point 3:I tried your suggestion but the voltage is always 1,5V!!?
point 4 you mean this:
http://www.diyaudio.com/forums/showthread.php?s=&postid=172707#post172707
so you agree that the 2SK170 should be GR and not BL??
Giorgio
point 4 you mean this:
http://www.diyaudio.com/forums/showthread.php?s=&postid=172707#post172707
so you agree that the 2SK170 should be GR and not BL??
Giorgio
Giorgio,
I did have a similar problem. My voltages on the 2SK170 in the MC section is 8.5V, not 15V. But I believe I only have 'BL' versions, which all had a Idss of about 8 mA as I recall. My other voltages check out the same.
I was getting 1.1 to 1.2V across R67 instead of 1V. What I did was to change the R5 from 6K8 to 15K. I'm using low noise HP red LEDs and their optimum is about 2 mA, which is what the 15K for R5 did: lowers their current (and brightness) from about 4.5 mA to 2 mA. The voltage across R67 is now 1.04V.
All voltages except the MC jfets match perfectly.
My one weird finding is that Q21/Q22 (the inverse signal diff pair) runs exceedingly hot. Too hot to touch, about 70°C. I put a large heat sink on them, but it seems strange. (BTW, they were hot when R59 was stock, before I replaced it with a pot to remove the output coupling caps.) Has anyone else noted how warm they are?
One other mod: 30,000 uF were added across each C21 and C29, and 10,000 uF across each C23 and C31. The PS now has about 90,000 uF. I also left off the muting relays. Since I leave the phono stage on 24/7, why bother to worry about a mute feature or whether the contacts will go bad in a several years? If you leave them off, nothing needs to be done since the relays simply shunt the signal to ground and do not directly affect the signal path.
Robert
I did have a similar problem. My voltages on the 2SK170 in the MC section is 8.5V, not 15V. But I believe I only have 'BL' versions, which all had a Idss of about 8 mA as I recall. My other voltages check out the same.
I was getting 1.1 to 1.2V across R67 instead of 1V. What I did was to change the R5 from 6K8 to 15K. I'm using low noise HP red LEDs and their optimum is about 2 mA, which is what the 15K for R5 did: lowers their current (and brightness) from about 4.5 mA to 2 mA. The voltage across R67 is now 1.04V.
All voltages except the MC jfets match perfectly.
My one weird finding is that Q21/Q22 (the inverse signal diff pair) runs exceedingly hot. Too hot to touch, about 70°C. I put a large heat sink on them, but it seems strange. (BTW, they were hot when R59 was stock, before I replaced it with a pot to remove the output coupling caps.) Has anyone else noted how warm they are?
One other mod: 30,000 uF were added across each C21 and C29, and 10,000 uF across each C23 and C31. The PS now has about 90,000 uF. I also left off the muting relays. Since I leave the phono stage on 24/7, why bother to worry about a mute feature or whether the contacts will go bad in a several years? If you leave them off, nothing needs to be done since the relays simply shunt the signal to ground and do not directly affect the signal path.
Robert
giolight said:
YES, for the point 3.
POINT 4:
since the led is a voltage reference for the current sources Q17 and Q16 what is important is the current through the transistor:
if you have a perfect 1 volt voltage across R67 and R53 like the schematic suggests, then the current through the transistors is 1/475= 2.1mA and 1/150= 6.6mA.
If you have a led that lefts 1.4-1.5 volt across the resistors you must recalculate the resistors value for the given current:
1.5/0.0066= 227 ohm
1.5/0.0021= 714 ohm
The Robert's suggestion works only if the voltage is slightly different like in his case 1.1 or 1.2 volts IMHO
...or change led
but it will thake not a long time....and I will stay wise about it.
Thank you Stefano for explanation, I would do as you said, I have one more problem that is in the other channel I have the four 2sk170 with a Vd of 3,2 which is too low and a drop over R45-48 of 200mV (98-100 the other channel) which is too much(10mA) , i changed Q14 and I got Vd 12v but still 200mV over R45-48.
Giorgio
Giorgio
rljones said:
It seems that your idle current through the diffamp Q21/Q22 is really to high. Try to measure voltage over R57 and calculate the current.
I biased my diffamp with 4,5 mA. 2,25mA for each FET (Q21/Q22).
