I'm designing a relatively high power pulse generator.

Basically I have a microcontroller generating a clean square wave (0-3.3 V or 0-5 V) which then goes into my amplification circuitry. I want to keep the microcontroller front-end as a signal generator for various pulse shape experiments.

The specs are :

  • V = 0-40 V (for unipolar pulses), +/- 20 V for bipolar pulses.
  • Current - up to 20 A peak.
  • pulse length is [10 us - 10 ms].
  • budget is a up to 50$ for the amplification circuitry (have the micro front end).
  • Load : resistive > 2 ohm (hence the 40 V, 20 A requirements)

So far I have a 2 basic designs using bjts:

  1. A pnp -> emitter follower for low output impedance which seems to do the job, but requires the level at the entrance to be shifted very high (around Vcc which is in my case ~40 V)

  2. Trying to get around the signal shifting issues, I've come up with a different topology where a npn -> emitter follower.

This time, my first common emitter stage is inverting, so I need to invert the input signal in terms of its polarity. The problem with this amplifier is that it has a huge quiescent current running through its load resistor Using a common emitter and an emitter follower: using the first one as a switch and the second one as an emitter follower for reduced output impedance (schematic attached). (bottom part). The good thing about this design is that allows me to feed in a pulse 0-1.5 which is convenient to get for my micro (micro=digital=robust, at least in my mind).

I could use some advice as for the following points:

  1. What topology would you use for this type of application?
  2. How could this unipolar (now) amplifier be extended to a bipolar, meaning that the pulses are centered around a center ground line?
  • 1
    \$\begingroup\$ I noticed you've mentioned a schematic being attached but there doesn't appear to be one. If you can edit to include a link to it someone will be able to edit the question to include as an in-line image. \$\endgroup\$
    – PeterJ
    Commented Mar 19, 2013 at 2:21
  • 2
    \$\begingroup\$ Need a schematic, but intuition says a MOSFET is a better solution than BJT. \$\endgroup\$
    – Matt Young
    Commented Mar 19, 2013 at 2:31

1 Answer 1


Use Power MOSFETs with very low Rdson. Look into an N-channel as well, to reduce Rdson. Don't forget to manage heat dissipation. Even at 90% efficient, you still need to dissipate 80W! Since your microcontroller probably can't drive the FET quickly enough, you probably need a preamp/level shifter stage as well.

See http://www.onsemi.com/PowerSolutions/product.do?id=NVD5890NL as an example.

You don't mention your application, but if you are trying to build an audio amplifer, open loop Class D amplifiers are limited in their THD.


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