Oscillation under load

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Thanks for the explanation Eva you must have had a good long look at the citrcuit for me , thanks!
couple of questions:

What is split feedback? it is my understanding that the large feedback loop to the op-amp through the low pas filter would filter out just about all signal frequencies and allow DC to pass to correct offset on the output of the amplifier. Increasing the capacitor to 1000uf would not make a lot of difference to the operation would it?

Also what is phase margin? i am aware that as frequencies are greater attenuated by a filter their phase is also moved further from the reference but will increasing the gain make this phase change more exaggerated and what can i tell by this?


Cheers
Craig
 
Split feedback means to take as a feedback signal some kind of ponderated sum of signals from two or more different places. In your circuit, the op-amp is taking as a feedback signal the sum of its output voltage high-pass filtered and the output voltage of the amplifier low-pass filtered.

This low-pass filtering is a progressive process, so the op-amp is still taking some feedback from the output of the amplifier at audio frequencies, maybe also above them (that's why I recommended increasing the value of the filter capacitor).

Phase margin is a concept that applies to amplifier stability. In general, an amplifier is stable when it produces less than 180 degrees phase shift at the frequency where the open-loop gain has decreased below the closed loop gain it has been asked to produce (ie: when feedback is no longer effective). The difference between the actual phase shift caused by the circuit and 180 degrees is called phase margin, because the smaller the margin the more oscillation is likely to happen.
 
Craig,

Several observations here:

1. There's no power filter/by-pass capacitors across the op-amp. This may be the cause of oscillation. Try to put a 0.1uF(non-polar) and 100uF(electrolytic) capacitor across the power pins.

2. Base of Q12 and Q16 are clamped by diode D12. This voltage is too low to operate Q12 and Q16 in their linear region. Comparing to the original ckt, this biasing ckt is very different. The original ckt make way more sense to me. You need around 1.2V bias voltage.

3. If there's 0.6V across D12, how come there's only about 20mV across D8.

4. Your voltage measurements indicates that there's no bias current in the CCS, Q9. Otherwise, there should be some voltage drop across R12. Let say the bias current is at 10mA. The voltage across R12 will be 1V.

5. Item 3 and 4 above may indicate you have a PCB layout problem. Have you check the accuracy of your PCB layout?

My recommendation to you is to bring up one stage at a time starting from the input. Remove the transistors on the right until their stage is tested to prevent damaging them. To begin with, bring up the op-amp stage with the CCS. Check the CCS output as you test the overall amp. But of course you have to be careful about what circuit to include and what not. Otherwise, the test result could be misleading.


Hope this helps and good luck.

a3k
 
Hi.

I've looked at the schematic, and what I have to suggest may work for either of the two feedback resistors: I don't know which would be better. Since the oscillation only appears under load, it may be the latter stage feedback that could use the modification. But since the wiring and solder joints may have some inductance (open circuit at high enough frequencies) the other resistor might need it more. If it were me I'd try the one closer to the power transistors first.

When my father was building a Tiger amplifier he kept frying power transistors because the thing had a big enough 1 MHz or so oscillation that it went into thermal runaway. So I suggested that he place a few (10-30) pf across the feedback resistor. This would make it a dead short at higher frequencies, and make the gain unity, which would make it stable. Much to my surprise, my advice worked! After that I think he reckoned I might have learned something at that school. If you have an oscillator and a scope you can measure the effect on your audio bandwidth, which should be nil. If you think about it, it's usually the stray capacitance feeding back from output to input of a transistor that limits its frequency response. So if you draw the analogy between your amp and a single transistor, you are just adding a bit of "stray" capacitance.

-Chris Sullivan
 
Hi Craig,

I researched more into this amp. I found that it's design is special that the power transistors are operating as a current dummping devices. It's not biased as we normally see in most amps. So point #2 in my previous posting is not invalid.

It's quite an interesting design.

Sorry for the confusion.

a3k
 
Consort_ee_um, thanks i thought i had that article but your one includes some extra equations around figure 1 that i dont recall mine having.

A3k thanks for the suggestions I will try the things you suggested, although i dont know if i can remove my transistors very easily as the board has been resoldered so many times now it might not stand much more.
I will also recheck the voltage measurements as this was an old picture i posted last time i tried to get this amp working.
I do actully have 100nf caps on the op amp supply, i just forgot to include them on the diagram.

Chris, yea i tried sticking in small value caps around my power transistors and drivers but they didnt stop the oscillation :(
 
Ok ill give this current increase thing a try, the hard thing being which part of the circuit is lacking in current.

Looking at the measurements I posted ages ago on this diagram it looks like very little current is flowing anywhere within the circuit, which is very wrong for a class A amplifier, right?
I have a feeling the problem lies between the 2N5551 and the base of Q15 (BD244C).

The CCS loading the first transistor should be providing 4mA to the 2N5551 under the original quad design.
1.05V at its base - 0.6V, leaves 0.45V dropped across the emitter resistor of the CCS so 4.5mA is flowing.

is this right? if so the idle current stays roughly the same or even less up until BD244C's base. And the supposedly class A drivers arent even switched on.
Where should i start?

Hope that made sense, i am still learning electronics.

Thanks all
Craig

Its unlikely that this thread is truely still active anymore ?? however I think it worthwhile my adding comments and observations as after a thorough read and a serious look at this schematic as provided by craig405 I can see several issues that may directly result in instability.

1. when moving C4 (120pF) to the emitter of Q10 and R8 (180R) then it is recommended that you remove Q3 completely (reducing loop gain and the likelihood of instability).

2. the schematic seems to have an error where Q3 and Q3(??) are direct coupled to C4 (120pF)/R8 (47R). NOTE there are several R8s and Q3s.... whereas the original QUAD cct couples the collectors of these transistors to the reactive bridge via 330pF and 3k3 to ground...

3. Without doubt the cct must have suitable cap decoupling on the opamp PS's. 0.1uF + 20uF is more than adequate.. too high and you may have startup/stutdown PS imbalance and thumping etc... the original LM301 was an absolute shocker.. it was NOT suitable for audio applications and also had a 3pF snubber cap installed to suppress oscillation..

4. low value bypass caps on all electros is recommended at all times.

Im sure Craig405 has learnt a lot about this current dumping cct by now - I used it myself in my early days when starting out in electronics (graduate) in the early 1980s. It was a great way to learn about REAL-WORLD electronics and get past TEXTBOOK idealisms... Ive leant a lot from this cct.. Peter Walker was a genius especially so when you consider the state of commercial electronics in the 1970s and can put things in appropriate context.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.