OK... so we have to try and get a handle on whats going on.
Meter lights are an obvious target to look at. You say you have 40 volts approx (per rail) and that the soft start relay operates. The question has to be "are the bulbs originals" ? If they are of a higher current rating (and thus lower resistance) then the relay would work but the bulbs would not see enough voltage to light. I cant see any other reason for them not lighting (assuming the series resistor chain is OK).
No click from the speaker relay... check the DC voltage at the collectors of the output transistors on both channels. They should be around zero give or take a few tens of millivolts.
Meter lights are an obvious target to look at. You say you have 40 volts approx (per rail) and that the soft start relay operates. The question has to be "are the bulbs originals" ? If they are of a higher current rating (and thus lower resistance) then the relay would work but the bulbs would not see enough voltage to light. I cant see any other reason for them not lighting (assuming the series resistor chain is OK).
No click from the speaker relay... check the DC voltage at the collectors of the output transistors on both channels. They should be around zero give or take a few tens of millivolts.
Well the bulbs are a minor issue really (at the moment) but yes, definitely worth a measure of them.
The voltages seem reasonable although 74 mv while not overly high in itself is perhaps a bit higher than expected but not enough to trip the protection.
Work backwards from the relay driver Q407 measuring the voltages. Q407 should be on (0.7 across B and E) Is Q403 and/or Q402 pulling it down ? Worth some voltage checks around there. If nothing conclusive then if you remove both those transistors does the relay operate OK ?
Q404 and 405 both have to be OFF as well. Again careful measurement.
The voltages seem reasonable although 74 mv while not overly high in itself is perhaps a bit higher than expected but not enough to trip the protection.
Work backwards from the relay driver Q407 measuring the voltages. Q407 should be on (0.7 across B and E) Is Q403 and/or Q402 pulling it down ? Worth some voltage checks around there. If nothing conclusive then if you remove both those transistors does the relay operate OK ?
Q404 and 405 both have to be OFF as well. Again careful measurement.
The voltage on the light bulb's is 0.6V, guess that explains why they don't light up.
Resistance is 2.2 Ohm (two bulb's in parallel)
Using the variable lab supply they light up at normal level at about 5-6V.
I'll try get those reading on the speaker relay transistors, hopefully later today.
Maybe I could try a old test speaker on the input side of the relay, just to hear if it's alive?
Resistance is 2.2 Ohm (two bulb's in parallel)
Using the variable lab supply they light up at normal level at about 5-6V.
I'll try get those reading on the speaker relay transistors, hopefully later today.
Maybe I could try a old test speaker on the input side of the relay, just to hear if it's alive?
Cold resistance of bulbs can be very low, the filament increases rapidly in resistance as it heats. Worth checking each bulb separately (one might have been changed) but as its a series circuit there's not much beside the bulbs tbh.
Yes, you could dangle a speaker off the amp output to test it. I would always add a series resistor at first (say 100 ohm) but there's no offset so that should be fine.
Yes, you could dangle a speaker off the amp output to test it. I would always add a series resistor at first (say 100 ohm) but there's no offset so that should be fine.
Q407 B and E is 590mV.
C on Q407 is 40V, connecting C to ground does make the relay click.
C on Q402/3 is 1.17V
C on Q406 is 33.3V, should be 0.9V.
C on Q401 is -3.7V
B on Q404 is 1V
B on Q405 is 94mV
1V on Q404's base that's strange, DC is still the same, D403 maybe?
Wondering if 40-0-40V was too low, I tried to bypass the 100W light bulb, rails went up to 54-0-54V, still no light in meters, speaker relay still not clicking.
All measurement was done whit the 100W bulb in circuit.
C on Q407 is 40V, connecting C to ground does make the relay click.
C on Q402/3 is 1.17V
C on Q406 is 33.3V, should be 0.9V.
C on Q401 is -3.7V
B on Q404 is 1V
B on Q405 is 94mV
1V on Q404's base that's strange, DC is still the same, D403 maybe?
