Long wiring to the gates is bad, have you included gate resistors? Try and make the wires shorter.
What decoupling have you included on the driver board and is there any on the output board/devices?
What decoupling have you included on the driver board and is there any on the output board/devices?
Yes I included the gate resistors, I will start rewiring it.
If by decoupling you mean capacitors on power rails, then yes I used the capacitors showed on the diagram.
About the zobel: I tried to use it on one channel, don´t know the values right now, but after powering up with the zobel the channel got blown...
If by decoupling you mean capacitors on power rails, then yes I used the capacitors showed on the diagram.
About the zobel: I tried to use it on one channel, don´t know the values right now, but after powering up with the zobel the channel got blown...
OK, but if I rewire the whole thing tomorrow and the oscillations don´t stop, what should I do next?
Add a capacitor(s) somewhere? Increase gate resistors? Use more decoupling?
Add a capacitor(s) somewhere? Increase gate resistors? Use more decoupling?
you can try to use bigger c24 a c25 ,but the high frequency gain margin will be lower.
the best solution is to remake it using shorter wirings to the output transistors.
the best solution is to remake it using shorter wirings to the output transistors.
I rewired the amp and it is stable when nothing is connected, but it starts oscillating at cca 5Mhz, ~0.5V when I connect bare speaker cable to the output.
The cable is about 3m long and has 200pF capacitance.
Any help to this?
The feedback resistor is paralleled with 220pF to decrase bandwidth above 20khz, so why does it behave like this?
The cable is about 3m long and has 200pF capacitance.
Any help to this?
The feedback resistor is paralleled with 220pF to decrase bandwidth above 20khz, so why does it behave like this?
kubeek said:
The expcected behaviour with such a large cap in paralell to the feedbackresistor is heavy oscillation. By decreasing closedloop bandwidth you increase the amount of feedback for higher reqs, destabilizing the amp.
You can use this cap in any size only if the amp is unitygainstable.
Reduce this cap below 22pF or skip it.
Dou have a zobel at the output? (100nf+10ohm) Without it many amplifiers start to oscillate when speakercable is connected.
Mike
I tried it with and without 10ohm+10nF zobel and it didn´t change anything.
Tomorrow I´ll disconnect the cap and see what it does, thanks.
Tomorrow I´ll disconnect the cap and see what it does, thanks.
kubeek said:
It should be 10ohm+100nf, thats the typical value here.
The 220pf cap in paralell to feedback resistor is quite a guarantee for oscillation.
Mike
try an 8 ohm resistor as a load first (8 ohm 100w noninductive), connected at the output terminals. then add capacitance in 1,2,5 steps starting at 10pf (10,20,50,100,200,500, etc...) up to about .01uf (10nf). as you add capacitance, the amp will become more prone to oscillation. the points to expect signs of oscillation are at the crossover point (zero line) and at clipping (as the output wave drops out of clipping).
most amps get compensation caps on the b-c junctions of the VAS transistors to make them stable.... try 33pf on the b-c junctions of T19 and T21.
most amps get compensation caps on the b-c junctions of the VAS transistors to make them stable.... try 33pf on the b-c junctions of T19 and T21.
So I stripped out the cap (it was 150pF) and the amp works great with no signs of oscillating even when clipping hard.
I don´t have the zobel connected right now because I don´t have 100nF cap for this voltage.
Thank you all!
I don´t have the zobel connected right now because I don´t have 100nF cap for this voltage.
Thank you all!
Does 150pF capacitor exist in your schematic ?
I known my problem. Volume is not soldered carefully 😀
I known my problem. Volume is not soldered carefully 😀
No the 150pf cap wasn´t in the schematic.
I used it because I thought that by decreasing gain at higher frequencies I would prevent the oscillations.
It seems this is not true..
I used it because I thought that by decreasing gain at higher frequencies I would prevent the oscillations.
It seems this is not true..
kubeek said:
the opposite is true, the 150pF capacitor decreases phase margin of the negative feedback and is the true reason of oscillations. Learn a litle bit about a Nyquist criterion of stability and you´ll understand.
