1 buffer for 2 channels or not

Hello.
I need to buffer 1 signal to 2 channels.
I need help to determine if I can use a single buffer opamp to send the buffered signal to 2 different channels ?
What is the optimal setup for lower noise ?
Thank you.
An externally hosted image should be here but it was not working when we last tested it.

EDIT: I will not use the lt1018 it just for illustration purpose.
 
Last edited:
Are the right hand LT1018's supposed to represent the various loads to be driven?
Are the loads in separate enclosures, with connectors and cables run separately?

Normally all you need to drive multiple loads is a low impedance source, but driving a cable's capacitance can be quite demanding - driving several cables from one opamp is perhaps a bad idea due to the loading.

Noise is not an issue here, everything is the same from a noise perspective really.

100nF is perhaps too small a coupling cap for driving an external line signal, as your output impedance is then very frequency-dependent, perhaps that's what you mean by noise though? (ie susceptibility to cable interference)

There's no need to ac-couple an output that's already purely ac, unless you expect failure of the driver for some reason. It makes sense that the load has the coupling capacitor since it knows its own input impedance and thus can choose a suitable capacitor value, and input impedances are generally large.


There's another reason not to have a high impedance on the output of the buffer, which is input-current non-linearity in the load - normally this is irrelevant when driven by a low impedance, but with high source impedances any input current non-linearity in the load is multiplied by the source impedance into a non-linear signal voltage...
 
Hi thank you.
The 2 channels are in the same enclose but are far from the buffer (around 20cm) each of the channels are wired with a shielded coaxial cable of 180pf/meter.
100nF is perhaps too small a coupling cap for driving an external line signal, as your output impedance is then very frequency-dependent, perhaps that's what you mean by noise though? (ie susceptibility to cable interference)
Yes it's not clear. Sorry.
What I meant by noise, is if the bias current of the CH1/CH2 opamp are both in one resistor, and maybe it will create 2 time the amount of noise than with 2 separate buffer/resistor to ground ?

Thank you it helped me a lot already.
 
EDIT: OK, so you can ignore part of this, as I didn't see the previous post before replying.

OP, please explain what you need this for - what kind of signal levels, cable lengths and input impedances are we looking at? Normally you should have little difficulty driving two 50-100 kOhm inputs with a moderate length of cable using any opamp with half-decent output driving capabilities. A 5532 or similar with 100-220 ohm of series resistance for each output should be plenty.

Not sure why noise would be of particular concern here. So let's assume the most critical case I can think of, trying to drive two power amps from one preamp. Even when using about the noisiest part commonly considered an audio opamp like the old AD712 at 18 nV/√(Hz), a unity-gain follower gives me only 2.6 µV of output noise. You will be very hard-pressed to find a preamp output with less than 3.5..4 µV all by itself. Even in a worst-case scenario, that's 1.9 dB more noise. Use a more sensible part at 5..10 nV/√(Hz), and that figure shrinks to 0.2 to 0.7 dB. If you can't accept that, maybe you have too much gain in preamp + power amps + speakers altogether and should consider inserting a voltage divider ahead of the buffer.

EDIT:
The 2 channels are in the same enclose but are far from the buffer (around 20cm) each of the channels are wired with a shielded coaxial cable of 180pf/meter.
A mere 20 cm? Piece of cake. I don't see any problem with that at all. 100 ohm of output R just in case, and you should be set.

Your greater worry may be internal ground loops.
What I meant by noise, is if the bias current of the CH1/CH2 opamp are both in one resistor, and maybe it will create 2 time the amount of noise than with 2 separate buffer/resistor to ground ?
Err, no. The bias resistor is in parallel with the (in general) much smaller source impedance, which is likely to ensure that current noise is a non-issue. Not to mention that whatever you are going to feed the opamp with is likely to output a whole lot of noise by itself. I mean, you're probably not trying to buffer an MM phono cartridge or condenser microphone capsule, are you?

Generally speaking, paralleling two opamps increases current noise by 3 dB and decreases voltage noise by the same 3 dB. Here, current noise is likely to be irrelevant either way, while voltage noise may or may not be.
 
Last edited:
Hi thank you.
The 2 channels are in the same enclose but are far from the buffer (around 20cm) each of the channels are wired with a shielded coaxial cable of 180pf/meter.
Yes it's not clear. Sorry.
In audio terms that's not far at all, single opamp is fine.

What I meant by noise, is if the bias current of the CH1/CH2 opamp are both in one resistor, and maybe it will create 2 time the amount of noise than with 2 separate buffer/resistor to ground ?
The noise is determined by the source impedance, not the bias resistor, as that is shorted out completely at AC by the very low impedance opamp output. The formula for Johnson noise voltage of an individual resistor applies to an open-circuit condition.


In any network of (perfect) resistors (at the same temperature) the noise voltage between any two nodes depends on the overall resistance between those nodes. An ideal voltage source acts as a zero resistance from a noise perspective, and an opamp output is close to being an ideal voltage source (but carrying a signal with its own existing noise already)


Thus the 100k resistor would add no voltage noise in this situation as its to ground, in parallel with the opamp output's voltage source. If it were in series with the opamp output it would contribute a large amount of noise.

Thank you it helped me a lot already.
 
Thank you both of you. I understand better now.
Last question, where should I ground the 0V of the +-15V.
I grounded the input connector to the chassis, and ran a bus wire to the first opamp ground.

What is the "standard common" best place to ground the 0V ?

If I ground it at the input jack will I have psu noise/imbalance in my signal ground ?
 
Is power supply ground already tied to signal ground elsewhere?

If so, I would run power ground (for any bypass caps etc.) back to the power supply alongside the +/-15 V, ideally twisting all three together. Signal ground should connect at some sensible place not too far away from where you need your connections. If you can afford to keep them separate, do it.

Giving good concrete advice is not easy without a diagram of all the existing ground (and perhaps power) connections in the unit. Just a bunch of rectangles with lines denoting connections either on the boards or between them, and any connections to chassis (maybe at the power supply or input jacks). Do you think you could draw one?
 
Thank you.

Let me work on drawing it in ltspice. But it's a frankenstein, a 50W 2 6L6 tube power section with a 2 channels opamp preamplifier.

Is a schematic better ?
I was too lazy to draw one for me, it's the right time to make one I guess.
I did not draw a schematic because it's my hands on test amplifier, always changing.