SC200 Amplifier Oscillation With PC Sound Card Input Destroys Q14 Output Transistor

After destroying output transistor Q14 twice and spending hours checking the circuit i put the output of the offending channel on the osciloscope. When i plug in the ouput from my PC sound card  in to the amp input the amplifier goes into osciltion and that is without any signal from the sound card. Just plugging the sound card into the amp sets the amp oscilating. I can see this on the scope. I have not worked out the exact frequency yet but it is several mega hertz. I have the amplifier running on a current limited power supply so the oscilations cant runaway and destroy anymore ouput devices. The other channel i have built seems fine and behaves normally when i plug it into my PC sound card. All the DC voltages seem fine on the oscilting amp. The DC output offset is 3mv, i can set the bias current to 4.4mv acros the output emitter resistors etc. I have no idea what the issue is, I suppose the place to start is changing the miller capacitor followed by all the other capacitors and them small signal transistors.


Has anyone had this issue before??

Regards
Heath

Hi Heath,
This along with your previous query have been forwarded to Editorial, we haven't overlooked them and hopefully we can get some faultfinding advice back. Interesting that one audio channel appears to be fine but the other is causing a problem. The SC200 is proving very popular, feedback and comments from any other constructors are very welcome too.
regards Alan W / EPE

I think this oscillating problem existed from the very start and has been the problem all along. Remounting the transistors has made no difference.
Regards
Heath

huddslad, I have some questions which would help diagnose your oscillation:
Did you build your circuits on printed-circuit boards provided for the SC200 design, or did you hand-wire your circuits? If you hand-wired, try moving the wires around while observing the oscillation on your oscilloscope. For both PCB and hand-wired, try moving your hand around close to the circuitry while observing the oscillation.
How big is the oscillation? Is it low level, or is it driving the output into clipping?
Does shorting the input cause oscillation the same way that plugging in the sound-card output does?Is the output open? Does putting a 6.8 ohm 3W resistor across the output change the oscillation?Is the ferrite bead on the input the same on both your amplifier channels?Is the 100 nF 250 V capacitor from the output to ground the same on both channels? Does paralleling another 100 nF 250 V capacitor change the oscillation?Note L2 is 2.2 uH in parallel with 6.8 ohms, in series with the output. Is the 6.8 ohm resistor the same in both channels? What kind of resistor is it? L2 and its resistor are there to prevent oscillations with some loads.

Hi Peter the amp was built on the SC200 circuits boards I purchased from EPEMAG.
The oscillation is about 4 volts peak to peak, at the amplifier output.
The frequency is about 1 mega hertz.The amplifier output was only connected to the oscilloscope when observing the oscillation. Shorting the amplifier input to ground causes the oscillation at the amplifier output to disappear.Leaving the amplifier input disconnected and open also causes the oscillation at the amplifier output to disappear.I have not installed the ferrite bead on either of the two boards.The output inductor L2 is installed correctly correctly in parallel with the 6.8 ohm resistor. The 6.8 ohm resistor is the same on both channels. The 6.8 ohm resistor is a wire wound resistor mounted on the underside of the boards.I have used BC556B and BC546B small signal transistors .I will carry out the following tests tonight after work and report back."moving my hand near the circuit and see if the oscillation changes" I will also install another 100nf 250V between the output and ground in parallel with the existing capacitor and see if the oscillation changes.
I will double check all the above observations are correct and report back tonight.RegardsHeath

HI Peter "moving my hand near the circuit and see if the oscillation changes"  does not change the oscillation amplitude or frequency in any way.I connected a 100nf 250V capacitor between the output and ground in parallel with the existing capacitor. This halved the amplitude of the oscillation to 0.2 volts peak to peak. The frequency did not change. (I meant to put 0.4V peak to peak in my previous post). I did not have a 6.8 ohm resistor so I used a 5 ohm high power resistor and connected this to the amp output. This imediately pulled the positive rail voltage down and led green led went out which mean the amp is drawing more than 230ma from this rail and my current limited supply. The negative rail green led stayed lit.

Hi Heath,I have a suggestion which may work. Connect a 1nF capacitor in series with a 100 ohm resistor, and connect these across collectors of Q1 and Q2. I have seen a number of designs with this connection, and in simulation, does stop high frequency oscillation in specific scenarios.
Else, is it possible another component which you did not replace has been damaged, but is still functional and degraded ?Regards,Richard.

