Attached is my amplifier circuit schematics. It is laid out on a breadboard. There are two isolated stages - preamp and poweramp.

Each stage works great on its own. They go crazy when interconnected. I guess it is called motorboating effect.

The input waveform is severely distorted. The output waveform is worse. The power supply oscillates between 18 to 20 V, 0.8 to 0.4 amps at around 1.2 Hz (visually observed).

I have already added power supply AC drains (at the end of rails) and supply line resistors across the stages. They don't seem to help much. What else can be done?

 

It may be the way it has been laid out on the breadboard, resulting in loose coupling (positive feedback) between the output and the preceeding stages.
Kinda an electronic version of putting the microphone too close to the speaker.
If there is a PCB layout specifically for this design, I would suggest you etch one using it.

 

Hi,

One thing I see is that you don't have a servo loop. The 47 mike cap in the feedback is going to cause problems because it will induce a big phase shift in the feedback path, which will cause your motorboating. You need a DC path to elininate this. Generally, the resistive feedback sets overall gain. There will be a capacitor to ground in the feedback path to emiliminate high frequency ringing.

 

Originally posted by beenthere@Mar 29 2006, 02:02 PM
Hi,

The 47 mike cap in the feedback is going to cause problems because it will induce a big phase shift in the feedback path, which will cause your motorboating. You need a DC path to elininate this.

[post=15582]Quoted post[/post]​

Ideally the 47u cap will be removed. But that won't happen until I figure out how to reduce the DC offset to zero at the preamp out.

 

It looks to me like C3 is not in the feedback path. The signal feeds forward from Q5/Q8 through C3 to Q11/Q13 AND feeds back to Q2 through RF/RS.
The thing I note from your problem description is that the power supply oscillates at 1.2Hz. What is the current limit of the supply? greater than 0.8Amps? If I read the note correctly 16Vpkpk or 8Vpk into 8 Ohm requires about 1Apk, the supply should be able to source and sink a little more than this without current limiting. The 1.2 Hz is suspiciously close to the time constant formed by RSS and C++/C--. If you have the luxury of two bench supplies try powering the preamp and postamp separately and see if this eliminates the oscillation. If you don't have a second supply try increasing Rss to 200 or 300 Ohm and see if the frequency decreases proportionally. It looks like the preamp doesn't need the full +/-20V supply range(RE5 might need a small reduction), how about a couple of 3 terminal linear regulators to power the preamp, you lose 2-3V but you get many dB of PSRR in the 1Hz-30kHz range.

I would leave C3 in there, any DC offset will pass through and show up across your speaker, not a good idea.

 

Originally posted by CoulombMagician@Mar 30 2006, 04:05 AM
It looks to me like C3 is not in the feedback path. The signal feeds forward from Q5/Q8 through C3 to Q11/Q13 AND feeds back to Q2 through RF/RS.
The thing I note from your problem description is that the power supply oscillates at 1.2Hz. What is the current limit of the supply? greater than 0.8Amps? If I read the note correctly 16Vpkpk or 8Vpk into 8 Ohm requires about 1Apk, the supply should be able to source and sink a little more than this without current limiting. The 1.2 Hz is suspiciously close to the time constant formed by RSS and C++/C--. If you have the luxury of two bench supplies try powering the preamp and postamp separately and see if this eliminates the oscillation. If you don't have a second supply try increasing Rss to 200 or 300 Ohm and see if the frequency decreases proportionally. It looks like the preamp doesn't need the full +/-20V supply range(RE5 might need a small reduction), how about a couple of 3 terminal linear regulators to power the preamp, you lose 2-3V but you get many dB of PSRR in the 1Hz-30kHz range.

I would leave C3 in there, any DC offset will pass through and show up across your speaker, not a good idea.

[post=15622]Quoted post[/post]​

The bench power supply can supply 1 amp max at 20 V. With or without C++/-- and Rss, power supply voltage and current will still oscillate. I did not take note of oscillation frequency. C++/-- and Rss were added in attempt to damp the oscillation.

The preamp was originally tested for +/- 13 ~ 15 V operation, so downsizing the voltage is no problem... if it helps to be more stable.

Linear regulator... like LM117 ? I am not sure if I have any voltage regulator in stock. (Everything has to come from existing stock.)

 

It turns out that I had my R1 set too low.

New parms:
Changed Rss = 0
Changed R1 = 9.7k
Changed Rf = 120k, connected to Vout instead of common emitter out.
Added C12 (10nF) across Q12 collector emitter
Added C14 (10nF) across Q14 collector emitter

This time, it worked beautifully for about a minute at full power. Then thermal runaway melts Q2




Forum glitch has prevented me from uploading the image directly.

 

thanks..............

 

thanksssss