oscillating audio amplifiers

[N_Cook]

If they are liable to oscillate (rather than howlround) why is it always
ultrasonic about 50KHz, not 20 to 30KHz or 80-100KHz ?

 

[William Sommerwerck]

If they are liable to oscillate (rather than howlround), why is it always
ultrasonic about 50KHz, not 20 to 30KHz or 80-100KHz?

Always?

Are you familiar with the use of negative feedback, and how an amplifier
needs to be designed to be stable?

No amplifier has infinite bandwidth. The response eventually starts to roll
off, and phase shift increases with the rolloff. If the phase shift hits 180
degrees before the gain falls below unity (1), you have the conditions
needed for oscillation -- /positive/ feedback (180 + 180), plus gain greater
than one.

To get around this, we add a single "compensating" pole to the amplifier's
transfer function (basically, a simple low-pass filter) several octaves
below the amplifiers normal rolloff point. Because of its "distance" from
the normal rolloff, it now dominates the response. Because a single pole can
never have more than 90 degrees phase shift, the amplitude response will
fall below unity (1) before the net phase shift (amplifier's native response
+ added pole) reaches 180 degrees. No more oscillation -- assuming the pole
has been well-placed.

Because negative feedback is used to reduce distortion and source impedance,
and the reduction is proportional to gain reduction, the compensating pole
has to be at least an octave above the audible band -- hence the tendency of
oscillation or instability to occur at about that frequency.

 

[N_Cook]

Always?

Are you familiar with the use of negative feedback, and how an amplifier
needs to be designed to be stable?

No amplifier has infinite bandwidth. The response eventually starts to roll
off, and phase shift increases with the rolloff. If the phase shift hits 180
degrees before the gain falls below unity (1), you have the conditions
needed for oscillation -- /positive/ feedback (180 + 180), plus gain greater
than one.

To get around this, we add a single "compensating" pole to the amplifier's
transfer function (basically, a simple low-pass filter) several octaves
below the amplifiers normal rolloff point. Because of its "distance" from
the normal rolloff, it now dominates the response. Because a single pole can
never have more than 90 degrees phase shift, the amplitude response will
fall below unity (1) before the net phase shift (amplifier's native response
+ added pole) reaches 180 degrees. No more oscillation -- assuming the pole
has been well-placed.

Because negative feedback is used to reduce distortion and source impedance,
and the reduction is proportional to gain reduction, the compensating pole
has to be at least an octave above the audible band -- hence the tendency of
oscillation or instability to occur at about that frequency.

I suppose over the years about 5 such amplifiers, the fifth this week. All
about 50 KHz , different makes , types and powers. I just would have
expected more variety between say 30 and 80 KHz for amps with transistor fTs
for audio use

 

[William Sommerwerck]

I suppose over the years about 5 such amplifiers, the fifth this week.
All about 50 KHz, different makes, types and powers. I just would have
expected more variety between say 30 and 80 KHz for amps with
transistor fTs for audio use.

Point taken. I'm surprised you've had a problem at all. You'd think the amps
would have been better-designed. Your speakers might also be "pathological"
loads.

 

[Phil Allison]

"William Sommerwanker"

Always?

Are you familiar with the use of negative feedback, and how an amplifier
needs to be designed to be stable?

** Got absolutely nothing to do with an amplifier's inherent stability with
NFB.

The Kook is talking about audio power amplifiers oscillating because of
accidental POSITIVE feedback - ie capacitive coupling from the output to
a non-inverting input.

Each case if different, but something like a 50 to 100pF cap from the
speaker out to the non-inverting input will send almost ANY tube or
transistor power amp into full power oscillation. ( A direct connection
typically turns the same amp into a low frequency ( 1 to 5 Hz) square wave
oscillator. )

The actual frequency of oscillation is determined by many things, but is
typically about 50 kHz ( with stray coupling) because of the combination of
input and output low pass filters, the amps slew rate and the available
gain. While not a common event these days, I still see occasionally power
amps damaged by HF oscillations. Bad wiring practices, like unshielded or
unbalanced input cables running alongside speaker cables are the usual
culprits.

Also, it is very easy to make a guitar amp oscillate at a high audio
frequency - just put the guitar near the speakers and turn the gain and
treble controls up. Capacitive coupling from the frame of the speakers
injects HF signals straight into the pickups and any unshielded wiring in
the guitar.

