The most common complaint about electronic limiters is that they affect the sound badly, One of the first things we do to so equipped small hi-fi amplifiers is ditch the limiters. For some designs, it's probably the most effective improvement.
Edit: thanks, that's massive driver gain - small wonder you ran into trouble with oscillation. I would suggest mid range Hfe grades for all amp parts except the input pair. Sometimes you have to match and have high Hfe here as well,
Edit: thanks, that's massive driver gain - small wonder you ran into trouble with oscillation. I would suggest mid range Hfe grades for all amp parts except the input pair. Sometimes you have to match and have high Hfe here as well,
Last edited:
With due respect for your successful repairs to commercial equipment, amptech, the intentionally wide stability margins and general ruggedness of those proven designs don't apply here. This is an "accidental" trial and error, simple domestic size one.
You can pull a lot of the safety measures on PA gear and they will work fine and sound better too - until you give them something they were designed to do - like cranking at full limiting into absurdly low, reactive loads.
Then you can tell us what the stupid little resistors were actually for.
The Zobel RC network is intended to damp oscillation alright but not really to make an unstable amplifier stable. The difference being that the 50mV oscillation is constant, I guess and the Zobel is meant to deal with transient situations due to the program, cable, load and level.
It's good you have control and a handle on the problems by using the scope. The amp should be quite stable without a filter, even though I've seen some behaving as you see. This is the point where we need the theory book and read Slone, Self or Cordell to find the stability relationship between the open loop bandwith, Nyquist stabi;ty, phase margin, compensation and a heap more amplifier variables that affect stability, This is complex, even if straightforward calculations or a simulation program can quiickly find the answers.
It's likely the oscillation is a local one if it remains at or below 50mV under load. A bunch of parallel heavy resistors with a rating overall of ~10W should be enough to find what the matter is. Wirewound resistor inductance is OK if you have enough in parallel and reducing the L.
Connect a resistance like this of around 8 ohms to the output but keep the input quiet by fitting a 1k shorting resistor across the input. The intention is to load the amplifier as an ideal speaker would, to see what happens with the oscillation level. If it remains steady or falls under load, you may need not be concerned as the Zobel should be fine at the job.
Of course, recheck with the Zobel in circuit too.
It's good you have control and a handle on the problems by using the scope. The amp should be quite stable without a filter, even though I've seen some behaving as you see. This is the point where we need the theory book and read Slone, Self or Cordell to find the stability relationship between the open loop bandwith, Nyquist stabi;ty, phase margin, compensation and a heap more amplifier variables that affect stability, This is complex, even if straightforward calculations or a simulation program can quiickly find the answers.
It's likely the oscillation is a local one if it remains at or below 50mV under load. A bunch of parallel heavy resistors with a rating overall of ~10W should be enough to find what the matter is. Wirewound resistor inductance is OK if you have enough in parallel and reducing the L.
Connect a resistance like this of around 8 ohms to the output but keep the input quiet by fitting a 1k shorting resistor across the input. The intention is to load the amplifier as an ideal speaker would, to see what happens with the oscillation level. If it remains steady or falls under load, you may need not be concerned as the Zobel should be fine at the job.
Of course, recheck with the Zobel in circuit too.
Last edited:
there's plenty of information regarding amplifier design work on rod elliot's audio page's... but agree cordall's/self's/ slones books also offer plenty of useful coverage on amplifier's... as for zobel net works they can differ from 2.2 ohms via 47nf to the more common type 10 ohms/100nf.. amplifier's that oscillate will burn up the zobel resistor.. 1 question what type of in cable are you using for the signal input... yes short the input and recheck...
Ordinarily, you don't mess with the compensation capacitors (the local Miller cap across the VAS or the Global NFB at C2. The 10pF cap tells us there was likely some HF oscillation in the original design that needed damping - even RF susceptibility. The impedance is only 12k in this circuit, so that's not much damping. You could temporarily increase it to 20 pF or so and check the effect on the oscillation level.
C15 miller cap has been changed drastically since your first schematic. I guess you did that to kill oscillation but it will also kill bandwidth and require other changes at the input stage to get the amplifier running correctly. Standard 100pF is usually quite sufficient. Try to get things original in the fundamentals then try simply reducing the gain by using original or similar spec parts. This is more likely to get your stability back.
Oscillation occurs by too much phase shift from input to the output. Each active device or reactive component shifts the signal phase so many degress unitil you get to the point where the feedback becomes positive (approaching 180 deg.) and full-on destructive oscillation occurs. Low level oscillation arises from local resonant circuits in the output stage, feedback network or anywhere the wiring is messy, gain is high and stray capacitance can have a resonant effect.
Broadly speaking, compensation adapts the feedback network to counter that shift but only up to a point and since it's all capacitively controlled, the amount of feedback will change from low to high frequency where it drops off and oscillation becomes increasingly possible. Hopefully, the HF gain (bandwidth) is dropping off faster than the feedback but if not.....!
C15 miller cap has been changed drastically since your first schematic. I guess you did that to kill oscillation but it will also kill bandwidth and require other changes at the input stage to get the amplifier running correctly. Standard 100pF is usually quite sufficient. Try to get things original in the fundamentals then try simply reducing the gain by using original or similar spec parts. This is more likely to get your stability back.
Oscillation occurs by too much phase shift from input to the output. Each active device or reactive component shifts the signal phase so many degress unitil you get to the point where the feedback becomes positive (approaching 180 deg.) and full-on destructive oscillation occurs. Low level oscillation arises from local resonant circuits in the output stage, feedback network or anywhere the wiring is messy, gain is high and stray capacitance can have a resonant effect.
Broadly speaking, compensation adapts the feedback network to counter that shift but only up to a point and since it's all capacitively controlled, the amount of feedback will change from low to high frequency where it drops off and oscillation becomes increasingly possible. Hopefully, the HF gain (bandwidth) is dropping off faster than the feedback but if not.....
!- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.