Hi Guys:
Looking for help to solve an issue of excessive DC Voltage imbalance >+30VDC
Picked up this Amp that had one channel non-functional....
So I viewed the board to see the what had failed & noticed the a few
resistors failed in the front end off the +71VDC and a few transistors seemed to measure a little off. The +71VDC rail voltage decoupling cap (1000uF 100V) was slightly bulging at the top vent so I replaced both +/- rail decoupling caps to be safe.
The rail voltage fuses (6A) are not affected by this excess DC imbalance, so there is not a shorted output device....
Using the 60Watt bulb in series with the 120VAC (I know its not ideal but I have fixed multiple amps using this method) to prevent excess current & blown Rail fuses. This lowers the Rail Voltages to +/-37 VDC for the following tests....
So I replaced the SA1016/SC2362 diff pairs & few resistors.... no change...
After a few rounds of parts swap out, I have replaced all the transistors including driver & output devices, all IN4148 diodes in the front end, as well as the IN4003 across the outputs, and most resistors associated with the transistors except 2kohm Pot & 2 associated resistors that measure good.
I have used the good channel amplifier fed by the bad channel's diode bridge & power supply caps and it worked perfectly.... so this rules out the bad channel's PS.
I have checked for any bad solder joints. I resoldered all the connections that terminated cables for the +/- VDC & ground cables going to Power Supply. I checked for any cracks in the board or solder bridges as well.
So I can adjust the bias voltage up to recommended 7mV and the DC offset would be reduced to +19VDC (using the bulb in series with AC rail voltage is only +/-37VDC)). If I increase the bias voltage (across testpoints T2 & T4)
to 20mV the DC offset is below 5mV BUT the bulb is glowing very brightly.... probably was forcing the AMP heavily into CLASS A operation & ???
For reference the good channel measures <5mV DC offset and a lower bias voltage of 2mV using the 60Watt light bulb in series with the AC Line voltage, so this rules out this not-so-optimum testing condition as a factor in the excess DC offset voltage in the bad channel.
The output devices (SA1492 & SC3856 & the input transistors SA1016 & SC2362) are all reasonably matched for reference..
My next step is to remove the light bulb out of the circuit and rely on the 6A fuses to protect the power amp....and redo the bias but I believe there is just something else causing this DC voltage imbalance.
The service manual for this amp is downloadable from: B&W Group North America Service & Support - Home
Any help would be appreciated.
Thanks in advance,
Looking for help to solve an issue of excessive DC Voltage imbalance >+30VDC
Picked up this Amp that had one channel non-functional....
So I viewed the board to see the what had failed & noticed the a few
resistors failed in the front end off the +71VDC and a few transistors seemed to measure a little off. The +71VDC rail voltage decoupling cap (1000uF 100V) was slightly bulging at the top vent so I replaced both +/- rail decoupling caps to be safe.
The rail voltage fuses (6A) are not affected by this excess DC imbalance, so there is not a shorted output device....
Using the 60Watt bulb in series with the 120VAC (I know its not ideal but I have fixed multiple amps using this method) to prevent excess current & blown Rail fuses. This lowers the Rail Voltages to +/-37 VDC for the following tests....
So I replaced the SA1016/SC2362 diff pairs & few resistors.... no change...
After a few rounds of parts swap out, I have replaced all the transistors including driver & output devices, all IN4148 diodes in the front end, as well as the IN4003 across the outputs, and most resistors associated with the transistors except 2kohm Pot & 2 associated resistors that measure good.
I have used the good channel amplifier fed by the bad channel's diode bridge & power supply caps and it worked perfectly.... so this rules out the bad channel's PS.
I have checked for any bad solder joints. I resoldered all the connections that terminated cables for the +/- VDC & ground cables going to Power Supply. I checked for any cracks in the board or solder bridges as well.
So I can adjust the bias voltage up to recommended 7mV and the DC offset would be reduced to +19VDC (using the bulb in series with AC rail voltage is only +/-37VDC)). If I increase the bias voltage (across testpoints T2 & T4)
to 20mV the DC offset is below 5mV BUT the bulb is glowing very brightly.... probably was forcing the AMP heavily into CLASS A operation & ???
For reference the good channel measures <5mV DC offset and a lower bias voltage of 2mV using the 60Watt light bulb in series with the AC Line voltage, so this rules out this not-so-optimum testing condition as a factor in the excess DC offset voltage in the bad channel.
The output devices (SA1492 & SC3856 & the input transistors SA1016 & SC2362) are all reasonably matched for reference..