Some "combo" amps have the speaker frames connected to the amp's chassis to
reduce this effect - though with tube amps it rarely damages anything.


.... Phil

 

[N_Cook]

While at it, assuming the amp outputs, Tx, fuses , zobell components etc
survive sustained 50KHz high level drive into speakers for a few seconds -
how would bass speakers fare as there would be no airflow cooling ? I assume
zobells would burn up to open cct promptly followed by the speakers.

 

[William Sommerwerck]

"William Sommerwanker"

** Got absolutely nothing to do with an amplifier's inherent stability with
NFB.

The Kook is talking about audio power amplifiers oscillating because of
accidental POSITIVE feedback - ie capacitive coupling from the output to
a non-inverting input.

Each case if different, but something like a 50 to 100pF cap from the
speaker out to the non-inverting input will send almost ANY tube or
transistor power amp into full power oscillation. ( A direct connection
typically turns the same amp into a low frequency ( 1 to 5 Hz) square wave
oscillator. )

The actual frequency of oscillation is determined by many things, but is
typically about 50 kHz ( with stray coupling) because of the combination of
input and output low pass filters, the amps slew rate and the available
gain. While not a common event these days, I still see occasionally power
amps damaged by HF oscillations. Bad wiring practices, like unshielded or
unbalanced input cables running alongside speaker cables are the usual
culprits.

Also, it is very easy to make a guitar amp oscillate at a high audio
frequency - just put the guitar near the speakers and turn the gain and
treble controls up. Capacitive coupling from the frame of the speakers
injects HF signals straight into the pickups and any unshielded wiring in
the guitar.

Some "combo" amps have the speaker frames connected to the amp's chassis to
reduce this effect - though with tube amps it rarely damages anything.


... Phil

 

[William Sommerwerck]

** Got absolutely nothing to do with an amplifier's inherent stability with
NFB.
The Kook is talking about audio power amplifiers oscillating because of
accidental POSITIVE feedback - ie capacitive coupling from the output
to a non-inverting input.

I don't know that that's what he /was/ talking about, but I understand what
you're saying. I've never seen it in a consumer audio system, but I can
imagine how it might occur on-stage.

By the way, supposedly correctly designed amplifiers /can/ become wildly
unstable with "pathological" loads. I've seen it.

PS: If you need to derogate my name, please find some cleverer way to do it.

 

[William Sommerwerck]

While at it, assuming the amp outputs, Tx, fuses , zobell components etc
survive sustained 50KHz high level drive into speakers for a few seconds -
how would bass speakers fare as there would be no airflow cooling ? I assume
zobells would burn up to open cct promptly followed by the speakers.

It depends. What's the woofer's impedance at 50kHz? If it's high enough,
there won't be enough current to overheat it.

 

[Phil Allison]

" Some wanker wrote"

" Nutcase Kook "

why is it
always ultrasonic about 50KHz, not 20 to 30KHz or 80-100KHz?



I don't know that that's what he /was/ talking about,

** HE does not know what he is talking about.

but I understand what
you're saying. I've never seen it in a consumer audio system, but I can
imagine how it might occur on-stage.

By the way, supposedly correctly designed amplifiers /can/ become wildly
unstable with "pathological" loads. I've seen it.

** The loads that sometimes make SS amps go loop unstable are moderate
values of capacitance - ie 10nF to 100nF at the speaker terminals. The
resulting oscillation may well be at 1MHz or higher and at a level the slew
rate allows.

Not at all likely to cause full power oscillation at 50kHz.


..... Phil

 

[Phil Allison]

"Nutcase Kook"

While at it, assuming the amp outputs, Tx, fuses , zobell components etc
survive sustained 50KHz high level drive into speakers for a few seconds -
how would bass speakers fare as there would be no airflow cooling ? I
assume
zobells would burn up to open cct promptly followed by the speakers.

** Only tweeters are at any risk. Bass and guitar speakers have lots of
inductance that makes their nominal impedance value rise by a factor of 5
to 10 at 50kHz.

This means the current flow is greatly reduced and the resulting heat
dissipation is but a small fraction of what you might wrongly suspect.


..... Phil