Wondering if 40-0-40V was too low, I tried to bypass the 100W light bulb, rails went up to 54-0-54V, still no light in meters, speaker relay still not clicking.
All measurement was done whit the 100W bulb in circuit.
It wasn't removing Q402/3 that did it, it was me testing the amp module outside the chassis.
When the heatsink touch the chassis, the relay don't click, no matter if Q402/3 is in place or not.
When it don't touch the chassis, the relay clicks, even whit Q402/3 in place.
Using the headphone jack, I hear a low level 50Hz hum, but that could be due to the missing ground.
Edit:
The measurement in post #586 was taken whit the power amp module in the chassis.
When the heatsink touch the chassis, the relay don't click, no matter if Q402/3 is in place or not.
When it don't touch the chassis, the relay clicks, even whit Q402/3 in place.
Using the headphone jack, I hear a low level 50Hz hum, but that could be due to the missing ground.
Edit:
The measurement in post #586 was taken whit the power amp module in the chassis.
Last edited:
OK...
I think the only way to work on this is with the amp built up correctly (but still using your linear PSU). We have to look at all the voltages from the relay working back to see whats going on.
So Q407 C should be zero volts (and of course if it were the relay would operate)
Q407 B to E should be 0.7 (fully on)
Q406 C as marked or even lower at 0.9 volts
Q406 B at around 1.4 volts.
If those conditions are met then the relay should operate.
If it doesn't then we need to measure the B-E volts of Q402 and Q403. Both should be below 0.3 volts and perhaps even slightly negative for Q402.
The B-E voltage of Q404 should also be below around 0.1 volts.
See if there are any obvious standout differences in those voltages when the amp is all screwed together correctly and all grounds etc are correctly in contact with each other.
I think the only way to work on this is with the amp built up correctly (but still using your linear PSU). We have to look at all the voltages from the relay working back to see whats going on.
So Q407 C should be zero volts (and of course if it were the relay would operate)
Q407 B to E should be 0.7 (fully on)
Q406 C as marked or even lower at 0.9 volts
Q406 B at around 1.4 volts.
If those conditions are met then the relay should operate.
If it doesn't then we need to measure the B-E volts of Q402 and Q403. Both should be below 0.3 volts and perhaps even slightly negative for Q402.
The B-E voltage of Q404 should also be below around 0.1 volts.
See if there are any obvious standout differences in those voltages when the amp is all screwed together correctly and all grounds etc are correctly in contact with each other.
I agree, the only reason I powered it up outside the chassis was that the PCB is connected to the chassis by a black wire screwed direct on the bottom of the chassis.
It's 0.59V.
It's way off, 33.3V
It's 1.17V (according to the manual it should be 1.2V, but measuring whitout the chassis ground, the relay operates when it reaches 1.28V)
B-E Q402 is -3.6V, B-E Q403 is -7V
Yep, spot on 0.1V
Well Q406 is way off, but I don't think that's the real problem, it'll come down when the 1.17V on it's B comes up.
I adjusted the DC offset down from the 71mV to -8mV, it did not change anything, the relay is still not operating.
I'm trying to learn as we go along, so I hope you don't mind some stupid questions
Q403 is marked "muting" in the manual, is it the ca. 10 sec. delay from power on to the relay operate?
Last edited:
Removed Q404 and Q405, no change.
Then tried removing Q401 (Q404 and 405 back in place) still no change.
Is it possible that 40-0-40V rail is just too low after all?
I did try bypass the 100W bulb, but perhaps not long time enough, the rail was 55-0-55V so i might have turned it off to quickly, as some cap's are only rated for 50V.
Then tried removing Q401 (Q404 and 405 back in place) still no change.
Is it possible that 40-0-40V rail is just too low after all?
I did try bypass the 100W bulb, but perhaps not long time enough, the rail was 55-0-55V so i might have turned it off to quickly, as some cap's are only rated for 50V.
Last edited:
Hmmm... those readings aren't quite as conclusive as I'd hoped. No matter
Lets go through how it works and hopefully that will reveal the problem.
