Hi,
I've tried a multplicity of combinations. None improved the situation.
Tried disconnecting half the outputs from their driver and ran as a 3pair output stage. no change.
reduced the bias down to ClassAB (Vre=23mV), no change.
then reconnected back to 6pair in ClassAB, no change.
That's 4 options that didn't help.
Connected the sinks to the audio ground and repeated those combinations from above, no significant change was detectable.
Removed the ceramic 94pF Cdom and inserted a 100pF polystyrene, slightly worse. Added a ceramic 27pF //100pF PS then repeated all the combinations from above.
None helped.
Disconnected the outputs from the feedback link to the PCB output terminal. This is now much worse than last week.
That polystyrene does not seem to work. Next will be back to 94pF ceramic for Cdom.
Now I'm thinking that I've got to change something more drastic.
1.) reduce the open loop gain?
at the first stage (BJT LTP) or (second stage JFET) or (over both first + second stages)
2.) reduce the closed loop bandwidth/phase?
That Cfb parallel to NFB resistor. @ 0pF and @ 22pF it does not work. What about a range of values between 2.2pF and 15pF?
3.) I've not yet tried decoupling the output collectors to the sinks?
4.) what others options are available?
I've tried a multplicity of combinations. None improved the situation.
Tried disconnecting half the outputs from their driver and ran as a 3pair output stage. no change.
reduced the bias down to ClassAB (Vre=23mV), no change.
then reconnected back to 6pair in ClassAB, no change.
That's 4 options that didn't help.
Connected the sinks to the audio ground and repeated those combinations from above, no significant change was detectable.
Removed the ceramic 94pF Cdom and inserted a 100pF polystyrene, slightly worse. Added a ceramic 27pF //100pF PS then repeated all the combinations from above.
None helped.
Disconnected the outputs from the feedback link to the PCB output terminal. This is now much worse than last week.
That polystyrene does not seem to work. Next will be back to 94pF ceramic for Cdom.
Now I'm thinking that I've got to change something more drastic.
1.) reduce the open loop gain?
at the first stage (BJT LTP) or (second stage JFET) or (over both first + second stages)
2.) reduce the closed loop bandwidth/phase?
That Cfb parallel to NFB resistor. @ 0pF and @ 22pF it does not work. What about a range of values between 2.2pF and 15pF?
3.) I've not yet tried decoupling the output collectors to the sinks?
4.) what others options are available?
Hi Mark,
nice to hear that you can still make time for the important things in life.
Yes, I used 6pair MJL4281/4302 for the output.
These are about the same speed as the MJE15030/1/2/3/4/5
I have read that the drivers should always be faster than the outputs. So, I decided to use the faster drivers that have done well in my other amps. 2pair of 2sb649/d669. VAS are 1pair 2sa1360/c3423.
You might win the bet. But, I cannot recall anyone else saying that 4281/4302 are not suitable for this Klone.
nice to hear that you can still make time for the important things in life.
Yes, I used 6pair MJL4281/4302 for the output.
These are about the same speed as the MJE15030/1/2/3/4/5
I have read that the drivers should always be faster than the outputs. So, I decided to use the faster drivers that have done well in my other amps. 2pair of 2sb649/d669. VAS are 1pair 2sa1360/c3423.
You might win the bet. But, I cannot recall anyone else saying that 4281/4302 are not suitable for this Klone.
Hi Andrew,
I don't know if this has been mentioned before but in the Symasym amplifier - I think they reduced the cap across the feedback resistor to 3.3pF to stop the fast output transistors from oscillating. You might want to give this a go since you have tried almost everything else.
Regards
Harry
I don't know if this has been mentioned before but in the Symasym amplifier - I think they reduced the cap across the feedback resistor to 3.3pF to stop the fast output transistors from oscillating. You might want to give this a go since you have tried almost everything else.
Regards
Harry
Hi Harry,Harry3 said:
Yes, post1781 suggested this next route.
But I'll take out the polystyrene first to try to get back to that ealier more stable state.
I fear the four stage amplification used in this Klone is exacerbating the problem.
I am hoping that Mark is wrong and that the "fast transistors" are not the cause. I have seen lots of stability solutions in three stage amplifiers but I don't have any experience of four stage.
I have some thoughts about your amp Andrew:
(but I must admit that I dont really know how far from the original your amp are)
Wandering if your problem has to do with your emitter resistors? Think I read in an earlier post that you where going to use 0.35 ohms emitter resistors for the powers, right? This change the whole driver stage and its bias value to dissipate much less bias, e.g. approx. only 60-65% from the original KSA 100. Reason for this is that your more "modern" powers are working with less bias per item (only 416mA of a totals of 2.5Amps). Thus giving a higher value for Hfe, maybe 100-150? Also this gives you a much lower value if locking at the total Voltage for the drivers emitter resistors, e.g. this is the Voltage measured at the powers Base to audio pos. output terminal (could be as low as 0.79V ? a bit far from the original, 1.3-1.5V?).
