Supply depended oscillation

[Thoru]

Hi all,

After fixing some other problems with my amp a new problem has turned up: oscillation! Read Slone's book again (it's roughly his figure 11.4 amp) and read all oscillation related threads here on diyAudio.com which I could find, but that didn't help.
My problem is a 3,9-4MHz oscillation. When I feed the amp with +-5V it's not noticable, but when I slowly increase the voltage it starts to appear. At +-8V the oscillation is some 300mVpp. When the supply is 11,5V it starts to repeatedly give of a DC offset at a rate of +-2 Hz. Seems people call this effect "motorboating", but I'm not sure it's the same.
I tried my lab supply instead of my home-brew unregulated supply, but it behaves exactly the same. I also grounded the heatsink but nothing changed. Tried single-pole instead of two-pole but the output was the same. Removed the Zobel-network and that increased both the voltage and frequency of the oscillation. It does not matter whether the speaker is connected or it's only my scope.
With respect to the changes I made to the original design: I seperated the single ground track in a signal and power ground. I added two 47Ohm resistors in the supply lines to have some lowpass filtering. I made the feedbacknetwork bandwidth limited to 50kHz. I replaced the CF stage with an EF one.

Does someone have any idea what might be wrong here?

Thanks in advance,

Remco Poelstra

[K-amps]

IS a load connected?

What happens at higher rails?

I have seen such oscillations in mass produced amps which have rails of +/- 55vdc.... they almost always oscillate at lower voltages (when a load is connected).

Maybe this might provide clues....

[Thoru]

Yes and no. Ussually I have my scope connected at powerup and then disconnect it to measure where the oscillation can also be found, but now always. With the speaker connected it's the same.

What do you mean with higher rails? Should I just continue to increase the supply voltage? I'm a bit scared of that because I really hate the sight of smoke from such a lot of work. But if you really think it will operate fine at +-30V (which is the final supply voltage), than I'll try

Thanks for your reply,

Remco Poelstra

[janneman]

Originally posted by Thoru
Hi all,

After fixing some other problems with my amp a new problem has turned up: oscillation! Read Slone's book again (it's roughly his figure 11.4 amp) and read all oscillation related threads here on diyAudio.com which I could find, but that didn't help.
My problem is a 3,9-4MHz oscillation. When I feed the amp with +-5V it's not noticable, but when I slowly increase the voltage it starts to appear. At +-8V the oscillation is some 300mVpp. When the supply is 11,5V it starts to repeatedly give of a DC offset at a rate of +-2 Hz. Seems people call this effect "motorboating", but I'm not sure it's the same.
I tried my lab supply instead of my home-brew unregulated supply, but it behaves exactly the same. I also grounded the heatsink but nothing changed. Tried single-pole instead of two-pole but the output was the same. Removed the Zobel-network and that increased both the voltage and frequency of the oscillation. It does not matter whether the speaker is connected or it's only my scope.
With respect to the changes I made to the original design: I seperated the single ground track in a signal and power ground. I added two 47Ohm resistors in the supply lines to have some lowpass filtering. I made the feedbacknetwork bandwidth limited to 50kHz. I replaced the CF stage with an EF one.

Does someone have any idea what might be wrong here?

Thanks in advance,

Remco Poelstra

Remco,

How did you limit the fb network bandwidth? I also think the 47 ohm in the supply line is too large, is this in the power outoput stage also? Looks like the amp tries to start up repeatedly only to be choked off as the supply collapses due to the 47 ohms.
Try to get back to the ORIGINAL design, get that going and THAN make the changes in a controlled way so you know exactly what causes what.

Jan Didden

[Thoru]

I replaced C6 with a value of 330pF. It should now form a lowpass filter with R10 (10k), with its 3dB point at roughly 50kHz.
47Ohm to big? I've seen it once operational with 100Ohm. The resistors are only in front of the smallsignal stages. So the OPS and the current limiting transistors are fed from the unfiltered supply. The smallsignal stages should only draw 12mA of current, which gives a voltage drop of 0.5V on 47Ohm. Is that really too much.
I did actually start with the unmodified circuit and than build on thing after another. That's why I'm also suprised with this oscillation. I've seen it run on +-30V before, and the only real change after that which I could remember is that I rerouted a track on the pcb. It was Q4's emitter who was connected to the unfiltered supply and I "fixed" that.
I've attached the PCB layout. The real situation can be found here: http://beryllium.net/~remco/foto-amp.pdf (536kb)

Thanks,

Remco Poelstra

[QSerraTico_Tico]

Any schematic? Not sure if I have Slone's book.

[Thoru]

Of course:
http://beryllium.net/~remco/versterker.pdf

Forgot to mention that I replaced the CF OPS with an EF design, but that's not visible on the photo taken...

Thanks,

Remco Poelstra

[Thoru]

Just thought it might be usefull to give the silkscreen with the pcb layout... Might help.

Remco

[QSerraTico_Tico]

to recapitulate: you have 2 problems
1) motorboating, low frequency oscillation
and
2) high frequency oscillation.
For problem 1 I would increase C7 to at least 680uF.
For 2 you have to find out if the output stage is oscillating or the whole amp.
See Linear Technology's Application Note 47 here:
http://linear.com/search/searchResul...eCriteria=an47
page 86 and further.
C8 seems a bit low to me, perhaps 47pF better?
It would not harm placing a simple low-pass filter at the input too.

[Thoru]

It seems that the low-frequency oscillation is a result of the high-frequency one, it only shows itself at +-11,5V supply lines. But I'll try your recommendations and see what they will do.
edit: Btw, there is a lowpass filter at the input.

Thanks,

Remco Poelstra