My next step is to remove the light bulb out of the circuit and rely on the 6A fuses to protect the power amp....and redo the bias but I believe there is just something else causing this DC voltage imbalance.
The service manual for this amp is downloadable from: B&W Group North America Service & Support - Home
Any help would be appreciated.
Thanks in advance,
If the output transistors seem OK, my guess is that there's something wrong in the front end, despite changing out most components. Time to make measurements .
You have a working channel which is great, it means you can make measurements in one channel and compare it to the other. Start at the input and work your way towards the output.
You have a working channel which is great, it means you can make measurements in one channel and compare it to the other. Start at the input and work your way towards the output.
Voltage Measurements have been documented on schematic
I have made measurements at critical points in the input & driver stages for your review. Please let me know what is the likely cause.
Note the rail voltages are lower (+/-35VDC versus +/-71VDC) as a result of a 60W light bulb in series with 120V line voltage.
Note: Using DMM in diode mode, all input/driver transistors measure good. All 1/4 Watt resistors have been replaced in input & driver stages. All solder joints have been re-soldered and has been fully checked for cracks in board and solder bridges.
Please review attachment for measurements.
I have made measurements at critical points in the input & driver stages for your review. Please let me know what is the likely cause.
Note the rail voltages are lower (+/-35VDC versus +/-71VDC) as a result of a 60W light bulb in series with 120V line voltage.
Note: Using DMM in diode mode, all input/driver transistors measure good. All 1/4 Watt resistors have been replaced in input & driver stages. All solder joints have been re-soldered and has been fully checked for cracks in board and solder bridges.
Please review attachment for measurements.
You don't mention the voltages appearing in the other channel. The comparison is what reveals the problems but since the rail voltages are only half the correct level, the current drawn by the amplifiers seems significant and possibly they are too low to allow correct operation, which might reveal more of the problem. If you must use direct connection to the PSU as your next step, fit 2-5 W resistors, around 100R, in place of the rail fuses as a precaution.
If you zero the bias reading and the rail voltages are still low, there is certainly another problem. A higher wattage bulb may suffice (100-120W) to give a clearer picture without further damage. Increasing the bias level and getting a glow simply supports this likelihood as Rotels generally feature quite low bias levels but obviously, there are 10 output devices and drivers to bias yet the TP measurement only checks one device. i.e it doesn't tell anything about the condition of the other output devices.
If you zero the bias reading and the rail voltages are still low, there is certainly another problem. A higher wattage bulb may suffice (100-120W) to give a clearer picture without further damage. Increasing the bias level and getting a glow simply supports this likelihood as Rotels generally feature quite low bias levels but obviously, there are 10 output devices and drivers to bias yet the TP measurement only checks one device. i.e it doesn't tell anything about the condition of the other output devices.
I agree with Ian.
35 V is probably too low for correct operation and comparison must be made to the correctly operating channel. Where measurements between the two differ (significantly) is probably where the problem is.
If the problem does seem to ly in an output device after all, measure the voltages across all 0.22 Ohm resistors, they should be more or less equal. If there is one that is significantly off, remove that output pair and try the amp again.
35 V is probably too low for correct operation and comparison must be made to the correctly operating channel. Where measurements between the two differ (significantly) is probably where the problem is.
If the problem does seem to ly in an output device after all, measure the voltages across all 0.22 Ohm resistors, they should be more or less equal. If there is one that is significantly off, remove that output pair and try the amp again.
I would tackle this by first shorting out the vbe multiplier Q617. This will force any output stage bias voltage to zero and force the output stage to be cut off. Be sure R631 and R633 are good. Also check D609 and D611. These can be removed and are only needed as protection driving highly reactive loads.
If the bulb still lights brightly after this then you have to first fix this issue. Any offset fault will still be present of course but first you have to stop the excess current draw.
Could the offset and current be caused by HF oscillation ? Have you scoped the output ?
If the bulb still lights brightly after this then you have to first fix this issue. Any offset fault will still be present of course but first you have to stop the excess current draw.
Could the offset and current be caused by HF oscillation ? Have you scoped the output ?