Q407 is used as a switch to drive the relay. There should be no "indeterminate" state. Its either fully on or fully off. Now this is where the 0.59 ? base to emitter voltage is slightly "indeterminate". It seems as if Q406 is passing a tiny current, not enough to turn on Q407 but enough to bring the volts across R414 up a little.
Q406 collector at 33 volts indicates that Q406 is essentially OFF and it should be ON. When Q406 is on current flows through R413 directly into the base of Q407 turning it on and clicking the relay.
So we look at the base of Q406 to see what is happening. R412 and R410 form a voltage divider and R411 supplies base current from this reference voltage to Q406. Q406 and Q407 are connected as a "Darlington pair" which essentially make one super high gain transistor.
Again the base volts of Q406 is a bit indeterminate unfortunately. You should see around 1.4 volts there. That's the combined base/emitter voltages of the two transistors.
So there are a few possibilities. Is C404 leaky and pulling the voltage down ? With no speakers connected you can just remove that cap and see if the relay clicks on instantly. C404 charges via R411 and its that time constant that gives the switch on delay. When the voltage on C404 reaches the magic 1.4 volts or so (the two B-E volt drops) then the transistors conduct and the relay clicks in.
Its possible one of the transistors is faulty but that wouldn't be my first suspect.
Q404 and Q405 can also pull this voltage down. These are the DC offset protection and when they conduct they instantly pull the voltage on C404 down and turn off the relay.
Here's what to do, and there are a lot of ways to attack this
1. Check C404 (remove it and with no speakers connected see if it all works)
2. If that shows nothing then measure the voltage across C403. It should be zero. If it isn't then there is a problem in the offset protection part of the circuit. If it is zero and the relay still doesn't operate then remove Q404 and Q405 and retest.
3. If that still doesn't work then to be 100% sure remove also Q402 and Q403 and retest. That then leaves just the relay drivers and associated components. If its still not working then we are looking at a low gain/leaky transistor, either Q406 or Q407.
Lets go through how it works and hopefully that will reveal the problem.
Q407 is used as a switch to drive the relay. There should be no "indeterminate" state. Its either fully on or fully off. Now this is where the 0.59 ? base to emitter voltage is slightly "indeterminate". It seems as if Q406 is passing a tiny current, not enough to turn on Q407 but enough to bring the volts across R414 up a little.
Q406 collector at 33 volts indicates that Q406 is essentially OFF and it should be ON. When Q406 is on current flows through R413 directly into the base of Q407 turning it on and clicking the relay.
So we look at the base of Q406 to see what is happening. R412 and R410 form a voltage divider and R411 supplies base current from this reference voltage to Q406. Q406 and Q407 are connected as a "Darlington pair" which essentially make one super high gain transistor.
Again the base volts of Q406 is a bit indeterminate unfortunately. You should see around 1.4 volts there. That's the combined base/emitter voltages of the two transistors.
So there are a few possibilities. Is C404 leaky and pulling the voltage down ? With no speakers connected you can just remove that cap and see if the relay clicks on instantly. C404 charges via R411 and its that time constant that gives the switch on delay. When the voltage on C404 reaches the magic 1.4 volts or so (the two B-E volt drops) then the transistors conduct and the relay clicks in.
Its possible one of the transistors is faulty but that wouldn't be my first suspect.
Q404 and Q405 can also pull this voltage down. These are the DC offset protection and when they conduct they instantly pull the voltage on C404 down and turn off the relay.
Here's what to do, and there are a lot of ways to attack this
1. Check C404 (remove it and with no speakers connected see if it all works)
2. If that shows nothing then measure the voltage across C403. It should be zero. If it isn't then there is a problem in the offset protection part of the circuit. If it is zero and the relay still doesn't operate then remove Q404 and Q405 and retest.
3. If that still doesn't work then to be 100% sure remove also Q402 and Q403 and retest. That then leaves just the relay drivers and associated components. If its still not working then we are looking at a low gain/leaky transistor, either Q406 or Q407.