So far I have replaced Q7. diode BAV21, miller cap 150pf, 1nf 63v cap connected to collector of Q7.I have also replaced the only two 220nf capacitors in the circuit.This has made no difference.I am not sure how I can isolate and narrow the problem down.I dont think modifyig the circuit is the answer especialy when one channel apears to be working and also i have been told that this amp is very popular and from that I assume lots and lots have been sold without. There are no other posts that i can seee from people with problms. Although i would like know how manyy of these actual boards have been sold in the u.k.

Has anyone got any sugestions.
RegardsHeath

I'm glad you used the PCB. That eliminates layout issues.The amplifier's misbehavior with a 5 ohm resistor load, suggests something is seriously wrong! What is the no-load no-signal DC output voltage? It should be zero. Does adjusting VR2 "offset adjust" permit setting the DC output to zero?Richard Shadbolt's 1 nF in series with 100 ohms across the collectors of Q1 and Q2 is definitely worth a try.The oscillation depends upon what is connected to the signal input, so it's the whole amplifier oscillating (rather than just the output emitter followers). To recapitulate: Input shorted no osc. Input open no osc. Input from sound card yes osc.With the sound card connected and the oscillation present, is there osc on the input? Also check with the scope both ends of the 10 ohm resistor between signal_in_low_side and amplifier_ground. Is there osc across the 10 ohm resistor?Try putting a 2.2 uH inductor where the ferrite bead goes.

Hi Peter the no load DC output is about 3 mV. I have done a away the VR2 and installed 50 ohm resistors. 3 mV DC output offset is fine I think.
To reiterate the behaviour  is as follows:-Input shorted I see no oscillation at the amp output.Input open I see no oscillation at the amp output.Input connected to sound card with no signal from the sound card I see oscillation.With sound card connected and oscillation present there is no oscillation on the amp input.There is no oscillation across the 10 ohm resistor. I removed this resistor on the off chance this would make a difference. It makes no difference to the oscillation.I have a theory which is because I used the higher gain BC546B and BC556B transistors these are causing the oscillation and making the amplifier unstable when the input in connected to a source. The Gain of these transistors are much high than BC546A and BC556A versions.The much higher Gain of the VAS stage using the higher gain transistors could be causing the oscillation??.Could this be cause?????.I have ordered BC546A and BC556A transistor version. I have a basic transistor gain measurement on my meter so i will compare the two versions.I think with the other channel I may have got lucky and installed a lower gain transistor samples especially in the VAS stage and the oscillations don't appear.What do think!!!.RegardsHeath

Hi Heath,Whilst waiting for the other transistors, maybe change Q7 which is a BC546 and Q8. It seems the negative swing circuit was damaged, and this may have fed through to the VAS stage. The BC546B has a maximum current of 100mA - which is not much. I would expect Q8 (KSC2690) would survive the excess current - but changing this will assure any degradation have been removed.
The fact that when you connect the PC sound card to the input, causes oscillation, indicates that the input pair circuit creates the pole to cause oscillation (could modify existing pole).
As i said earlier - try clipping a series 100nF and 100 ohm resistor across the collectors of the input pair. Or a 68k resistor - this will cause no damage, and will change the pole.
It is possible that B type transistors cause the oscillation, but based on typical value for hfe, and the variation of hfe for the BC546A - then i would not expect this to be the problem (the highest hfe for BC546A is close to the typical hfe value for the BC546B). Regards,Richard.

Peter would the transistor DC current gain hFE and small signal current gain current gain hfe make a difference???????.Am I grasping at straws??????.Regards
Heath

Hi Peter yesterday i already replaced Q7. diode BAV21, miller cap 150pf. Today i replaced  Q8 and there was no change in the oscilation frequency or aplitude.After this I connected a 100nf in series with a 100 ohm resistor across the collectors of the input pair as you sugested. I then put the amp output on the scope. There was no load connected. The fequency of the noise apeared to be 1khz and was a strange waveform,
I then put the scope on the amplifier input which showed oscilation frequency of 1 mega hertz , about 0.25V peak to peak. There was no load connected to the amplifier output.Maybe the input pair are damaged????.
Unfortunately this forum wont let me attach any photos of the scope plots.RegardsHeath

