I have a Counterpoint SA-100 power amplifier of the 1994 year. This device exposes various problems:
1. Bias power supply 36V zeners (5W) overheating. I consider change the 470 ohm 1w resistors and the zeners for a constant current source (CCS). The CCS would be a transistor with a base red led and 270 emitter resistance connected to 50V and a 10K base resistance connected to ground. Therefore, the zener from the collector to ground would be of 1 watt with a best voltage regulation. Are there best ideas?
2. Instability problems. The DC offset and bias current changes with the temperature. Actually I adjusted the bias current (via a 1 ohm resistor in place of a fuse) to 400 mA. Is this correct?
3. Motorboating and/or ripple noise. This is very disappointing. Should I change decoupling capacitors and tubes?
4. Should I change the design of the tubes section? Is there a schema of the changes?
This my first post. Thank you very much and sorry for my English. If anything is not clear please just ask.
Andres Fontalba.
1. Bias power supply 36V zeners (5W) overheating. I consider change the 470 ohm 1w resistors and the zeners for a constant current source (CCS). The CCS would be a transistor with a base red led and 270 emitter resistance connected to 50V and a 10K base resistance connected to ground. Therefore, the zener from the collector to ground would be of 1 watt with a best voltage regulation. Are there best ideas?
2. Instability problems. The DC offset and bias current changes with the temperature. Actually I adjusted the bias current (via a 1 ohm resistor in place of a fuse) to 400 mA. Is this correct?
3. Motorboating and/or ripple noise. This is very disappointing. Should I change decoupling capacitors and tubes?
4. Should I change the design of the tubes section? Is there a schema of the changes?
This my first post. Thank you very much and sorry for my English. If anything is not clear please just ask.
Andres Fontalba.
Hi Andres,
Yes, those amplifiers have many design problems.
1. Yes, use a CCS to bias those zeners at a lower current. You will find that this solves a few problems. Those 10 uF caps are bad also.
I love it when people think about things! Repace the zeners also.
2. The bias will never really settle down on these. I can't remember the exact bias off-hand, but look for about 100 mA per output. For two devices this would be 200 mA. Your power transformer will last longer.
3. The heater filter caps may be defective now. Just change them. Motorboating may be due to excessive gain. If you have VR3 (pots), short them out. They go noisy.
4. You could. The amplifier should run as it is after doing steps 1 through 3. There are no official schematics for a modified tube section on these.
-Chris
Yes, those amplifiers have many design problems.
1. Yes, use a CCS to bias those zeners at a lower current. You will find that this solves a few problems. Those 10 uF caps are bad also.
I love it when people think about things! Repace the zeners also.2. The bias will never really settle down on these. I can't remember the exact bias off-hand, but look for about 100 mA per output. For two devices this would be 200 mA. Your power transformer will last longer.
3. The heater filter caps may be defective now. Just change them. Motorboating may be due to excessive gain. If you have VR3 (pots), short them out. They go noisy.
4. You could. The amplifier should run as it is after doing steps 1 through 3. There are no official schematics for a modified tube section on these.
-Chris
Hi Wavebourn,
Look up Michael Elliot. He is the designer of this product. He does tend to guard his designs for a price.
But you are correct, there are some fundamental errors in this design. Ilimzn gave a very good suggestion for a start. The initial post exchange pointed out a couple others. The incremental improvements would result in a complete revamp of this design, not just a modification.
-Chris
Look up Michael Elliot. He is the designer of this product. He does tend to guard his designs for a price.
But you are correct, there are some fundamental errors in this design. Ilimzn gave a very good suggestion for a start. The initial post exchange pointed out a couple others. The incremental improvements would result in a complete revamp of this design, not just a modification.
-Chris
Schema
The schema of the Counterpoint SA-100 is in the post #4 of this thread:
http://www.diyaudio.com/forums/showthread.php?postid=852504
Searching for the broskie's Aikido:
http://www.tubecad.com/2004/blog0011.htm
I will study it. Thank you very much.