This gives me about 10mA idle current at the output stage (Q8/Q19). Same idle current as used at the output stage (Q7/Q18) of the riaa amp.
The inverter works perfectly and temperature isn't any problem.
toshiba
toshiba,
I measured the voltage across R57 and it is 5.5V. This means the current through Q20 is 5.5V/221R, or 25mA, not 4.5mA.
I was looking at this portion of the schematics and it looks like the same CCS that is used in the Aleph amps, such as the Aleph 5. In the service manual for the Aleph 5, an IRF9610 is used instead of the 9110, but otherwise the values are the same. This manual states that this sets up a bias of 20mA. 9V zener across the gate to the positive rail drops about 4V across the gate, leaving 5V across the 221R resistor, or 20mA (5V/221R). (BTW, I get 22V, rail - 9V, over R68, meaning that the 9.1V zener is set at 4.6mA. Side question: why isn't a cap put in parallel with this zener to reduce noise?)
If we want 4.5mA for the diff pair Q21/22, then R57 should be around 1K2, not 221R, since then the voltage drop will still be 5.5V, but this divided by 1K2 will then give 4.6mA. Did you use a resistor greater than 221R for R57, and do you have around 5V dropped over R57? Since I don't think there are any typos on the schematic, and the listed values should set Q20 at 20mA, how did you decide on 4.5mA?
Even if there is 25mA across Q20, that would mean that there is 30V-5V (PS- R57) times 0.025A, or 0.625W power across Q20. This is within the 1.3W rating of the IRFD9110. I would think this wouldn't be so hot. Q21/22 should each be dissipating 1/2 of this. Is this reasoning incorrect?
Thanks, Robert
I measured the voltage across R57 and it is 5.5V. This means the current through Q20 is 5.5V/221R, or 25mA, not 4.5mA.
I was looking at this portion of the schematics and it looks like the same CCS that is used in the Aleph amps, such as the Aleph 5. In the service manual for the Aleph 5, an IRF9610 is used instead of the 9110, but otherwise the values are the same. This manual states that this sets up a bias of 20mA. 9V zener across the gate to the positive rail drops about 4V across the gate, leaving 5V across the 221R resistor, or 20mA (5V/221R). (BTW, I get 22V, rail - 9V, over R68, meaning that the 9.1V zener is set at 4.6mA. Side question: why isn't a cap put in parallel with this zener to reduce noise?)
If we want 4.5mA for the diff pair Q21/22, then R57 should be around 1K2, not 221R, since then the voltage drop will still be 5.5V, but this divided by 1K2 will then give 4.6mA. Did you use a resistor greater than 221R for R57, and do you have around 5V dropped over R57? Since I don't think there are any typos on the schematic, and the listed values should set Q20 at 20mA, how did you decide on 4.5mA?
Even if there is 25mA across Q20, that would mean that there is 30V-5V (PS- R57) times 0.025A, or 0.625W power across Q20. This is within the 1.3W rating of the IRFD9110. I would think this wouldn't be so hot. Q21/22 should each be dissipating 1/2 of this. Is this reasoning incorrect?
Thanks, Robert
I had time to work on the board this morning. I replaced R57 with a 499R. Voltage drop across this is now 5.7V on each channel, or about 12mA for the diff pair. Voltage drop across Q20 measures at 22.5V, meaning it now dissipates about 270mW (rated at 1300mW). Q21 and Q22 are now running at 6mA each and have voltage drops (measured) of 29.5 and 34V, respectively. Q22 is thus dissipating around 200mW. All are now warm, but not hot, to touch.
Voltage drop across R65 is still around 0.6V (I used 49R9 instead of 47R for R65), meaning the inverse output stage is still running at 12mA. So all seems in order; the only significant deviation is changing R57 from 221R to 499R to avoid excessive heat. (This also means the formula or rationale from the previous post for how the CCS works, makes sense.)
Regards, Robert
Voltage drop across R65 is still around 0.6V (I used 49R9 instead of 47R for R65), meaning the inverse output stage is still running at 12mA. So all seems in order; the only significant deviation is changing R57 from 221R to 499R to avoid excessive heat. (This also means the formula or rationale from the previous post for how the CCS works, makes sense.)
Regards, Robert
Hi Robert,
sorry for late answer, but at the moment to much work.
I'm a friend of low bias current in the frist gain stage!
You are right. I don't work with the original values of the schematics.