Thanks for the walk through, is nice to know how it works, rather than just replace in blind
The last thing i tried last night was to raise the rails a bit whit another bulb in parallel to the 100W bulb, that made the rail go up to 44-0-44V and now the relay clicks after 14 sec.
The amp plays, the speaker is a old PC speaker, but it sounds like both pre and power amp are working as they should
Unfortunately my ISP decided to do some maintenance at that time, so I couldn't post it here
14 sec. for the relay to click is a bit more than normal, could it be C404 being leaky?
And the same leaky C404 could be the reason that 40-0-40V wasn't enough?
Anyway the plan is to replace all cap's, but I'd like to hear it play first and get to know it's sound, so I can tell if recaping was worth it.
Even the Wattmeters are working fine, but there's still no light, R804 (150 Ohm/7W) get pretty hot, too hot to touch, but not burning hot, my guess is 60-70 C, not sure if that's related to the "no meter light problem"
The last thing i tried last night was to raise the rails a bit whit another bulb in parallel to the 100W bulb, that made the rail go up to 44-0-44V and now the relay clicks after 14 sec.
The amp plays, the speaker is a old PC speaker, but it sounds like both pre and power amp are working as they should
Unfortunately my ISP decided to do some maintenance at that time, so I couldn't post it here
14 sec. for the relay to click is a bit more than normal, could it be C404 being leaky?
And the same leaky C404 could be the reason that 40-0-40V wasn't enough?
Anyway the plan is to replace all cap's, but I'd like to hear it play first and get to know it's sound, so I can tell if recaping was worth it.
Even the Wattmeters are working fine, but there's still no light, R804 (150 Ohm/7W) get pretty hot, too hot to touch, but not burning hot, my guess is 60-70 C, not sure if that's related to the "no meter light problem"
Great to hear you have it working and interesting too So the rails did need to be a bit higher and that does explain the "indeterminate" or half way house readings. I never thought it would be so critical on that but I can see now why. Its R410 at 2.2K (the lower part of the voltage divider). There wasn't quite enough voltage to turn on the transistors.
14 seconds does sound a long time but we could do some maths and see if its in the right ball park. I'd have to draw it out.
The bulbs are odd and I still can't see any other possible reason other than incorrect bulbs being fitted. Its a pure series circuit feeding them. Are R803 and R805 OK
So the rails did need to be a bit higher and that does explain the "indeterminate" or half way house readings. I never thought it would be so critical on that but I can see now why. Its R410 at 2.2K (the lower part of the voltage divider). There wasn't quite enough voltage to turn on the transistors. 14 seconds does sound a long time but we could do some maths and see if its in the right ball park. I'd have to draw it out.
The bulbs are odd and I still can't see any other possible reason other than incorrect bulbs being fitted. Its a pure series circuit feeding them. Are R803 and R805 OK
14 seconds is too long... 100uf and 35k (ish... its complicated by the fact the cap is fed from a divider) should give a delay of around 4 seconds but that delay can be overridden by Q404/405 and Q402/403
The voltage on C404 rises exponentially as it charges and the two relay driver transistor base/emitter junctions "clamp" the cap voltage at around 1.4 volts.
Change the cap first
The voltage on C404 rises exponentially as it charges and the two relay driver transistor base/emitter junctions "clamp" the cap voltage at around 1.4 volts.
Change the cap first
Maybe the time will go down as the rails goes up, it's still 5V short.
Anyway the cap will be changed
Since using the amp as it is now, is a bad idea, the rails will likely collapse when current gets too high, and the bulb starts to light up, I guess I should focus on the PSU.
Reading how John's PSU died after a few months doesn't makes me optimistic.
http://www.diyaudio.com/forums/solid-state/220179-faulty-sony-ta-f6b.html#post3175922
and
http://www.diyaudio.com/forums/solid-state/220179-faulty-sony-ta-f6b-2.html#post3178125
I'm pretty sure those 2023's I ordered are fake, the SANKEN logo doesn't look right, the legs are thinner, they weigh less than the originals (2.6 grams vs. 1.9 grams).
What I've done so far on the PSU is a complete recap, tested all transistors and resistors, all OK except for the blown 2023R's.
The linear regulator works when fitted whit 2 original 2023R's
I have 4 good original 2023R's
So my options are:
1) use them in the switching section and find something new for the linear regulator.
2) use two original 2023R's in the linear reg. and find something new for the switching section.
3) find something new for both sections.
4) ?
As the remaining 2023R's have bin stressed 3) or 4) will likely be the best option.
I asked ON semi for a replacement, they came up whit this:
https://www.onsemi.com/PowerSolutio...=EE&category=EE6&question=4680&includedC=true
Searching mouser for "2sc2023" gives this result:
NTE394 NTE | Mouser
But It's not available in my region
In this post
http://www.diyaudio.com/forums/solid-state/181121-sony-ta-f6b-psu-repair-58.html#post3622944
amc184 uses FSC2335 but we don't know how that turned out yet.
You (or was it Jaycee) mentioned MJE13007 and TIP50 way back in this thread.
John tried BUT11A, it did not work.
What really bothers me is that both John's and mine surviving 2023R's all measure a very low hFE (7-10) and all the substitutes (and the datasheet on the SANKEN 2023 no mention of an "R" version) says min. 30
SONY did use a special "R" or "O" grade 2023.
Could it be that the "R" grade is a low hFE version, needed in this design?
So that the 2023's not only needs to be closely matched but also very low hFE?
Anyway the cap will be changed
Since using the amp as it is now, is a bad idea, the rails will likely collapse when current gets too high, and the bulb starts to light up, I guess I should focus on the PSU.
Reading how John's PSU died after a few months doesn't makes me optimistic.
http://www.diyaudio.com/forums/solid-state/220179-faulty-sony-ta-f6b.html#post3175922
and
http://www.diyaudio.com/forums/solid-state/220179-faulty-sony-ta-f6b-2.html#post3178125
I'm pretty sure those 2023's I ordered are fake, the SANKEN logo doesn't look right, the legs are thinner, they weigh less than the originals (2.6 grams vs. 1.9 grams).
What I've done so far on the PSU is a complete recap, tested all transistors and resistors, all OK except for the blown 2023R's.
The linear regulator works when fitted whit 2 original 2023R's
I have 4 good original 2023R's
So my options are:
1) use them in the switching section and find something new for the linear regulator.
2) use two original 2023R's in the linear reg. and find something new for the switching section.
3) find something new for both sections.
4) ?
As the remaining 2023R's have bin stressed 3) or 4) will likely be the best option.
I asked ON semi for a replacement, they came up whit this:
https://www.onsemi.com/PowerSolutio...=EE&category=EE6&question=4680&includedC=true
Searching mouser for "2sc2023" gives this result:
NTE394 NTE | Mouser
But It's not available in my region
In this post
http://www.diyaudio.com/forums/solid-state/181121-sony-ta-f6b-psu-repair-58.html#post3622944
amc184 uses FSC2335 but we don't know how that turned out yet.
You (or was it Jaycee) mentioned MJE13007 and TIP50 way back in this thread.
John tried BUT11A, it did not work.
What really bothers me is that both John's and mine surviving 2023R's all measure a very low hFE (7-10) and all the substitutes (and the datasheet on the SANKEN 2023 no mention of an "R" version) says min. 30
SONY did use a special "R" or "O" grade 2023.
Could it be that the "R" grade is a low hFE version, needed in this design?
So that the 2023's not only needs to be closely matched but also very low hFE?
Given you have four original 2023's I think I would use those in the invertor and fit replacements for the linear reg. The two transistors in the reg are in parallel with no current sharing resistors and so these should be closely matched devices to ensure each shares the current equally. I wouldn't like to say what the R or O grades denote. As to them measuring low hfe on a meter, well it might be worth testing and comparing them in a simple circuit by injecting a fixed base current and measuring the collector current. The meter will test at really low current and any differences in vbe could make a massive difference given the low volt/current meter test conditions.