Andypairo said:
That may be the solution. I never had the luxury of a lot more heat sink than I needed on production units.
I need to address an issue that has come up with the "safety resistor", R35 on the 100W schematic, R25 on the 50W schematic. Two people have reported that they had to remove that resistor to get the amp to bias up where they wanted it. That resistor was not in the production units or any units I built. It was suggested as a way to keep the output bias from going too high in the event of an open in the bias string. It seemed like a good idea, so I ran simulations with it and it worked as planed. Unfortunately, in the real world with real units, it seems to prevent the bias circuit working as it should. I would recommend leaving out that resistor.
I think I have posted this before, but this is a good time to repeat it.
After construction, set R26 to 30-50 ohms measured across R5. After power up, adjust R26 for minimum DC at The base junction of Q7 and Q10. Be careful not to short anything. One end of R11 will also work for this measurement. Make sure to use the end that is common to the base junctions.
Next, set both R32 and R33 to minimum. Adjust either R32 or R33 for minimum offset at the output. Only one will need to be adjusted. The other is left set to minimum. This is done without a load. After the offset has been set, set the bias either for minimum distortion or at the operating point you feel is optimum.
Check the settings and adjust if needed. Let the amp run with no input and no load for 30 minutes then recheck and readjust.
After construction, set R26 to 30-50 ohms measured across R5. After power up, adjust R26 for minimum DC at The base junction of Q7 and Q10. Be careful not to short anything. One end of R11 will also work for this measurement. Make sure to use the end that is common to the base junctions.
Next, set both R32 and R33 to minimum. Adjust either R32 or R33 for minimum offset at the output. Only one will need to be adjusted. The other is left set to minimum. This is done without a load. After the offset has been set, set the bias either for minimum distortion or at the operating point you feel is optimum.
Check the settings and adjust if needed. Let the amp run with no input and no load for 30 minutes then recheck and readjust.
I've got the voltage gain stage for the 100W (+) version made and it tests fine at rails from +/- 24VDC to +/- 60VDC. The input transistors were Vbe matched to about 1mV using a test jig from J. Graeme's "Applications of Operational Amplifiers." The output was drifting around a bit but I haven't given it much chance to settle. I just put a bit of thermally-conductive epoxy across the diff pair to thermally couple them. Will see how that works.
The TO-220 transistors (Q4 and Q6) get warm - not too hot to touch but enough for a small heatsink.
Phil
The TO-220 transistors (Q4 and Q6) get warm - not too hot to touch but enough for a small heatsink.
Phil
I've now tested the voltage gain stage (VGS) up to +/- 75VDC. I'm using raw DC from a separate power supply, no regulation. I put small pieces of brass stock on Q4 and Q6 for some heatsinking and thermal coupling. I also epoxied the diff pair together. At +/- 75 VDC rails, I get about 44V RMS out at 1 kHz before clipping. The positive side clips a little before the negative peak but clipping appears clean. I'll probably use +/- 66 VDC rails on the VGS for my prototype.
I let the VGS warm up a few minutes then adjusted offset to about 1 mV measured with a Fluke 287 DMM (resolution 0.1 mV). After 30 minutes, it drifted off to about -13 mV and I reset it to less than 1 mV. 30 minutes later (60 minutes total into the test) the offset had drifted less than 2 mV absolute. The circuit was powered down and the experiment was repeated with similar results 6 hours later. These results could be obtained with no signal or with an output voltage of 10V pk-pk at 1 kHz.
The total resistance from the emitter of Q4 (2SA1837) to positive rail is about 29 ohms, close to the 30 ohms Steve Dunlap predicted, with about 12 mA standing current in the VAS.
I let the VGS warm up a few minutes then adjusted offset to about 1 mV measured with a Fluke 287 DMM (resolution 0.1 mV). After 30 minutes, it drifted off to about -13 mV and I reset it to less than 1 mV. 30 minutes later (60 minutes total into the test) the offset had drifted less than 2 mV absolute. The circuit was powered down and the experiment was repeated with similar results 6 hours later. These results could be obtained with no signal or with an output voltage of 10V pk-pk at 1 kHz.