Heath, I don't think using BC546B and BC556B transistors instead of lower-current-gain BC546A and BC556A transistors made any difference.
The gain of the Q1 Q2 Q3 Q4 stage is the same for the A or B transistors, because the circuit impedances at the bases are low enough to make the base currents have little effect on the circuit's operation.
All of the SC200 is designed so that each transistor's base current is small compared to other currents flowing in the base circuits. This was done to eliminate effects of the base currents. To my eyes, the SC200 is an excellent design.
Heath, like most of us, including me, learned to think of bipolar junction transistors BJTs as base-current-controlled devices. This was a useful viewpoint back when transistors had barely enough current gain, too much cost, and were used without feedback. Today's transistors have plenty of current gain, very low cost, and in the SC200 are being used with lots of feedback.
Today, it is often more useful to think of BJTs as base-emitter-voltage-controlled devices, and treat the base current as an avoidable nuisance. The transconductance Gm (change in emitter current divided by change in base-emitter voltage) of a silicon BJT is given by the Ebers-Moll model:
Gm = Iemitter / Vthermal
where Iemitter is the emitter current, and Vthermal is the thermal voltage proportional to temperature in Kelvins (degrees above absolute zero), and is about 26 millivolts at room temperature. The Ebers-Moll model holds for all silicon BJTs, so the Gm does not need to be specified on the BJT datasheets.
For those used to working with JFETs, calculating the Gm of a BJT is enlightening.
The Gm, being proportional to emitter current, is exponentially nonlinear. The resistors in series with the emitters of Q1 Q2 Q3 Q4 reduce the effective Gm and make the effective Gm more linear.

I wonder how the SC200 compares to the NXP Semiconductor TDA8954 integrated-circuit 210-watt stereo power amplifier? Infineon and Texas Instruments also make power amplifier chips, though not as powerful.

Hi Peter thanks for your reply. I can confidently continue to use the BC546B and BC556B and swap out any failed and suspect transistors with the same. Although i did not need to buy any BC546A and BC556A versions and these were  actually harder to get hold  of and cost more than the B version strangely.
RegardsHeath

The TDA8954 apears to be a class D chip a different animal altogether. Distortion prformance appears to be apalling.
RegardsHeath

I forgot about the existence of class D chips. The Texas Instruments LM3886 is only 68 watts, but is all-analog and at least 19 years old. You might enjoy comparing the circuitry of the LM3886 and the SC200. What are the distortion specs of the SC200?

Back to your oscillation... With the input open, try connecting a jumper wire from the shell (ground) of the input connector to the shell (ground) of the sound card output. Does that make it oscillate? If so, does moving the jumper wire around change the oscillation?
There may be magnetic coupling (mutual inductance) between the 2.2-uH inductor and the input circuitry. I found a photo of the SC200. In the photo, the 2.2-uH inductor is wound on a plastic form 1 centimeter in diameter and 1.5 centimeters high, which  maximizes its magnetic-field output and hence maximizes mutual inductance. Is your 2.2-uH inductor the same construction?Wave a piece of sheet steel close to the SC200. Does that affect the oscillation? Try sheet aluminum also.
The magnetic coupling should be the same on both your SC200 boards, so would be only part of the explanation for your oscillation.
Check the 10-ohm resistor between the input connector and ground, as the resistor might be open. Check the 6.8-ohm resistor across L2 the 2.2-uH inductor. The inductor is there to provide high impedance at radio frequencies so as to force high-frequency currents (such as your oscillation) into the 6.8-ohm resistor. The resistor is there to kill the Q and damp the oscillations.
Does the SC200 PCB have a ground plane?

Hi Peter I realised last night that the ground/shield tab on the phono connectorwas nisulated so that when I plugged in the sound card the sound card shield/ground was notconnected to the ground on the amplifier. This was the problem all along i think,With only the live connected to the sound card there was highfrequency noise entering the amplifier and the amplifier was going bananas.DOH DOH DOH.I have compared this to other channel and my commercial CYRUS and QUAD amplifiers exhibitthe same behaviour if you don't connect the ground/shield.
More fool me.
RegardsHeath

Congratulations, Heath! An open input-ground connection would explain the oscillation.