ground loop problem after all.
I have a couple of questions:
1) Adding a small resistor between PSU and signal ground solved ground loop noise. Why?
2) I'm pretty sure op amps don't like a low load impedance such as my Sennheiser HD201 (24 Ω @1 KHz). Anythign I can do to improve this circuit? Except paralleling more op amps, that is.
3) Output coupling capacitor is so large (-3 dB @ 2 Hz). Do you think the op amp sees it as a capacitive load and should I be concerned about it?
1) Op amps don't seem to share the load, as in the ground channel doesn't contribute to output power.
2) Power supply noise was a ground loop problem after all.
I have a couple of questions:
1) Adding a small resistor between PSU and signal ground solved ground loop noise. Why?
2) I'm pretty sure op amps don't like a low load impedance such as my Sennheiser HD201 (24 Ω @1 KHz). Anythign I can do to improve this circuit? Except paralleling more op amps, that is.
3) Output coupling capacitor is so large (-3 dB @ 2 Hz). Do you think the op amp sees it as a capacitive load and should I be concerned about it?
An externally hosted image should be here but it was not working when we last tested it.
you really should just use higher output current op amps - the O2 project's NJM4556 are fine if you want to avoid surface mount parts - TL072 have particularly poor output stages for such low Z load
the 47 Ohm current sharing R are too big - single digit Ohm R drop less supply V, reduce problems with C load from the cable
and why inveritng?
the 47 Ohm current sharing R are too big - single digit Ohm R drop less supply V, reduce problems with C load from the cable
and why inveritng?
inverting gain loads the source more heavily which can cause distortion, reqiuires bigger DC blocking C for the same frequency response
too high a value of feedback R can allow another destablizing effect to become a isssue - the input parasitic C of the op amp
noise V and in the case of bjt input bias current errors are bigger with higher values of feedback R
too high a value of feedback R can allow another destablizing effect to become a isssue - the input parasitic C of the op amp
noise V and in the case of bjt input bias current errors are bigger with higher values of feedback R
Would you consider this an improvement over the last circuit? How low should I go for feedback resistor? Lower resistors implies larger capacitors, which is the reason I aimed high initially.
Also, if I can get my hands on a pair of NJM4558 or OPA2134 should I prefer them to NE5532?
Also, if I can get my hands on a pair of NJM4558 or OPA2134 should I prefer them to NE5532?
An externally hosted image should be here but it was not working when we last tested it.
I think virtual grounds are perfectly OK to work with, but you don't want to sink any serious currents into it. This you would do in the headphone application shown a couple of posts before.
There is an elegant way out, though, which would also increase headroom when driving from a single low voltage supply: of the four output opamps, connect two in positive and two in inverting mode, and bridge the outputs. No drive currents would flow back into the virtual ground, and presto.
Disadvantage is that you halve the available drive current, so you might wish to have some more opamps in the output if you worry about not having enough.
vac
There is an elegant way out, though, which would also increase headroom when driving from a single low voltage supply: of the four output opamps, connect two in positive and two in inverting mode, and bridge the outputs. No drive currents would flow back into the virtual ground, and presto.
Disadvantage is that you halve the available drive current, so you might wish to have some more opamps in the output if you worry about not having enough.
vac
I just wanted to add my support for Conrad's position, on both counts.
A virtual ground is a approximation of a real ground, that approximation is as good as the performance of the amplifier vs. the work it has to do. The impedance at high frequency may be an issue, stability may be an issue, noise may be an issue, and it is for sure more complex and less reliable than the real thing.
It's a valid trick for creating a split power supply from a single supply, and it is useful for battery powered or low budget headphone amp applications where true split supply is not feasible.
However, for hifi application, just use a transformer with two secondary windings. Problem solved: real ground obtained (circuit common, COM, as I prefer to call it.) with no mucking around.
Second point: parallel op amps will lower the noise and increase the output current, but will not get around the fact that you are using op amps to drive headphones: ultra high feedback (in unity gain config), very lean output bias current - i.e. mostly class B operation. A discrete output stage (single ended, or heavily class-A biased push-pull, or any of the power-follower circuits) is hardly much more effort to build, and can be properly tailored to drive low impedance, inductive/capacitive loads.
/rjm
Last edited:
Tapped it. Needs extra filtering, but will do the trick while the project is still on the proto-board. Abandoned biasing altogether. Will build a proper PSU when it's done.You don't need to. The PC power supply has +12, -12v, and ground. The molex has +12, +5 and ground. You can tap the -12v from the board connector.
This is probably the best advice. Learned a lot. Adopted the basic idea. Enough said.
Here is a cheap TDA2030 headphone amp with single supply rail from amlinear regulator. I just tried out and it's very decent sounding.
http://www.diyaudio.com/forums/chip-amps/310816-tda2030-coffee-money-headphone-amp.html#post5150348
Works quite nicely.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.