I have designed a circuit with 5V single-supply op amps, but since the inputs to the circuit are AC, I need a biasing voltage to replace the ground node on the equivalent dual supply circuit. Without getting into unnecessary circuit details, this biasing voltage may need to sink or source up to 10 mA.

The circuit amplifies signals of around 2 MHz, and originally I had used a wide bandwidth voltage regulator (TI TPS71701) to generate this biasing voltage, but I found out the hard way that this regulator is unable to sink current.

I've also tried using another, non-wide-bandwidth regulator (Diodes Inc. AP2202K), which did appear to work, but its datasheet makes no mention of being able to sink current, so I don't intend on relying on an undocumented feature.

Also I've tried using a low speed (1 MHz GBW) op amp (TI LMV342) configured in unity gain buffer topology. Although it does appear to work in practice, SPICE simulations indicate that there should be large (100 mV+) output voltage swings when sourcing/sinking a few mA of current at 2 MHz, which is unacceptable in this circuit. These swings disappear when frequency is reduced to 2 kHz. Hence I'm a little hesitant to put this into mass production.

I'm looking for suggestions of other low-cost circuits capable of generating a 2 V reference from a 3 V or 5 V supply, and which have no problem sourcing and sinking currents up to 10 mA at 2 MHz with little loss of regulation.

EDIT: I tried adding a load resistor to the output of the TPS71701 regulator, such that the amplifier circuit is able to sink current through this resistor. This did work, but now I have a constant 10 mA load (and this is a battery powered device). Actually, when in use, this product draws on the order of 1 A, so 10 mA might not be so bad, but still, I'd feel better with a circuit without a vampire current draw.

  • \$\begingroup\$ Why are you relying on the regulator to deal with the 2MHz rather than using a BFC (big fat capacitor) to reduce the requirements on the regulator? \$\endgroup\$ Aug 12 '15 at 16:53
  • \$\begingroup\$ While a capacitor would smooth out zero-mean (pure AC) transients, if the signal has a DC value below the regulator’s output voltage, then there’ll be a net sink of current which the regulator is unable to deal with. \$\endgroup\$
    – swineone
    Aug 12 '15 at 16:59
  • \$\begingroup\$ Yes, I realize that, but without the high bandwidth requirement you could simply use a termination regulator such as an LT1118 or just about any op-amp with a BFC and some compensation to keep it stable. \$\endgroup\$ Aug 12 '15 at 17:02

EDIT: I didn't see OP's edit where he says he already tried this. Anyway I'll leave this answer up for future readers.

You are correct that most voltage references cannot both source and sink current. One option is to either use a shunt reference or to pre-load a series reference:


simulate this circuit – Schematic created using CircuitLab

In this circuit, normally the reference IC provides 13 mA to R1. If the actual load (connected at OUT) needs to source current, that current will flow through R1 and U1 will just reduce its output current. I've shown both a filter capacitor (C1) to maintain stability of the reference/regulator, and bypass capacitance (C2) to take up most of the switching current and maintain steady voltage at the load under transients.

However, pre-loading is fairly wasteful of power (about 26 mW burned in the pre-load resistor, plus the extra current through the series reference). Andd 10 mA is a lot to ask from a precision reference circuit.

But if ~50 mW doesn't bother you, and you don't need sub-1% accuracy, it should work okay and doens't cost much.


How about this:

enter image description here

V1 is your voltage reference which gets buffered by the cheap OpAmp. The OpAmp loads a bunch of capacitors to your reference voltage. R1 makes sure that the OpAmp stays stable even with capacitive load.

I did a simulation using x7r capacitors with low ESR for C1 to C4. There is a loading effect on the voltage, but I see a maximum ripple of around 5mV here. That may or may not be good enough for your needs.

C4 is probably overkill and can be cut.

  • \$\begingroup\$ Should point out that if OP wants to use a different op-amp, change the capacitance values, etc., they'll need to check the stability of this circuit, since most op-amps don't like capacitive loads. A capacitor connected directly from the op-amp's output to the inverting input may improve stability. \$\endgroup\$
    – The Photon
    Aug 12 '15 at 15:54
  • \$\begingroup\$ @ThePhoton good point! Stability has to be checked. The LM358 was just chosen because it's my default OpAmp for cheap and noncritical single supply stuff. I didn't really picked it for any other reason or parameter. \$\endgroup\$ Aug 12 '15 at 15:58
  • \$\begingroup\$ I’m wondering how this circuit would fare if the resistor on the op amp output were replaced by an inductor (i.e. an LC filter). Would stability be an issue? Because this would eliminate the load regulation issue, and provide additional filtering. \$\endgroup\$
    – swineone
    Aug 12 '15 at 17:00
  • \$\begingroup\$ @swineone That is in fact a very good question. \$\endgroup\$ Aug 12 '15 at 17:52

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.