4
\$\begingroup\$

I'm looking for an existing design for a good yet simple class D audio amplifier using discrete semiconductors, no ICs. The purpose is educational, with a final product that actually could be used.

  • not bad quality (but we don't need top quality, or to compete w audiophile gear)
  • enough power to drive modest size speakers (it's hard to be quantitative about this)
  • no exotic parts.
  • within reach of serious hobbyists, EE students
  • not too simple; not for beginners or kids to make.
  • need not be commercially viable or survive outside an electronics hacker's home.
  • small transistor count prefered, but no need shave every penny or cut corners.
  • illustrates nicely the operation of class D amplifiers
\$\endgroup\$
  • 2
    \$\begingroup\$ I think getting rid of ICs completely conflicts with your goal of "illustrates nicely the operation of class D amplifiers". \$\endgroup\$ – endolith Feb 25 '11 at 2:37
  • \$\begingroup\$ Build a class A amp. You just need a voltage source, 1 inductor, a few capacitors, a bunch of resistors, and some bjt's (make sure to get a high current rated bjt for the final stage). It's not efficient but it sounds good. \$\endgroup\$ – user3770 Apr 7 '11 at 22:22
  • \$\begingroup\$ endolith, I would have thought so as well. Except until today, when I found something: electronics.stackexchange.com/questions/10668/… \$\endgroup\$ – zebonaut Apr 19 '11 at 14:37
9
\$\begingroup\$

I've done this before (though only in simulations) with a quad high-speed high slew rate op-amp. So there is one IC in it, but if you're really anal about that, you could implement it at the transistor level. Anyway, two op-amps are used as a triangle/ramp wave generator, which is then fed into a third op-amp which compares the audio signal with this ramp - this generates a PWM drive, which is applied to a MOSFET gate driver (a simple transistor driver I found on the 'net) which drives a MOSFET which passes current on request and filtered as required using inductors and capacitors. I don't have schematics, but this is how I would implement it.

The PWM frequency decides the quality of the audio. Higher frequency PWM will increase MOSFET heating and decrease efficiency due to increased switch rate but it will improve audio quality.

| improve this answer | |
\$\endgroup\$
  • 2
    \$\begingroup\$ I 1upped this because it is exactly how I would try it, too. Except, maybe, for the power MOSFETs' gate drivers, they can be built using discretes with not much additional effort (and some great insight into how switching stages can be built). But (fast) OpAmps for generating the sawtooth or triangle and for comparing the audio input will definitely be the best way if you don't want the project to be too complicated. \$\endgroup\$ – zebonaut Apr 8 '11 at 7:02
8
\$\begingroup\$

I've just found a reference design for a class D design using discrete components only. It is published and patented by NXP (TCFKA Philips).

I have not built it to verify the design, but they seem to use a very clever idea of making the differential amplifier and switching power stage self-oscillating by adding a capacitive feedback path.

Here are the links to...

I found it while looking for other stuff at nxp.com, but dames en heren, the design looks so cool I wish I had the time to start hacking it on a breadboard right away...

For educational purposes, I believe it is o.k. to build a patented circuit without trouble, but I'm not a lawyer. Other than that, it seems to be an absolutely perfect match for the requirements you mentioned.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ How would the THD be at say 90W say 20KHz? How would things be into a reactive load like a speaker passive crossover system. \$\endgroup\$ – Autistic Mar 21 '16 at 10:31
3
\$\begingroup\$

Probably this would too be difficult to do without ICs. Class D is definitely more complex than traditional ones.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Perhaps a comparitor chip could be allowed, or a simple current regulator. But I'm aiming for zero chips as much as can be reasonable. \$\endgroup\$ – DarenW Feb 25 '11 at 1:49
1
\$\begingroup\$

This doesn't satisfy your "no-IC" constraint, but may be is a good starting point:

International Rectifiers Audio Reference Designs

You can get the schematics or buy one of the kits (~US$300).

| improve this answer | |
\$\endgroup\$

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.