I have an old amplifier unit that runs on A.C and I'm trying to make it a battery powered amp I understand that the A.C voltage get rectified to 15+/15- D.C with a common ground so I used a 36v li-ion battery pack and the following circut to get a virtual ground

enter image description here

the circuit works but when it's connected to the amp power rails the virtual ground voltage drops to 0.8 V , is it because of capacitive load instability ? is there any solutions ?

Note : I'm using a LM324N Opamp I.C and the unit uses a NJM4558 DUAL OPERATIONAL AMPLIFIER I.C

  • \$\begingroup\$ Most likely there's something besides the 4558 drawing current through the ground. Also, that 1K resistor greatly limits the amount of current that you virtual ground can supply. \$\endgroup\$
    – JRE
    Aug 29 '16 at 14:41
  • 2
    \$\begingroup\$ Your 'virtual ground' would need to source & sink as much current as your power amp can drive. Even if you removed that 1k resistor an LM324 or similar will never do the job. \$\endgroup\$
    – brhans
    Aug 29 '16 at 14:41
  • \$\begingroup\$ What kind of amplifier is this, anyway? \$\endgroup\$
    – JRE
    Aug 29 '16 at 14:41
  • \$\begingroup\$ There are manufacturers that make better solutions for virtual ground than this. \$\endgroup\$
    – Voltage Spike
    Aug 29 '16 at 15:33
  • \$\begingroup\$ So does your old amp have three connections, +V, GND, -V? How much power does it supply? \$\endgroup\$
    – gbulmer
    Aug 29 '16 at 16:35

The circuit you show is almost guaranteed to oscillate. The 1K resistor makes it worse. It also can't handle much ground current before the op-amp saturates (due to the 1K).

Something more like this has a better chance of working:


simulate this circuit – Schematic created using CircuitLab

Since we don't know how much ground current there is, it's hard to design it properly. You may need a beefier buffer than an LM324 which is only capable of 10s of mA. If the load is grounded vs. a bridging amplifier it's probable unsuitable for anything more than a headphone amplifier. On the other hand, if the 4558 directly drives the output, it may be okay.

  • \$\begingroup\$ Can you go into detail about WHY it is "almost guaranteed to oscillate"? It seems fine to me, and is just an error amplifier with a gain of 1+1k/R_load. When I look at your configuration, there is nothing but AC current going through R2, as there is no DC path to ground, giving no DC feedback for your amplifier. How does this generate a stable reference voltage at the output? What am I missing? \$\endgroup\$ Aug 29 '16 at 21:52
  • \$\begingroup\$ @AndrewSpott you're missing the capacitive load hidden in the amplifier. It won't oscillate without that (at least an LM324 won't) \$\endgroup\$ Aug 29 '16 at 22:29
  • \$\begingroup\$ You mean the miller capacitance that sets the amplifier's low pass filter? Or some other capacitive load? How does your circuit get around that? \$\endgroup\$ Aug 29 '16 at 22:35
  • \$\begingroup\$ @AndrewSpott There's a pole added by the load capacitance (hidden on the amplifier schematic) and the Ro of the amplifier (hidden inside the op-amp model, typically around 50 ohms). The circuit isolates the load capacitance with 100R and establishes an AC feedback path through the 10n. \$\endgroup\$ Aug 30 '16 at 9:20

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.