The original KSA 100 had other working levels for the driver stage as for example: 58% more bias for the drivers (could be a factor for errors in a class-A amp) and much higher emitter resistors for the power stage, e.g. 1.0 ohm.
This was only some thoughts from my earlier experience with class-A amps constructions................
To correct this error there is two ways to do this: lower the emitter resistors for the drivers or rise the value for the powers emitter resistors (e.g. up to 0.68-1.0 ohm), or a combination of both.
If I may suggest something it would be this: change the power stage back to the original, e.g. change the emitter resistors to be as original KSA 100: 1.0 ohm. Make some measurements with only 4 pair of powers to begin with. If ok, then you might go further and try it with all 6 pair connected.
If this doesnt help at all, then I would like to suggest you make some measurements with no power transistors connected, e.g. drive and measure the main board only. If ok, then try to change your power transistors to some which is more like the originals (as the MJ15003/4 pair).
At least, after approx. 20 years of servicing audio gears there has been not only once but more cases I have experienced oscillating solid states power stages and that this error was caused by a faulty power transistor and/ or a faulty driver transistor.
Hope this gives you some ideas but Im not an expert
Hmmmmmmm, regarding "Expert" something suddenly came in to my mind. Do you know what is the definition of an expert? if not:
An expert is one who knows more and more about less and less until he knows everything about nothing (Nicholas Murray Batter)
Regards
(but I must admit that I dont really know how far from the original your amp are)
Wandering if your problem has to do with your emitter resistors? Think I read in an earlier post that you where going to use 0.35 ohms emitter resistors for the powers, right? This change the whole driver stage and its bias value to dissipate much less bias, e.g. approx. only 60-65% from the original KSA 100. Reason for this is that your more "modern" powers are working with less bias per item (only 416mA of a totals of 2.5Amps). Thus giving a higher value for Hfe, maybe 100-150? Also this gives you a much lower value if locking at the total Voltage for the drivers emitter resistors, e.g. this is the Voltage measured at the powers Base to audio pos. output terminal (could be as low as 0.79V ? a bit far from the original, 1.3-1.5V?).
The original KSA 100 had other working levels for the driver stage as for example: 58% more bias for the drivers (could be a factor for errors in a class-A amp) and much higher emitter resistors for the power stage, e.g. 1.0 ohm.
This was only some thoughts from my earlier experience with class-A amps constructions................
To correct this error there is two ways to do this: lower the emitter resistors for the drivers or rise the value for the powers emitter resistors (e.g. up to 0.68-1.0 ohm), or a combination of both.
If I may suggest something it would be this: change the power stage back to the original, e.g. change the emitter resistors to be as original KSA 100: 1.0 ohm. Make some measurements with only 4 pair of powers to begin with. If ok, then you might go further and try it with all 6 pair connected.
If this doesnt help at all, then I would like to suggest you make some measurements with no power transistors connected, e.g. drive and measure the main board only. If ok, then try to change your power transistors to some which is more like the originals (as the MJ15003/4 pair).
At least, after approx. 20 years of servicing audio gears there has been not only once but more cases I have experienced oscillating solid states power stages and that this error was caused by a faulty power transistor and/ or a faulty driver transistor.
Hope this gives you some ideas but Im not an expert
Hmmmmmmm, regarding "Expert" something suddenly came in to my mind. Do you know what is the definition of an expert? if not:
An expert is one who knows more and more about less and less until he knows everything about nothing (Nicholas Murray Batter)
Regards
Hi Flod,
thanks for the feedback.
Yes, I changed driver Re to 51r (from 75r) to match the higher hFE of the output stage and to match the lower Vre from the 0r35.
Now that I have been able to measure actual stage currents I think Dre=47r would be better.
I like your idea of removing the outputs and testing the drivers+voltage amp into a higher value load (4ohm*120~=470r for the test load).
Another thought.
What are the symptoms of inductive wire wound emitter resistors?
thanks for the feedback.
Yes, I changed driver Re to 51r (from 75r) to match the higher hFE of the output stage and to match the lower Vre from the 0r35.
Now that I have been able to measure actual stage currents I think Dre=47r would be better.
I like your idea of removing the outputs and testing the drivers+voltage amp into a higher value load (4ohm*120~=470r for the test load).
Another thought.
What are the symptoms of inductive wire wound emitter resistors?
Inductive emitter resistors (wire wound) would normally not be problematic to use due to the over all total inductance in cables and speaker are far more higher. The inductance in the resistor only add a very small part of this totals.