Kubeek, some good advice has been given already, but there have been many cases where this Crescendo based design has had instability problems. To summarize, here is what I would do
1. Insert an output inductor of around 1-2uH with parallel 2.2 Ohm resistor across the inductor. This should be placed as close as possible to the output devices.
2. Zobel network 10Ohms in series with 0.1 uF cap (use a good quality 250VAC cap here - not a cheap ceramic). Some people have a preference for putting the Zobel before the output inductor, others after the inductor. Try it before the inductor. You need the inductor to isolate the amp from the cabling capacitance.
3. Remove or reduce drastically the value of the cap that is in parallel with the feedback resistor. If the value is too high, it will cause oscillation for the reasons explained earlier in this thread - a small value can help to reduce hf distortion though (example, I use about 10pF in parallel with 4k7 feedback resistor). My advice: best left off until you have the amp completly stable - then tweak it through listening/scope observation. If you have access to distotion measuring gear, then this should also be applied to fine tune your amp.
4. The amp uses R-C compensation on the diff amp collector loads, but I think is would benefit from some miller capacitance from the VAS Cascode collectors to the cascode amplifer bases. Start with 68pF and reduce it - you should be able to get it down to 15pF - 22pF with the amp remaining stable. Use good quality caps here as well - not cheap ceramics. (I use silver mica, but some people don't like them). Another approach here would be to increase the diff amp emitter degeneration resistors - try going up 20-30%. This reduces the loop gain.
5. Mosfet wiring. Gate stoppers of at least 100 Ohms along with very short wiring is absolutely mandatory. If you continue to see oscillation at frequencies above 5MHz, I wouyld increase the gate stoppers - you can probably go to 330 Ohms or a bit higher.
6. Test your amp with it delivering a square wave output signal of around 2V at 1KHz into 2uF in parallel with 8 Ohms. It should be stable with no oscillation.
7. Some pointers. If your amp is oscillatiing between say 200 - 2MHz it is most likely a loop compensation issue. If its much above this, it may well be output stage parasitics - though mosfet output stage amps can also show loop stability problems at frequencies well above 2MHz because of their wider bandwidth.
Good luck with your project.
1. Insert an output inductor of around 1-2uH with parallel 2.2 Ohm resistor across the inductor. This should be placed as close as possible to the output devices.
2. Zobel network 10Ohms in series with 0.1 uF cap (use a good quality 250VAC cap here - not a cheap ceramic). Some people have a preference for putting the Zobel before the output inductor, others after the inductor. Try it before the inductor. You need the inductor to isolate the amp from the cabling capacitance.
3. Remove or reduce drastically the value of the cap that is in parallel with the feedback resistor. If the value is too high, it will cause oscillation for the reasons explained earlier in this thread - a small value can help to reduce hf distortion though (example, I use about 10pF in parallel with 4k7 feedback resistor). My advice: best left off until you have the amp completly stable - then tweak it through listening/scope observation. If you have access to distotion measuring gear, then this should also be applied to fine tune your amp.
4. The amp uses R-C compensation on the diff amp collector loads, but I think is would benefit from some miller capacitance from the VAS Cascode collectors to the cascode amplifer bases. Start with 68pF and reduce it - you should be able to get it down to 15pF - 22pF with the amp remaining stable. Use good quality caps here as well - not cheap ceramics. (I use silver mica, but some people don't like them). Another approach here would be to increase the diff amp emitter degeneration resistors - try going up 20-30%. This reduces the loop gain.
5. Mosfet wiring. Gate stoppers of at least 100 Ohms along with very short wiring is absolutely mandatory. If you continue to see oscillation at frequencies above 5MHz, I wouyld increase the gate stoppers - you can probably go to 330 Ohms or a bit higher.
6. Test your amp with it delivering a square wave output signal of around 2V at 1KHz into 2uF in parallel with 8 Ohms. It should be stable with no oscillation.
7. Some pointers. If your amp is oscillatiing between say 200 - 2MHz it is most likely a loop compensation issue. If its much above this, it may well be output stage parasitics - though mosfet output stage amps can also show loop stability problems at frequencies well above 2MHz because of their wider bandwidth.
Good luck with your project.
- Status
- Not open for further replies.