I have assembled a Servo-amplifier and there's a slight difference between channels.
It sounds fantastic, so whatever is going on one can't hear it.
I use a 2x40 VAC 2x600 VA transformer for OPS and a 2x12 VAC 1A for driver PCB, and high voltage rails are in series with OPS rails, but with it's own CRC regulation.
Then the regulators leave +/- 60V for driverstage, OPS have +/- 53V with 940mA bias / channel.
Instead of 2N5401/5551 I use KSA992 and KSC1845, and as predrivers 2SA1480 and 2SC3790 instead of 2SA1210 and 2SC2912.
Outputstage use 2SJ79 / 2SK216 as drivers and 6 pairs 2SJ55 / 2SK175.
IPS current is set to 2mA, there's 5mA / 18mA on VAS stage and 28mA on mosfet drivers.
Now to the problem...
One channel have slightly more gain, about 1V from 10 Vrms and up to clip level at 31,5 Vrms (Bad channel) and 32,5 Vrms (good channel) output in 3,9 ohms. When it clips the "bad" channel starts clipping at negative side, positive side clips the same as the good channel.
Distortion at 1 kHz and 20 Vrms out is 0,0035% / 0,0060%, 10 kHz 0,0190% / 0,0496%, and at 20 kHz 0,0364% / 0,0902%, so the "bad" channel looks a bit worse. At 30Vrms it looks about the same, but a bit higher.
I have not been able to measure any differences before VAS-stage, but the difference appears on VAS output.
So is it Vbe or Hfe depending, or may one expect a faulty transistor ?
Best regards.
Figge
It sounds fantastic, so whatever is going on one can't hear it.
I use a 2x40 VAC 2x600 VA transformer for OPS and a 2x12 VAC 1A for driver PCB, and high voltage rails are in series with OPS rails, but with it's own CRC regulation.
Then the regulators leave +/- 60V for driverstage, OPS have +/- 53V with 940mA bias / channel.
Instead of 2N5401/5551 I use KSA992 and KSC1845, and as predrivers 2SA1480 and 2SC3790 instead of 2SA1210 and 2SC2912.
Outputstage use 2SJ79 / 2SK216 as drivers and 6 pairs 2SJ55 / 2SK175.
IPS current is set to 2mA, there's 5mA / 18mA on VAS stage and 28mA on mosfet drivers.
Now to the problem...
One channel have slightly more gain, about 1V from 10 Vrms and up to clip level at 31,5 Vrms (Bad channel) and 32,5 Vrms (good channel) output in 3,9 ohms. When it clips the "bad" channel starts clipping at negative side, positive side clips the same as the good channel.
Distortion at 1 kHz and 20 Vrms out is 0,0035% / 0,0060%, 10 kHz 0,0190% / 0,0496%, and at 20 kHz 0,0364% / 0,0902%, so the "bad" channel looks a bit worse. At 30Vrms it looks about the same, but a bit higher.
I have not been able to measure any differences before VAS-stage, but the difference appears on VAS output.
So is it Vbe or Hfe depending, or may one expect a faulty transistor ?
Best regards.
Figge
Check R22 in the bad channel, might be misplaced with a 10k io 100k
A(R22=100k) = 11.1x (bal) - 20.2x (unb)
A(R22=10k) = 12.0x (bal) - 21.1x (unb)
It's a slight difference, but does account.
A(R22=100k) = 11.1x (bal) - 20.2x (unb)
A(R22=10k) = 12.0x (bal) - 21.1x (unb)
It's a slight difference, but does account.
Thank's. I will check that.
Well, you know that one thinks one have doubblechecked everything, but anyway there's always something...
I'll get back in a couple of days.
BR
Figge
Well, you know that one thinks one have doubblechecked everything, but anyway there's always something...
I'll get back in a couple of days.
BR
Figge
Now I have gone thru the amp.
All components are ok, but I found out that one rail were a bit lower and the LM336 cascode regs were between 4v2 and 6v4. Looks like they had drifted. So I pulled the servo OP-amps and started.
Adjusted rails, cascodes, and Iq on IPS.
Measured AC out and it's exactly the same.
BR
Figge
All components are ok, but I found out that one rail were a bit lower and the LM336 cascode regs were between 4v2 and 6v4. Looks like they had drifted. So I pulled the servo OP-amps and started.
Adjusted rails, cascodes, and Iq on IPS.
Measured AC out and it's exactly the same.
BR
Figge
The current from and to D2 and D3 is running trough P2-R13. P2 is a little risk in the design: if it starts giving problems, it will cause unwanted effects in the settings and signal.
Check C4-5 for identical values and P10-11 for identical Vdc values between the A+ and K- nodes. The absolute values over the P2-R13 combo will be different, as this results from different Vgs's of the K170-J74 diff's.
SW1b might cause problems too. To be checked. Compare R19 between the channels.
Check C4-5 for identical values and P10-11 for identical Vdc values between the A+ and K- nodes. The absolute values over the P2-R13 combo will be different, as this results from different Vgs's of the K170-J74 diff's.
SW1b might cause problems too. To be checked. Compare R19 between the channels.
I have readjusted everything again.
The LM336's are very sensitive to adjust.
Potentiometers are Vishay, so no crap.
I also notised that the rail regulators are almost as sensitive to adjust.
Anyway, after readjusting both channels they measure the same, and puts out 32Vrms into 3R9 dummyloads, both channels driven. Distortion measures -100dB. S/N measures 100 dB to.
Listened to different music yesterday and today, and I must say that it sounds fantastic driving my backloaded horns with 2x Beyma 8G40 and a CP380.
BR
Figge
The LM336's are very sensitive to adjust.
Potentiometers are Vishay, so no crap.
I also notised that the rail regulators are almost as sensitive to adjust.
Anyway, after readjusting both channels they measure the same, and puts out 32Vrms into 3R9 dummyloads, both channels driven. Distortion measures -100dB. S/N measures 100 dB to.
Listened to different music yesterday and today, and I must say that it sounds fantastic driving my backloaded horns with 2x Beyma 8G40 and a CP380.
BR
Figge