Hi Thimios,
I have been chasing my tail all day with this new modded CFA. Checked everything I can think of. I had matched hFe on all transistors except for Q3/4 and Q7/10. I have about a .5V offset. I was able to dial it out with the pots but it took a lot of adjustment after which the bias on the VAS seems quite high. Square waves look good and it plays music but there is something still not right. I'm posting a schematic with voltages in hopes that another set of eyes might help find the issue. Your success have given me hope so I have not given up.q I did change out Q3/4 with hfe matched devices with no change.
Blessings, Terry
Attachments
Last edited:
Something wrong just at input. Voltage at Q1,Q23 base +0,526, voltage at Q2 emitor -1,111, so Ube Q2 is 1,63V- faulty Q2
R14 current shold be about 1,8mA (40V/22k), but voltage at emitor Q1 is 50mV , so Q1 must by closed-shorted C7, C14?? And voltage difference across R28, R27 is 0V, so no VAS current?
R14 current shold be about 1,8mA (40V/22k), but voltage at emitor Q1 is 50mV , so Q1 must by closed-shorted C7, C14?? And voltage difference across R28, R27 is 0V, so no VAS current?
Last edited:
OK , some mods "bring something to the table" (some do not).
(in reference to BELOW)
C7/C8 +R8 = the major part of the PSRR gain. (OPS R114/115 @ 10R adds 3db)
CM1/2 fed back to Q1/2 emitters = 1/4 thd , THD stays <10ppm right
up to clipping - Dual MIC is gut
.
Extra resistors (R13/14) , if too high ... reduce the CCS's ability to trim
the offset - 2.2k seems to work well (no THD rise).
Original 12-15V zener low voltage CCS's add a whole 6db PSRR (I added them
again) ... also no THD rise. Edit- red led Vf stays much more constant w/ a regulated supply.
I raised the gain of the hawksford/CM to push me into single digit PPM (R15/16- 20/21) ... CLG rise.
Could see no benefit to resistors between input pair and Hawksford/CM (10K's) ????
So , I now have 8-10ppm at high levels , clips good , awesome pssr ...
and it is only 2 cut traces from the original. It also offsets like the original ,
CCS trimmers are very responsive (10-20R = many 10's of mv offset).
Slew is master (...ful - 220V/us+)
Had a lot of trouble even getting the full modded sim to mv level.
Feel free to mod either way - this would be my way.
OS
(in reference to BELOW)
C7/C8 +R8 = the major part of the PSRR gain. (OPS R114/115 @ 10R adds 3db)
CM1/2 fed back to Q1/2 emitters = 1/4 thd , THD stays <10ppm right
up to clipping - Dual MIC is gut
.Extra resistors (R13/14) , if too high ... reduce the CCS's ability to trim
the offset - 2.2k seems to work well (no THD rise).
Original 12-15V zener low voltage CCS's add a whole 6db PSRR (I added them
again) ... also no THD rise. Edit- red led Vf stays much more constant w/ a regulated supply.
I raised the gain of the hawksford/CM to push me into single digit PPM (R15/16- 20/21) ... CLG rise.
Could see no benefit to resistors between input pair and Hawksford/CM (10K's) ????
So , I now have 8-10ppm at high levels , clips good , awesome pssr ...
and it is only 2 cut traces from the original. It also offsets like the original ,
CCS trimmers are very responsive (10-20R = many 10's of mv offset).
Slew is master (...ful - 220V/us+)
Had a lot of trouble even getting the full modded sim to mv level.
Feel free to mod either way - this would be my way.
OS
Attachments
Last edited:
C7/C8 +R8 = the major part of the PSRR gain. (OPS R114/115 @ 10R adds 3db)OK , some mods "bring something to the table" (some do not).
(in reference to BELOW)
C7/C8 +R8 = the major part of the PSRR gain. (OPS R114/115 @ 10R adds 3db)
CM1/2 fed back to Q1/2 emitters = 1/4 thd , THD stays <10ppm right
up to clipping - Dual MIC is gut
Extra resistors (R13/14) , if too high ... reduce the CCS's ability to trim
the offset - 2.2k seems to work well (no THD rise).
Original 12-15V zener low voltage CCS's add a whole 6db PSRR (I added them
again) ... also no THD rise. Edit- red led Vf stays much more constant w/ a regulated supply.
I raised the gain of the hawksford/CM to push me into single digit PPM (R15/16- 20/21) ... CLG rise.
Could see no benefit to resistors between input pair and Hawksford/CM (10K's) ????
So , I now have 8-10ppm at high levels , clips good , awesome pssr ...
and it is only 2 cut traces from the original. It also offsets like the original ,
CCS trimmers are very responsive (10-20R = many 10's of mv offset).
Slew is master (...ful - 220V/us+)
Had a lot of trouble even getting the full modded sim to mv level.
Feel free to mod either way - this would be my way.
OS
I can't see these R114-115
Hi BV,
You made me curious about Q1 & Q2 so I just went out and measured them Both have .578 vbe so I think they are ok. It is 11:20PM here so I am done for the day. I will take new measurements in the morning. When I took those measurement above the IPS was not attached to the OPS. It is now and I just checked a few things and noticed that the - rail is pulled down to something like -64V while the + rail is at +70V. Pot R5 has been adjusted to make the output 0 and I believe that is what is pulling the -rail low. Should I take measurements with the output at nil or should I take them with the pots equalized? I will also measure the vbe of each transistor and get the ma of each resistor.
Thanks, Terry
Terry did you try adjusting the trimers in the opposite rotation each other for offset adjustment?
Last edited:
saturation ....
Yes , short pulses of 8-10ma on Q1/2 -6/9.
An analogy would be the beta enhancer on the "blameless" , 220R is enough
to mitigate this ...
But the cascode still clips nicely.
Those pulses won't exceed Ic/soa for the ksa/c's.
PS - if they are going to "overuse" this amp , solder in a ksc2690/ksa1220
pair for input , easier than adding additional components .
Keep it simple , too many parts "slows things down" . (figure that out).
I DID find out why there is "no free lunch" ( reduced slew).
OS
Yes , short pulses of 8-10ma on Q1/2 -6/9.
An analogy would be the beta enhancer on the "blameless" , 220R is enough
to mitigate this ...
But the cascode still clips nicely.
Those pulses won't exceed Ic/soa for the ksa/c's.
PS - if they are going to "overuse" this amp , solder in a ksc2690/ksa1220
pair for input , easier than adding additional components .
Keep it simple , too many parts "slows things down" . (figure that out).
I DID find out why there is "no free lunch" ( reduced slew).
OS
Yes I know the math shows that. One of the problems of just measuring voltages. Touching the probe will at times cause voltage to move around a little. Tonight I measured across b-e and it was .578 on each. I will make a more sensible schematic tomorrow.
OS, I see minimum of 4 trace cuts. See attached.
Attachments
Last edited:
