The first post shows an LTspice schematic. I thought it was your work. 50pF caps everywhere as if it is just a preliminary work. This kind of circuit is not easy to get working.
4 MHz? I guess you have a GM (Gain Margin) issue there. This is unfortunately, can only be solved if you understand the Nyquist. But the easiest thing to do, if you already have a PCB/circuit that might work, is to increase the feedback resistor (without changing the feedback value). For example, twice the current values are 3k (2x1500) and 200 (2x100).
The main compensation is the one on the VAS of course. Did you really have A1360 there?
Hi, AndriyOL!
The source of oscillation can be the output transistors themselves.
In this I was convinced, when I used the transistors IRF9530 in the output stage of the amplifier. I caught transistor specimens that were very prone to auto-generation and no measures with PCB and corrective capacitors allowed to eliminate the auto-generation.
I eliminate auto-generation only by changes in the Zobel chain.
The source of oscillation can be the output transistors themselves.
In this I was convinced, when I used the transistors IRF9530 in the output stage of the amplifier. I caught transistor specimens that were very prone to auto-generation and no measures with PCB and corrective capacitors allowed to eliminate the auto-generation.
I eliminate auto-generation only by changes in the Zobel chain.
The original schematic is not my work. I only draw it in LT Spice with some changes (TL431, lower output devices quantity, compensation), but I'm not familiar with simulation and there are no specific transistor models available.
Do you mean increase the feedback resistor like this: 6K (2X3000) and 400 (2x200)?
Yes, as on the schematic, layout and post 81: A1360 on the positive side and C3423 on the negative side of the VAS.
Zobel filter is intended to neutralize the effects of the driver’s voice coil inductance, but I doubt it can cure the oscillation of output transistors. It was reported somewhere that zobel resistor can even catch a fire in severe oscillation condition. In this case I think better approach would be to increase base stoppers of output transistors.
Last edited:
Yes. Regarding the design I personally wont use low supply rail for circuit like that.
I've tried this feedback change, but the amp board became even more unstable.
I also thought about PS voltage change as per original schematic. I noticed the amplifier becomes a bit more stable when the supply voltage changes to 2-3V higher due to mains fluctuations. Capacitors I’m using in PSUs are rated 50V for both. I think the maximum I can raise the supply rails are 50V for front end and 40V for main power supply.
What do you think about Audio1Man consideration in post 134? Does this method will do the same thing, limit the GB but in different manner, or it's completely different possible solution?
The picture on the scope isn't differing much with 20MHz switch
Last edited:
The reason for oscillation in an amplifier is:
- The gain is more than 0dB, at the frequency of 180 degree fase change.
As I see, the 180 degree frequency is approx 3MHz. At this frequency, the gain is approx 12dB.
In other words, it should be oscillating.
Base stopper has little effect unless you use high value resistor (it is usually used for trimming only). Capacitor is the main tool for compensation.
If you have the ASC file, actually you can trim the gain margin (GM) to increase it to above 20. Then it will be easier to troubleshoot the built circuit.
There are many ways to achieve the goal. This is no difference with working with speaker crossovers. Only the inductors and capacitors are very low value (and thus taking different forms: cob, trace, etc.).
Let's examine your circuit's compensations;
the b-c cap across 2SC/A1360 is very gross. It has the highest effect but we are working with many poles here so do not expect a throw of a big cap there will solve the problem. The proper location of the compensation is imho from base of the top transistor (TO-59) to the collector of the 2SC/A1360. Around 10 pF I guess is maximum I want to add. After the addition of this cap, there would be a new peak at higher frequency, at the output of the VAS...
A gross way to tame this peak is to add capacitance to ground right at the base of the output triple (base of 2SC/A1360). You already have an RC (100 Ohm - 220pF) at the bottom. But I think it is in WRONG position. Your circuit has the RC not at the base but at the collector.
A gross way is to use the base capacitor(s) at both 2SA and 2SC and without the 100 Ohm resistor. The series resistor is critical, it may add a terrible peak if used without calculation. It might correspond to the 10 MHz issue.
So how about doing these steps:
1) Do not change the FB resistor too much. Increasing 2x is fine (2k2 for the series resistor is fine).
2) Move the RC compensation to the base of bottom 2SA1360 and check the scope.
3) If there is improvement, lower the R (e.g. to zero) and/or put also at the base of 2SC.
4) Combine with moving the VAS compensation as stated before (10p from base of TO59 to collector of 2SC/A1360).
But the easier way is to do it in simulation, to make sure you have GM higher than 20. If you post the ASC may be people can help.
