For the past 6 months I've been designing and building a 30W 3MHz audio modulated Tesla Coil driver.

For this, I've created a +-15V linear regulated power supply, an AM modulator using an MPY634, and a class AB output amplifier which uses a 2SC5200 / 2SA1943 BJT pair.

The only thing that is still giving me trouble is the class AB output amplifier, which clips the positive half wave when the input from my signal generator goes above a certain voltage.

Here are the schematic and a scope shot:

Schematic of class AB output stage amplifier

Scope shot of clipped wave

I know clipping can occur when the output voltage reaches the rail voltage, but the rail voltage is +15V, while the positive output wave is only at +2V when it starts clipping.

Meanwhile, the negative wave remains a perfect sine.

I tried to simulate the circuit in EveryCircuit, but in the simulation everything looks fine.

Any idea what I am doing wrong here?

Thanks in advance for any help!

EDIT: Decoupling caps C1 and C2 were NOT present during my test, and my signal generator is an MHS-5200A

EDIT 2: Like @Transistor suggested, I measured the voltage between the two emitters, removed the signal generator from the circuit, and shorted the input terminal to ground.

The result is this 8.5Vpp high frequency (even though the crappy scope says 0 Hz) noise!

8.5Vpp noise

Is my power supply causing this? (A higher quality PSU is already on its way) Or is it indeed a problem with the biasing circuit?

If the latter, can someone please explain to me how to determine a proper biasing circuit? I've followed all kinds of tutorials, and apparently they are all missing something crucial.

ALSO, it seems like R2 is getting much warmer than R1.

It also seems like the waveform clipping is reduced as frequency is increased.

EDIT 3: I triple checked the connections, and they are correct. Then I measured the resistors, and they both show around 470 Ohms. Finally, I replaced the 1N4148 diodes with the much more bulky HER305 diodes, and now there is no clipping anymore...

However, these HER305 diodes have a forward voltage drop of 0.5V, which might not be ideal for biasing the transistors, which usually need 0.7V.

I ordered some 1N4937's to see if those would be a better fit (I am planning to push this amp up to 3 MHz, so I need fast diodes).

Thanks everyone for your help!

  • \$\begingroup\$ Switch the scope to DC coupling and have a look at the signal on the emitters. Post another photo. What is the voltage there when input is shorted to ground? \$\endgroup\$
    – Transistor
    Mar 21, 2020 at 10:28
  • \$\begingroup\$ Where did you measure, on emitters or after the dc blocking cap? \$\endgroup\$ Mar 21, 2020 at 10:46
  • \$\begingroup\$ The problem may well be in your driver stage aka "siggen" in that schematic, and we don't have any idea what's in there. As Dave points out, C1 may not be helping. \$\endgroup\$
    – user16324
    Mar 21, 2020 at 12:40
  • \$\begingroup\$ I'll try DC mode. I measured before and after the 50 Ohm load resistor RL. Decoupling caps C1 and C2 were NOT present during my test, and my signal generator is an MHS-5200A. \$\endgroup\$
    – kregus
    Mar 21, 2020 at 16:25
  • \$\begingroup\$ @Transistor I did what you suggested and posted a picture in the question above. Seems like there is 8.5Vpp noise in between the emitters, even when no signal is being applied... any idea how I can fix this? \$\endgroup\$
    – kregus
    Mar 21, 2020 at 21:05

3 Answers 3


I'm not sure why you AC-coupled your input and output, given that you have balanced power supplies, which makes it unnecessary.

The point that the commentors are making is that your bias network doesn't do what you think it does. You're assuming that the junction between the two emitters is close to zero volts with no signal. But in fact, the actual voltage is a function of the relative current gain of the two different transistor types.

Since the gain of an NPN transistor is typically significantly higher than the gain of a similar PNP, that junction is actually sitting quite close to the positive rail, even with no input. You need to fix your biasing scheme, possibly by employing some negative feedback at low frequencies/DC.

But just removing the input capacitor would also fix the problem, assuming that the source impedance is low enough.

  • \$\begingroup\$ Hi Dave, thanks for your answer! This is the first ever amplifier I've ever built, and I find very contradicting information about biasing the amp correctly, so if you could point me in the right direction, that would be amazing! And I forgot to mention that C1 and C2 were NOT present during my test, so this result is WITHOUT caps. \$\endgroup\$
    – kregus
    Mar 21, 2020 at 16:28

Q1 emitter-base short circuit. Check out Q1 or replace it! The effect is also apparent in the absence of 15 volts of the Q1 collector.

  • 2
    \$\begingroup\$ I was skeptical but a CircuitLab simulation of the circuit (with the capacitors in place - which the OP now says are not in the actual circuit under test) showed that this could produce the symptoms. I suspect that Dave Tweed may be correct though. This answer could do with a bit more detail to help the OP why this would give the observed affect. \$\endgroup\$
    – Transistor
    Mar 21, 2020 at 16:38
  • \$\begingroup\$ This was an interesting idea, and the effect indeed shows up in simulations, but I changed out the 2SC5200 transistor for a brand new one, and unfortunately the same things happened. I also used a multimeter, and there was no short. \$\endgroup\$
    – kregus
    Mar 21, 2020 at 20:52
  • \$\begingroup\$ @kregus Please make sure the installation is correct and + 15V is present of the collector of Q1. If possible, please show us detailed pictures of the Q1 installation! \$\endgroup\$
    – Peter MP
    Mar 22, 2020 at 7:54

After replacing the 1N4148 diodes with the much stronger HER305 diodes, the clipping disappeared, even though the 1N4148's seemed in fine working order when tested with a multimeter.

Still not sure why this worked, but it did.

I was able to push the output wave up to 13.7V @ 3 MHz.

Scope shot

And for those who were asking, this is what my amplifier looks like (before I changed out the diodes).


The 2SA1943 is on the left, and the yellow wire is the -15V.

  • \$\begingroup\$ Why on the first test the frequency is 1 kHz and on the last test 3MHz? It is interesting to see what happens at 1kHz again. \$\endgroup\$
    – Peter MP
    Mar 22, 2020 at 20:46
  • \$\begingroup\$ At 1 kHz the same thing happened as before, minus the clipling. \$\endgroup\$
    – kregus
    Mar 22, 2020 at 21:08
  • \$\begingroup\$ According to my simulations on LTspice Q1 is out - open collector. \$\endgroup\$
    – Peter MP
    Mar 22, 2020 at 21:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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