The total resistance from the emitter of Q4 (2SA1837) to positive rail is about 29 ohms, close to the 30 ohms Steve Dunlap predicted, with about 12 mA standing current in the VAS.
Steve --
To paraphrase a former co-worker at Oak Ridge: "Those transistors work for me, I don't work for them!"
But more seriously, you are certainly correct and I hope to post heatsinking suggestions for a 125W or so Krill soon. Please don't report me to the Semiconductor Abuse League just yet.
Phil
To paraphrase a former co-worker at Oak Ridge: "Those transistors work for me, I don't work for them!"
But more seriously, you are certainly correct and I hope to post heatsinking suggestions for a 125W or so Krill soon. Please don't report me to the Semiconductor Abuse League just yet.
Phil
unmibh said:
Hi Rom
I don't want to put words in Phil's mouth - but I might be able to save you a little time here. Phil is using Steve's PWB's and on these boards R9 goes to GND. The schematic that shows R9 going to the negative rail is not correct and needs to be updated (Steve D. is working on that as fast as handicap will let him). I'm unsure what value Steve is using for R9 on the 100W amps tho. Perhaps Steve or Phil will chime in on that. It's still early in New Mexico - the state that Phil resides in - most likely he is still asleep.
Thanks Andrea/c2cThomas,
I just want to make sure we are using the same value and circuit as reference.
Hi Andrea, I'm not too far from you I'm down south Puglia region.
I just ordered some transistor as the shop did not have any left after my last purchase. Hopefully in a week time and I can set up the 100W version.
Regards
Rom
I just want to make sure we are using the same value and circuit as reference.
Hi Andrea, I'm not too far from you I'm down south Puglia region.
I just ordered some transistor as the shop did not have any left after my last purchase. Hopefully in a week time and I can set up the 100W version.
Regards
Rom
Hello Rom --
I'm out of bed now, but maybe not completely awake. I am still working on my first 300 mL of coffee. I'm glad that Andrea and c2cThomas have already answered your question. I am using Steve's printed circuit boards.
I have a few extra comments. I did not have 51k resistors so I used 47.5k resistors which of course is not a significant change. As Andrea said, as long as the current through the diode string is about 1mA you are ok. And I tested the voltage gain stage over a wide range of voltages so I had from 0.5 - 1.5 mA through that diode string and the circuit still worked fine.
If you are not using Steve's boards, I suggest you connect R9 between the resistor string and the negative supply rail and calculate R9 according to your supply voltages to provide 1 mA through the diode string. I will modify my boards to do this once I decide on the supply voltages I will use for the final amplifier.
Good luck with your 100W version! I will start the output stage today so hope to have some results soon.
Phil
I'm out of bed now, but maybe not completely awake. I am still working on my first 300 mL of coffee. I'm glad that Andrea and c2cThomas have already answered your question. I am using Steve's printed circuit boards.
I have a few extra comments. I did not have 51k resistors so I used 47.5k resistors which of course is not a significant change. As Andrea said, as long as the current through the diode string is about 1mA you are ok. And I tested the voltage gain stage over a wide range of voltages so I had from 0.5 - 1.5 mA through that diode string and the circuit still worked fine.
If you are not using Steve's boards, I suggest you connect R9 between the resistor string and the negative supply rail and calculate R9 according to your supply voltages to provide 1 mA through the diode string. I will modify my boards to do this once I decide on the supply voltages I will use for the final amplifier.
Good luck with your 100W version! I will start the output stage today so hope to have some results soon.
Phil
unmibh said:
Be sure to get the polarity right. There are those that believe that having the color bands facing the right direction can produce a subtle but audible difference when used in a DC voltage circuit. For AC circuits you should use a non-polarized type.
Let's see some of those simulator kids play around with that one!