Well, at last a comment on why using or not ground planes on power amps.
Short distance paths between components is much more important. That certainly can be easier with dual-sided pcbs.
Separating grounds can also be important, but for lowering distortion, not preventing oscillations.
And of course keeping input and output paths well apart. THAT can cause oscillations.
Short distance paths between components is much more important. That certainly can be easier with dual-sided pcbs.
Separating grounds can also be important, but for lowering distortion, not preventing oscillations.
And of course keeping input and output paths well apart. THAT can cause oscillations.
That is why there are two files. If the one with a lot more parts worked there would be only one file.
I should notice, it was a mistake when I reported yesterday that with feedback resistors change to 3K\200R the amp become more unstable. Today when I wanted to change back 1.5K\100R I saw that one 1.5K was not soldered. Sorry about that. When I soldered it and checked the amplifier again, this feedback change showed little bit better stability than before, but not much. Than I soldered back 1.5K\100R and started to change compensation capacitors.
Following your advices:
- I changed the positions and value of compensation capacitors at VAS to 12pF from bases of small VAS transistors to collectors of TO-126 VAS transistors,
- shorted the 100R on RC at the base of predriver at negative side of the pcb (in fact RC was on the base of predriver, my mistake when I draw the schematic, sorry).
The result on the scope was better than before I guess. First picture.
- Then I added 200pF capacitor from the base of predriver on positive side of the pcb to input GND. After this modification the amp couldn't even start, my light bulb current limiter was activated periodically with 2 sec interval. When I removed this capacitor and checked again the scope showed even better result. Second picture. Btw, my scope shows almost the same signal as on the second picture when the amplifier is turned off. I've checked it again and found out that the signal on the scope keeps changing to signal as on the first picture but with less overshoot and eventualy almost straight line.
Should I try also to change feedback resistors to 2.2K\150R?
However, there is one kind of instability has left. When I touch by DMM lead to gnd or to voltage terminals, some input emitter resistors, oscillation appears immediately and current limiter is activated.
I should notice, it was a mistake when I reported yesterday that with feedback resistors change to 3K\200R the amp become more unstable. Today when I wanted to change back 1.5K\100R I saw that one 1.5K was not soldered. Sorry about that. When I soldered it and checked the amplifier again, this feedback change showed little bit better stability than before, but not much. Than I soldered back 1.5K\100R and started to change compensation capacitors.
Following your advices:
- I changed the positions and value of compensation capacitors at VAS to 12pF from bases of small VAS transistors to collectors of TO-126 VAS transistors,
- shorted the 100R on RC at the base of predriver at negative side of the pcb (in fact RC was on the base of predriver, my mistake when I draw the schematic, sorry).
The result on the scope was better than before I guess. First picture.
- Then I added 200pF capacitor from the base of predriver on positive side of the pcb to input GND. After this modification the amp couldn't even start, my light bulb current limiter was activated periodically with 2 sec interval. When I removed this capacitor and checked again the scope showed even better result. Second picture. Btw, my scope shows almost the same signal as on the second picture when the amplifier is turned off. I've checked it again and found out that the signal on the scope keeps changing to signal as on the first picture but with less overshoot and eventualy almost straight line.
Should I try also to change feedback resistors to 2.2K\150R?
However, there is one kind of instability has left. When I touch by DMM lead to gnd or to voltage terminals, some input emitter resistors, oscillation appears immediately and current limiter is activated.
Attachments
![20171225_163341[1].jpg](/community/data/attachments/601/601835-d09e445637fae62ae7e64b69d8e03446.jpg)
![20171225_172925[1].jpg](/community/data/attachments/601/601846-1761080d967d150906dec11d5cf893ac.jpg)
Last edited:
Just check by probe psu rails directly? No, I'm first, from what I know. I'm not familiar with crowbar approach. What's this?
If I understand you correctly, ground one end of the cap to the amplifier gnd, another end of the cap connect to the probe tip, probe ground to the amp ground and check each stage?
On the pcb l I was modifying there was Cdom capacitors on predrivers 100pF (the other channel has 50pF) and the amp behaved as I described. When I changed these capacitors to 50pF the amp become completely unstable with full scale oscillation on the scope.
If I understand you correctly, ground one end of the cap to the amplifier gnd, another end of the cap connect to the probe tip, probe ground to the amp ground and check each stage?
On the pcb l I was modifying there was Cdom capacitors on predrivers 100pF (the other channel has 50pF) and the amp behaved as I described. When I changed these capacitors to 50pF the amp become completely unstable with full scale oscillation on the scope.
Last edited: