Oscillation (?) upon connecting speaker

[richie00boy]

I've just constructed an amp of my own design, a basic Lin-style with Darlington output devices. All is nice with no load, but when I connect a speaker my power supply starts current limiting on the positive rail. Because this causes the foldback to kick in, the rail shuts down and I can't see if something is oscillating.

Is it likely to be RF getting into the amp via the speaker lead? The lead is about 1m long basic figure-8 cable. I have a zobel on the amp. Do Darlingtons usually need a base-collector cap?

[clem_o]

Yep, that does seem like oscillation. You can put series incandescent lamps on the rails while you debug the amp - every time a lamp lights you know the amp's still unstable.

Darlingtons are fairly slow devices. They can introduce a phase shift that would cause that instability. Slow down the amp further at the VA and / or via the feedback (i.e. capacitor parallel with the fb resistor). Have you followed 'grounding' rules?


Cheers!

[richie00boy]

It just seems strange that it only happens when I connect a speaker, no signal has been applied.

Grounding is fine and has been proven on another amp I recently built.

[jaycee]

What about a resistor-inductor damping in series with the output ?

Just out of interest, is the schematic available ?

[richie00boy]

Maybe it does need an inductor, but with no signal and just connecting a load, I don't think adding something that helps with dynamic conditions is where I should be looking at the moment.

Here is the schematic, ignore the transistor type numbers they are just the models I have.

[Johan Potgieter]

Richie00boy,

After agreeing with clem_o, my tuppence worth would be a suggestion that you have some unwanted common resistance (could be very low) in the return lead of your load. This will be difficult to advise on without a photo of your amp, which may also not be of much help unless the tracks on the printed circuit are visible.

One possible way to see what is going on on a scope is to try making the load variable from say 100 ohms downward and see whether you can get the oscillation before everything shuts down at a lower load resistance, i.e. without a loudspeaker as the (variable impedance) load. I would certainly not expect any induced rf with such a short loudspeaker lead.

More suggestions are that a series inductor at the output could very well make a difference depending on the nature of your loudspeaker. Then I do not notice smaller polyester capacitors (about 100 - 470 nF) over the power supply caps C7 - C10, while a dominant capacitor C4 of 100 pF (value not very clear on my screen) may be a little on the high side. But I presume this circuit worked elsewhere. I also presume you were able to set the "idling" current correctly with R12. Furthermore I have found some semiconductor designs to be non-tolerant of the Zobel load, strange as that may sound.

But my approach would firstly be to try my suggestion in paragraph 2. If you can make the oscillation (if indeed that is the problem) visible, that would obviously help a lot. Also, if you can do this test with a small square wave input signal (about 5 KHz), it might warn even earlier of impending trouble. Start by looking at it with no load.

Wishing you luck.

[clem_o]

Originally posted by richie00boy
It just seems strange that it only happens when I connect a speaker, no signal has been applied.

Hi richie00boy,

Nothing strange about it oscillating 'only after' connecting the speaker. It's just a stability continuum - really bad amp oscillates without load, some will oscillate only when the load is connected, some will be fine until it's loaded with heavy capacitance, some only show stability problems when hit with a square wave input...

Cheers!

[richie00boy]

Thanks for everybody's input.

Originally posted by Johan Potgieter
After agreeing with clem_o, my tuppence worth would be a suggestion that you have some unwanted common resistance (could be very low) in the return lead of your load. This will be difficult to advise on without a photo of your amp, which may also not be of much help unless the tracks on the printed circuit are visible.

The load is returned by it's own ground lead to the PSU common.

Originally posted by Johan Potgieter
More suggestions are that a series inductor at the output could very well make a difference depending on the nature of your loudspeaker. Then I do not notice smaller polyester capacitors (about 100 - 470 nF) over the power supply caps C7 - C10, while a dominant capacitor C4 of 100 pF (value not very clear on my screen) may be a little on the high side. But I presume this circuit worked elsewhere. I also presume you were able to set the "idling" current correctly with R12. Furthermore I have found some semiconductor designs to be non-tolerant of the Zobel load, strange as that may sound.

The amp is totally my own design and I'm prototyping it. Leads are quite short and my last amp (of a totally different design though) was fine with this level of decoupling.

Yes I could set Iq fine with the pot R12.

I might try adding an output inductor, but really it's something I'm trying to avoid do to trying to make an easy to build amp. Also I see plenty of similar amps around without the inductor.

[lumanauw]

Hi, Richie,

Maybe you can try base stoppers 100-220ohm (R16-R17)?

Hi, Johan,

How are you

Furthermore I have found some semiconductor designs to be non-tolerant of the Zobel load, strange as that may sound.

I once experience exactly this. The amp is not stable WITH zobel.
Why is that?

Johan, I wanted to ask you something. Many people test amp with 4ohm//2.2uF.

I have an amp that can pass resistive load test, but oscilate with 2.2uF load (parrarel or stand alone capacitor).

Do I have to make it stable (compensate more) until it is stable with 2.2uF or not?

Where does this 2.2uF test comes from, it is absolute necessary when making amps (tobe stable with 2.2uF cap test)?

[john_ellis]

Darlingtons are two-for-one in the same can, which means that as they have the same frequency roll-off you get 12 dB/octave instead of 6 with a single. Ft's are around 4 MHz, so it is important to make sure that unity gain is reached at or below this to ensure that the amp. is unconditionally stable.

Your frequency limit is set with Re and Ccomp 47 ohm, 100pF=33 MHz if I am not mistaken! You may find that changing Re's to 100 ohm and using 150 pF might bring some order to your world (this is 10 MHz). Then you may need to think about reducing Re in the VAs to boost the gain a little to compensate for the reduction going from 47 to 100 ohm.

I agree you need also to fit an inductor to run amps into capacitive loads - they often still oscillate, even if they ought to be unconditionally stable. Soemwhere around 5 uH should be fine.

Suggest you also reduce the feedback resistance a little - too high values will give rise to higher distortion. (you have to change both this and the gain ratio resistor to keep the gain the same)

Cheers
John