This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
This is the amp I have built.
I think I got rid of the noice but we will se in the future.
I have another question.
Since I use +/-58 Volts for supply (transformer 40 Volts) I get around 108 Volts across Q8 base emitter. And with 35mA going through Q8 it gets quite hot. Heatsink will get about 50C. I was wondering if I could change the values on R5 and R6 to have more voltage over R5 and R6 instead of Q8?
If I put R5=15k and R6=470 I get about 78 volts over Q8 instead. I guess that would lower the heat some.
It would be ok if I just keep the ratio between R5 and R6 the same?
Through R7 and R8 it goes about 1mA each.

And also I heard something about a servo for the feedback. Or was that for the DC level to the mosfets?
The mosfets are 1058 and 162.

Glad for help
Q8, 108 Volts?? I believe it's half the voltage. Do you have zero output voltage?

It's good to have good current capability to drive the fet's. The load is approx. 1,5-2 nF unlinear. You can calculate slew rate as I/C = V/s

You should have at least 15 mA. One way to save power consumption is to have a buffer between the high gain stage and the fet's.

If you want a DC-servo, there are at least 3 easy ways to implement it.

1 Inject current into the minus-input
2 Inject current into the plus-input (can be tricky
3 Create unbalance in the first diff stage by injecting current into (or from) the collector loads.

If you check my monster phono amp (see link below) you can see how I have done. I have servo model 1 (RIAA stage) and 3 (pre amp).

Heatsink temp at 50 degrees C is high but far from dangerous.

If you have noise (and feedback) you probely have some problems with oscillation. The gain is only 22. You should have approx. 22-50 µV noise at the output. The S/N would be at least 100 dB.

If you check my monster headphone amp, you can see that I have 0,63 µV input noise with ordinary BC-transistors. You should have 1-3 µV I think.
Hi peranders

Yeah sorry its half that voltage, but they still get pretty hot =(
I know 50 C is not that hot for silicon but would be nice if could be a little cooler.
The current mirrors work so that if Q3 sets a current of 2mA then R7 and R8 has 1mA each and that would give R5 and R6 a ratio of 11. So if R5 is 2200 and R6 is 68 that is 2200/68=32 so about 32mA will go through Q8, and I guess that is enaough to drive one mosfet on each rail. I wonder what would happen if I like put 4400 and 136 for R5 and R6 instead. That would give the same ratio so the same current. But maybe there is something Im missing. I know more current gives better noice rejection or so.

Thanks for the info about DC servos

The circuit shown is the Borbely 60W amplifier that was offered in TAA 1982. I know, I have built two stereo and one mono block amplifiers. And the transistors do get hot. I have been using my 60W amps on and off since the late '80s and I have not had any problems. Also I had raised the bias of the output mosfets to 170 Ma. and still no problems. And if I remember correctly the bias on the transistors you talking about is suppose to be about 30 Ma. cause Mr. Borbely believes this value is to keep the maximum linearity to drive two mosfets in the output stage. You should contact Mr. Borbely if you have any questions at I deal with him all the time and he always has time to send me an e-mail if I have questions about his designs.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.