5
\$\begingroup\$

I have DAC 3.3V generated narrow square pulses that I need to amplify to 50Vdc. For this purpose an OPA551 single supply opamp works fine, but it's quiescent current of ~7mA @ 50V consumes too much power.

I hoped to use a LT1013 at 44V (trade off) and 0.5mA Iq, but it seems to only swing to the negative rail.

Because high voltage single supply opamps are hard to find and expensive, should I go with a transistor or FET amp circuit to amplify the 3.3V pulses? Any suggestions on how to achieve this? Or any other design ideas?

Update: Some more details on the circuit as requested. It is battery powered, the 50V is generated by a DC-DC converter, is stable and works ok with the OPA551. The pulse frequency is 1kHz and are 50µS wide and the required output current doesn't exceed 10-15mA. The OPA551 was a good fit besides the current consumption. That is to say I want to preserve as much power as possible because of the battery operation.

And indeed the LT1013 as a low power amp also had problems with the slew rate.

\$\endgroup\$
21
  • 1
    \$\begingroup\$ You're working with 50V swings, but 7mA is too much current from the amp? 50V: but what is the load? Across what kind of impedance does 50V have to be produced? \$\endgroup\$
    – Kaz
    Commented May 3, 2013 at 21:24
  • \$\begingroup\$ What accuracy is the DAC? Actually, what accuracy do you need for the 50V output? What frequency are the pulses at? Watch out for slew-rate limitations in low current opamps. \$\endgroup\$
    – markrages
    Commented May 3, 2013 at 21:29
  • 1
    \$\begingroup\$ If you're not going to provide the requested details -- all of them -- then the question will have to be closed as "not a real question". \$\endgroup\$
    – Dave Tweed
    Commented May 3, 2013 at 22:34
  • 2
    \$\begingroup\$ No, not yet. For example, you are using a "DAC" to generate the pulses, which implies that you want to have some control over their amplitude. How accurate does that need to be? And if the other parameters are not critical to you, you need to state that, giving some numbers. Would it be OK if the risetime on a 50 us pulse was 20-30 us? Would it be OK to use a low-voltage amplifier driving a transformer to increase the output voltage? \$\endgroup\$
    – Dave Tweed
    Commented May 3, 2013 at 22:56
  • 1
    \$\begingroup\$ Can the operating load be isolated from ground, i.e. not connected to the generating circuit's ground reference in any way? Or does it need to be referenced to the same ground, directly or indirectly? \$\endgroup\$
    – Anindo Ghosh
    Commented May 4, 2013 at 4:26

3 Answers 3

Reset to default
4
\$\begingroup\$

OK, so you basically need an audio amplifier circuit but without the high-power stage. So here's a op-amp + Sziklai output stage amplifier that should supply 50V swing at low currents.

enter image description here

It simulated OK in my head, but I tried it in LTspice to get the supply current and transient response, and tune the feedback for stability.

enter image description here

Blue trace is 3V supply current, and red is 50V supply current. The other two are input and output.

You could save some quiescent current by using a 1.5 V supply for the emitter voltage instead of the 40k+40k divider. 20k to a normal rail-splitter circuit would work.

\$\endgroup\$
2
  • \$\begingroup\$ This looks like a good solution. I will give it a try. Are the transistors and opamp critical as I don't have these laying around. Thanks for your extensive feedback! \$\endgroup\$
    – artelse
    Commented May 6, 2013 at 19:36
  • \$\begingroup\$ There's nothing very special about those transistors but they have a high Vceo. Many general-purpose transistors are usually only rated for 40 volts or so. You'll need more than 50, of course. \$\endgroup\$
    – markrages
    Commented May 6, 2013 at 20:09
2
\$\begingroup\$

Since you only want to make bigger pulses, not arbitrary waveforms, you don't need a general purpose amplifier anyway. All you need is the ability to control a high side switch, and maybe a low side switch depending on what the output drive requirements are.

A PFET with a pullup resistor on the gate, zener to limit the gate voltage, and a low side switchable current drive can turn the FET on and off with no quiescient current at all when off.

Above is a basic topology that works. If you need faster turn-off time than a pullup provides, you can add a PNP clamp that is turned on for a µs or so when trying to turn off the FET. You may also need a low side driver versus a simple passive pulldown on the output depending on what the sink current requirement is. You said 15 mA, but we don't know if that is only for sourcing or also for sinking.

\$\endgroup\$
2
  • \$\begingroup\$ Have been thinking about the high side PFET solution, but couldn't picture in my mind how to control it exactly. Do you have a concrete schematic example at hand? This is my first post here and am impressed by the high level of feedback I get here! \$\endgroup\$
    – artelse
    Commented May 6, 2013 at 19:41
  • \$\begingroup\$ To add to my above comment: I only need sourcing, not sinking. \$\endgroup\$
    – artelse
    Commented May 6, 2013 at 22:32
1
\$\begingroup\$

I finally got around fixing the quiescent current problem. I slapped my forehead not thinking about it before, but the answer is simply to make use of the quite extreme 1:20 duty-cycle of my application. Dynamically switching the op-amp on/off when needed reduces current consumption accordingly. I also found the OPA454 op-amp with a much better Iq and has an enable option which made interfacing to my 3.3V logic simple.

\$\endgroup\$

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