If using a high value (1.0 ohm) as in the KSA 100, and this without any problems, even those resistor have a much higher inductance value than yours 0.35 ohms .
Most power amp manufacturer uses those types of emitter resistors (cheapest available).
if inductance would be a problem you would observe it as a ringing on trailing and leading edges when amplifying square waves only, but its only a guessing here.
Regards
If using a high value (1.0 ohm) as in the KSA 100, and this without any problems, even those resistor have a much higher inductance value than yours 0.35 ohms .
Most power amp manufacturer uses those types of emitter resistors (cheapest available).
if inductance would be a problem you would observe it as a ringing on trailing and leading edges when amplifying square waves only, but its only a guessing here.
Regards
Frankly I would be surprised if differences in squarewave response would be audible. IIRC a lot of people experimented on the KSA50 klone and it didn't make much of a difference even though it changed (remember the 390pF miller caps?) the squarewave significantly.
Rather go too low with the bandwidth (ie. rounded squarewave) and forget about it. Measure up the spectrum and THD (easy nowadays with PC software and a half-decent soundcard) and see how little it makes a difference. HF instability will be filtered out by the card's steep cutoff filters so will not be visible, therefore limit the bandwidth so that the 10kHz square shows a slightly rounded response.
Rather go too low with the bandwidth (ie. rounded squarewave) and forget about it. Measure up the spectrum and THD (easy nowadays with PC software and a half-decent soundcard) and see how little it makes a difference. HF instability will be filtered out by the card's steep cutoff filters so will not be visible, therefore limit the bandwidth so that the 10kHz square shows a slightly rounded response.
Hi Pwatts,
I am trying to follow the message contained in your posting to Flod.
Are you saying that the sound quality effect of using wire wound emitter resistors is of no significant consequence?
or
That instability in the amplifier does not matter if the input signal is sufficiently filtered to remove the HF artefacts that may cause that instability to become apparent.
or
That the fuses blowing in my amp are not a problem because input filtering will prevent "something".
Help me understand.
I am trying to follow the message contained in your posting to Flod.
Are you saying that the sound quality effect of using wire wound emitter resistors is of no significant consequence?
or
That instability in the amplifier does not matter if the input signal is sufficiently filtered to remove the HF artefacts that may cause that instability to become apparent.
or
That the fuses blowing in my amp are not a problem because input filtering will prevent "something".
Help me understand.
My message was actually meant for you, and not in relation to Flod's comments. I have, however, never seen any noticeable difference between emitter resistor quality especially if they're of low value. You can try Dales or Mills but I doubt it will make a difference, and even more that it will address your problem. However, Krell used 5W Dales.
I seriously doubt that your problem is due to HF on the input since your 680pF input shunt cap will take care of that. If you're uncertain, increase that cap to 2.2nF and see what happens. Of course when it starts attennuating the audio band the square wave will become more rounded, but it will at least tell you whether that's where the problem lies.
I seriously doubt that your problem is due to HF on the input since your 680pF input shunt cap will take care of that. If you're uncertain, increase that cap to 2.2nF and see what happens. Of course when it starts attennuating the audio band the square wave will become more rounded, but it will at least tell you whether that's where the problem lies.
Thanks P.
It is easy to parallel an extra RF cap on the input, but that 0.68uS is what I have settled on to suit my ears/speakers.
It ocurred to me this morning that having grounded the main heatsinks that the drivers' sink is still isolated. It is common to the four drivers & Vbe & two VAS (but not cascode). It will be easy to add another ground link once I drill and tap a new hole.
That's another complete dismantle and reassemble.
Or, could I attach a grounding lug under the Vbe fixing bolt?
Would that be just as effective in grounding this smaller sink?
It is easy to parallel an extra RF cap on the input, but that 0.68uS is what I have settled on to suit my ears/speakers.
It ocurred to me this morning that having grounded the main heatsinks that the drivers' sink is still isolated. It is common to the four drivers & Vbe & two VAS (but not cascode). It will be easy to add another ground link once I drill and tap a new hole.
That's another complete dismantle and reassemble.
Or, could I attach a grounding lug under the Vbe fixing bolt?
Would that be just as effective in grounding this smaller sink?
Hmm...
Would you know it, it seems I already payed for some diodes!
http://www.diyaudio.com/forums/showthread.php?postid=1138702#post1138702
I believe the ones for Europe were going to be sent to someone in Europe for distribution over here?
Would you know it, it seems I already payed for some diodes!
http://www.diyaudio.com/forums/showthread.php?postid=1138702#post1138702
I believe the ones for Europe were going to be sent to someone in Europe for distribution over here?