For the inverter I use 6,8V Zener, R57=620 and R59=2.21k.
Alltough I have placed a small compensation capacitor over R62.
Value is circuit independent. Scope the output with square wave and examine the value.
To achieve low offset for bypassing C9, it is inevitable, that the drain currents of Q21 and Q22 are exactly the same. You can achieve that by varying the values of R57 or R59.
toshiba
sorry for late answer, but at the moment to much work.
I'm a friend of low bias current in the frist gain stage!
You are right. I don't work with the original values of the schematics.
For the inverter I use 6,8V Zener, R57=620 and R59=2.21k.
Alltough I have placed a small compensation capacitor over R62.
Value is circuit independent. Scope the output with square wave and examine the value.
To achieve low offset for bypassing C9, it is inevitable, that the drain currents of Q21 and Q22 are exactly the same. You can achieve that by varying the values of R57 or R59.
toshiba
TheMyxin,
its good to hear we have similar readings.
i too am not presently using the inverting side, but plan on it soon. my main concern is in maximizing longevity of the parts. i thought it too warm (i also believe it is about 70°C, with some readings slightly greater)--even with a heat sink. and since no heat sink was mentioned, i got to worrying about leaving it on (i run my electronics 24/7). mr. pass warms about parts running too hot not surviving for many years, and i'd just as soon not worry about stressing them.
regards, robert
its good to hear we have similar readings.
i too am not presently using the inverting side, but plan on it soon. my main concern is in maximizing longevity of the parts. i thought it too warm (i also believe it is about 70°C, with some readings slightly greater)--even with a heat sink. and since no heat sink was mentioned, i got to worrying about leaving it on (i run my electronics 24/7). mr. pass warms about parts running too hot not surviving for many years, and i'd just as soon not worry about stressing them.
regards, robert
On the emitter of Q24 I have 27V which seem to be correct.
If I leave off Q14 on R18 I have exact half of it but if I put back Q14 I have a drop of 20V on it and the rest on R16.
Vd Q10-13 start at 1,5V and in some second stabilize on 6,2V.
The other channel goes directly to 12,7V and don't move from it.
Giorgio
If I leave off Q14 on R18 I have exact half of it but if I put back Q14 I have a drop of 20V on it and the rest on R16.
Vd Q10-13 start at 1,5V and in some second stabilize on 6,2V.
The other channel goes directly to 12,7V and don't move from it.
Giorgio
@giolight
Measure dc-voltage over R31 ( gate-voltage of Q10-Q13). Result should be nearly zero.
In case of zero :
Are Q10-Q13 BL-Types or do you use V-Type ? Anyway, it seems that Q10-Q13 of the wrong channel have very high IDSSmax.
To get the preamp work, increase R45-R48 ( test it with a value between 100-200 ohm ). IMO sum of drain current should be about 8-10mA.
Another way to check the general function : Short R66 with jumper and halve the value of R40. The cascode Q14 should reach stable values. Anyway, 40mA is really to high.
In case of non zero and correct placement of the fets :
Locate a possible source, shorten traces etc. which supplies the gates.
toshiba
Measure dc-voltage over R31 ( gate-voltage of Q10-Q13). Result should be nearly zero.
In case of zero :
Are Q10-Q13 BL-Types or do you use V-Type ? Anyway, it seems that Q10-Q13 of the wrong channel have very high IDSSmax.
To get the preamp work, increase R45-R48 ( test it with a value between 100-200 ohm ). IMO sum of drain current should be about 8-10mA.
Another way to check the general function : Short R66 with jumper and halve the value of R40. The cascode Q14 should reach stable values. Anyway, 40mA is really to high.
In case of non zero and correct placement of the fets :
Locate a possible source, shorten traces etc. which supplies the gates.
toshiba
Hi all,
after carefully board ceck I found R37 unconnected to R38, I don't know how that happened, rejointed and everything went ok.
I pretty don't understand what happened but that' s it.
I use 2SK170BL I have some GR and even 2SK369V could be interesting to try.
Thank you for help.
Giorgio
after carefully board ceck I found R37 unconnected to R38, I don't know how that happened, rejointed and everything went ok.
I pretty don't understand what happened but that' s it.
I use 2SK170BL I have some GR and even 2SK369V could be interesting to try.
Thank you for help.
Giorgio
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.