The schema of the Counterpoint SA-100 is in the post #4 of this thread:
http://www.diyaudio.com/forums/showthread.php?postid=852504
Searching for the broskie's Aikido:
http://www.tubecad.com/2004/blog0011.htm
I will study it. Thank you very much.
¡Hola!
The schema is a postscript file that can be converting to pdf: http://www.ps2pdf.com/convert/convert.htm
The circuit of the SA-100 hasn’t power mosfets source degeneration resistors. Would be advisable include then? And feedback?
Thank you very much. Andres Fontalba.
The schema is a postscript file that can be converting to pdf: http://www.ps2pdf.com/convert/convert.htm
The circuit of the SA-100 hasn’t power mosfets source degeneration resistors. Would be advisable include then? And feedback?
Thank you very much. Andres Fontalba.
Hi Andres,
Thanks, that is the later model.
Source degeneration resistors will not sound very good. Also, this design does have some feedback through V2a. I don't know if it's intended or not. It is a no overall feedback design, so adding feedback will change the sound.
One of the big deals about this design was the very close matching between the outputs. This is due to the lack of source resistors. Michael did find that source resistors had a negative impact on sound quality with this design. He is correct as far as this design is concerned.
I think ilimzn's ideas have merit, or leave it stock save the zener biasing.
-Chris
Thanks, that is the later model.
Source degeneration resistors will not sound very good. Also, this design does have some feedback through V2a. I don't know if it's intended or not. It is a no overall feedback design, so adding feedback will change the sound.
One of the big deals about this design was the very close matching between the outputs. This is due to the lack of source resistors. Michael did find that source resistors had a negative impact on sound quality with this design. He is correct as far as this design is concerned.
I think ilimzn's ideas have merit, or leave it stock save the zener biasing.
-Chris
anatech said:
Unfortunately, they are necessary for good current sharing, whichg is why, without them, VERY tight (I would even risk saying, impossibly tight) matching of the output pairs is necessary.
The schematic also implies some mods, one of them being replacing the original pairs with IRFP parts, i would recomend IRFP9240 and IRFP340, WITH source resistors. Another possibility would be the 2SJ200/2SK1530 pairs, but without source resistors even those would have to be matched tightly. Lateral MOSFETs would also be possible but with somewhat reduced maximum current. They have the aded advantage of somewhat self-balanced current sharing assuming proper idle current is selected. Finally, a redesign of the output stage with BJTs may also be feasible - this implies emitter resistors as well as an additional driver stage.
IRFP have signifficantly higher gm than other mentioned types, so source resistors should have the least impact.
The biassing arrangement is more than a bit 'fishy'.
There is really no need to run that high a current through the zener diodes, since there is less than 1 mA drawn through the bias network. The high emitter resistor of the bias servo BJT is also suspicious. I would use a Vgs multiplier with a small HEXFET here.
The tube part is a bit perplexing. The gain is quite high too - using the tubes in an 'aikido' arrangement would lower this signifficantly, perhaps even a bit too low (preamp may be mandatory), but IMHO it would work much better.
The PSU is one important part that needs attention.
The output stage supply should really be consolidated because splitting the grounds that start off as a single ta on the transformer, into two, and then back into one, creates completely unnecessary ground loop problems.
The power transformer is very much undersized for the declared power output (especially into 4 ohms), the SA-100 I had cooked it's transformer, and I am told thi is not a rare occurence. Power rail sag can also be responsible for motorboating by introducing a LF positive feedback mechanism into the tube stage. This is almost invariably destructive to the output stage, and possibly, it's load. I would look into adding a separate transformer for the B+ and perhaps heater supply of the tube part. Regarding the DC heater supply, using a CRC filter with a very large C after the bridge, tends to run the bridge into huge current peaks, resulting in it heating up and cooking the PCB, first filter cap and often itself. In addition, I found the heater voltage to be quite on the high side - over 7V in the case of the amp I had. Most of these problems can be fixed by converting the CRC filter to a RC (with the two Cs in parallel) or RCRC filter. The front R signifficantly reduces surge currents in the rectifier, which takes care of all the associated problems.
Hi ilimzn,
Your observations about matching are correct. It takes me about two days to come up with a set or two out of 40 pieces of each. Not fun.
I did have an opportunity to look at a good matched set from Michael Elliot before installing them. I felt better about my own work as I match the devices much closer than he does. With this design it is mandatory.
The bias control is rather loose as you have pointed out. It only gets worse, ragged at lower currents. It takes a fair amount of time to get the bias right as there is a long time lag between adjustment and stabilizing.
If you look at the series resistance of the triode you will see that some signal gets through to form a feedback element. Adjustment of VR3 is able to vary the gain because of this. Unfortunately the signal path through V2a is a very non-linear one. Make sure that VR3 is removed and replaced with a wire link. BTW, gain differences between channels depending on matching tubes!
I totally agree with your other comments. The transformer will work well with a redesigned output stage that draws much lower bias current. This part is a common thing to fail if the bias is too high, ventilation is poor or the owner is "blasting" his stereo.
Best regards,
-Chris
Your observations about matching are correct. It takes me about two days to come up with a set or two out of 40 pieces of each. Not fun.
I did have an opportunity to look at a good matched set from Michael Elliot before installing them. I felt better about my own work as I match the devices much closer than he does. With this design it is mandatory.
The bias control is rather loose as you have pointed out. It only gets worse, ragged at lower currents. It takes a fair amount of time to get the bias right as there is a long time lag between adjustment and stabilizing.
If you look at the series resistance of the triode you will see that some signal gets through to form a feedback element. Adjustment of VR3 is able to vary the gain because of this. Unfortunately the signal path through V2a is a very non-linear one. Make sure that VR3 is removed and replaced with a wire link. BTW, gain differences between channels depending on matching tubes!
If you were to look at your amp, you would find that the second stage is running in a current starved condition. If you correct this, the gain will rise quite high resulting in motorboating. The current draw through the outputs also rises in concert with the motorboating, so you could blow the amp up by doing this! So the current starving is a "patch" to make the amp work without motorboating.
I totally agree with your other comments. The transformer will work well with a redesigned output stage that draws much lower bias current.
This part is a common thing to fail if the bias is too high, ventilation is poor or the owner is "blasting" his stereo.Best regards,
-Chris
Hola.
First, thank you very much for the answers. Excellent help!
Well, summary of changes in the schematic:
1. Adjust the bias current for about 100 mA per output MOSFET, total 200 mA.
2. Replace R62, R63, D60 and D61 (470 ohm 2W resistor and 36V 5W zener diodes) by a CCS and 36V 1W zeners.
3. Insert a resistor between bridge rectifiers and first capacitor C51 in the DC tubes heater supply.
4. Adjust VR3 for minimum gain to reduce motorboating.
5. Time to listen!
Advanced changes:
6. Add a separate transformer for the tubes B+ and heater supply.
7. Total redesign of the tubes section. See Broskie’s Aikido schematic.
8. Replace power MOSFETs by IRFP9240 and IRFP340 with source resistors. One question: how many transistors for output?
Again, thank you very much.
Andrés Fontalba.
First, thank you very much for the answers. Excellent help!
Well, summary of changes in the schematic:
1. Adjust the bias current for about 100 mA per output MOSFET, total 200 mA.
2. Replace R62, R63, D60 and D61 (470 ohm 2W resistor and 36V 5W zener diodes) by a CCS and 36V 1W zeners.
3. Insert a resistor between bridge rectifiers and first capacitor C51 in the DC tubes heater supply.
4. Adjust VR3 for minimum gain to reduce motorboating.
5. Time to listen!
Advanced changes:
6. Add a separate transformer for the tubes B+ and heater supply.
7. Total redesign of the tubes section. See Broskie’s Aikido schematic.
8. Replace power MOSFETs by IRFP9240 and IRFP340 with source resistors. One question: how many transistors for output?
Again, thank you very much.
Andrés Fontalba.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.