2
\$\begingroup\$

I am working on an audio project that needs to drive 16 balanced xlr cable runs with stereo differential audio. For this purpose I am using the following op-amp circuit to produce a low impedance differential output:

enter image description here

In this circuit R20 functions as a current limiting resistor, and effectively sets the output impedance at 75 ohms, while C32 and R24 decouple the audio signal from the DC bias.

In my application I am generating a stereo test signal and sending it to all available outputs (alternating left and right for even and odd), so I just need a way to safely buffer the output a bunch of times in such a way that the outputs can't interact with one another.

My question is whether I can safely drive multiple lines off the same op-amp by simply attaching multiples of the R20, C23, R24 current-limiting + decoupling circuit to the same op-amp outputs. Theoretically with an ideal op-amp circuit this should work fine, since the op-amp will hold the voltage constant at it's output, and the output resistors will limit current preventing voltage drop, but I am wondering if this can lead to problems with noise, or other interaction between the output lines when put into physical practice. Furthermore, seeing as I'm planning to drive up to 16 lines, is there a way I can calculate a max channel cutoff for when I should break this method out to multiple op-amps?

This is meant to be a piece of test equipment, so having it be reliably low noise is important, but being able to have a single op-amp package rather than 16 of them would be a big reduction in cost and board space, so it's worth exploring.

Thanks!

\$\endgroup\$
4
  • \$\begingroup\$ I assume load is between 10k and 10M so no limit. But STP cable with source shld.gnd only if going far away to a different and. LF response limited by Load R * C =T \$\endgroup\$ May 26 at 22:02
  • \$\begingroup\$ How long is the cable? what is the effective load impedance with everything in parallel ? \$\endgroup\$
    – Autistic
    May 27 at 3:49
  • \$\begingroup\$ Note that you must shift the mic signal to 4.5 V DC point before giving it to the opamp. \$\endgroup\$
    – tobalt
    May 27 at 3:59
  • \$\begingroup\$ @tobalt it's coming from another amplifier circuit that is using the same bias, so it is already shifted \$\endgroup\$ May 27 at 13:16
0
\$\begingroup\$

You need to look at several things:

  1. power supply voltage determines the peak voltage at the output
  2. the input impedances of the devices you are going into.

The 5532 is an excellent audio op amp, and will drive (from the datasheet 600R at +/-15V. But sometimes people run it a bit higher, like +/-17V to get more headroom. The real determining factor is the maximum current out of the op amp.

If your destination devices have an input impedance of say 10k, you are right at the edge - in fact a bit over it as your + output already drives one 10k load (the other stage). 10k/17 is 588R. (This doesn't allow for the 75R source resistor but it shows things are getting a bit tight - also don't forget that the lower the paralleled destinations look, the more signal you lose across that 75R rsistor, which is needed for stability, given possible cable capacitances.)

If it were me, I would provide two or four outputs, and divide the destinations between them. Sixteen is just a bit on the high side.

EDIT : I just saw you are running at +9V, so the situation is a bit easier. Your output voltage swing is only about 3.5V pk, so output current probably not an issue. However you DO need an artificial mid rail as others have stated (virtual ground).

\$\endgroup\$
1
  • \$\begingroup\$ Mid rail is just ground because C23 and C29 decouple the output from the DC bias. Connector is on another hierarchical sheet so is not pictured here. \$\endgroup\$ May 27 at 13:14
1
\$\begingroup\$

The limit will be the loading by the cable capacitances. You can amass 10s of nF of load capacitance if you use many long cables.

This would be like a load of only a few 100 Ohm at 20kHz and the opamp will eventually not be able to drive this due to the lack of output current capabilities, producing distortion.

\$\endgroup\$
2
  • 1
    \$\begingroup\$ recommend tens or dozens. 10ns can seem like 10 nanoseconds on here \$\endgroup\$
    – DKNguyen
    May 27 at 4:09
  • \$\begingroup\$ Yes sorry that was a typo \$\endgroup\$
    – tobalt
    May 27 at 5:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.