Just last night i had to fix a HK 870 with 7 volts of offset in the one out ...Nightmare to remove piggy board of drive to measure and/or exchange transistors with no success but while looking equipped with a second pair of eyes i noticed a good crack on the pcb solder it fixed in seconds
you may inspect that also
As Ian said the best approach is to measure always looking at the working Chanel for reference ,
Often there is a case where a transistor measures quite properly with a DVM or even a hfe tester but fails under 50-60 volts you must observe that also
Finally if the offset is positive you may start swapping parts of the +rail from the working channel but always with caution .
starting swap point should be transistors Q613 -619 - 635 ( this will be of course after verifying proper voltage conditions )
NOTE :
( at the specific amp consumption is fairly low also bias is fairly low .... 60 W bulb may drop the mains a few volts but not half of it .Behavior like that means that one of your transistors is conductive fully open when powered and that brings us to the point where a semi looks ok outside the amp but not when powered with 50-70 volts )
Kind regards
sakis
you may inspect that also
As Ian said the best approach is to measure always looking at the working Chanel for reference ,
Often there is a case where a transistor measures quite properly with a DVM or even a hfe tester but fails under 50-60 volts you must observe that also
Finally if the offset is positive you may start swapping parts of the +rail from the working channel but always with caution .
starting swap point should be transistors Q613 -619 - 635 ( this will be of course after verifying proper voltage conditions )
NOTE :
( at the specific amp consumption is fairly low also bias is fairly low .... 60 W bulb may drop the mains a few volts but not half of it .Behavior like that means that one of your transistors is conductive fully open when powered and that brings us to the point where a semi looks ok outside the amp but not when powered with 50-70 volts )
Kind regards
sakis
Last edited:
The differentials current source allow good operating point
down to voltages lower than the one measured with the bulb.
Since there s lot of output transistors , the voltage drop is normal
in a 60W bulb.
Voltages measurements should be done with two decimals as
it allow better diagnostics.
It appears that the current source Q605 conduct double the current
compared to the other side current source Q611 , wich shouldnt be the case ,
so check both theses transistors as it seems odd that they produce different
absolute emitter voltages while having the same base absolute voltages.
down to voltages lower than the one measured with the bulb.
Since there s lot of output transistors , the voltage drop is normal
in a 60W bulb.
Voltages measurements should be done with two decimals as
it allow better diagnostics.
It appears that the current source Q605 conduct double the current
compared to the other side current source Q611 , wich shouldnt be the case ,
so check both theses transistors as it seems odd that they produce different
absolute emitter voltages while having the same base absolute voltages.
I would try to use 100W bulb, or 2x60W parallel, to let a bit more voltage for the amplifier. And I would set the bias to 0 on the faulty channel, to reduce the power consumption.
Sajti
Update
I have taken new measurements and I will post later today.... To make it work with a 60 Watt bulb, I disconnected other channel while measuring.
I got rail voltages near 60V versus 70V so we should be more accurate.
The good channel worked as expected with the correct voltages with a Bias of 4.5mV versus the recommended 7mV, while the bad channel had +45VDC on the output. The bulb stayed very dim so I believe I could have just bypassed the bulb for the measurements as I believe the outputs are fine....
Stay tuned.
I have taken new measurements and I will post later today.... To make it work with a 60 Watt bulb, I disconnected other channel while measuring.
I got rail voltages near 60V versus 70V so we should be more accurate.
The good channel worked as expected with the correct voltages with a Bias of 4.5mV versus the recommended 7mV, while the bad channel had +45VDC on the output. The bulb stayed very dim so I believe I could have just bypassed the bulb for the measurements as I believe the outputs are fine....
Stay tuned.
I can set the bias voltage to 4.5mV on the faulty channel and the DC offset goes down to 45VDC versus 58VDC when the bias is set at minimum 0.1mV.
Stay tuned...
Looking at the voltages written in de attachment of post #3, I can't help but think that things go wrong around Q615, R627, R629, VR601.
It's as if there isn't enough current flowing from the negative rail through these components wich causes high positive voltages to appear on the base of Q617.
Being in the middle of almost identical voltage dividers between positive and negative rails, shouldn't this measure ~0 V-ish?
Please compare measurements around these components with those of the good channel.
It's as if there isn't enough current flowing from the negative rail through these components wich causes high positive voltages to appear on the base of Q617.
Being in the middle of almost identical voltage dividers between positive and negative rails, shouldn't this measure ~0 V-ish?
Please compare measurements around these components with those of the good channel.
Last edited:
Measurements
Here are the measurements for the faulty & good channels.
Once the bias is increased on the faulty channel from 0.1mV to 4.5mV the DC Offset decreases from 58VDC to 45VDC.
The voltages in the input change very slightly if I increase Bias from 0.1mV to 4.5mV but the output decreases.
Please review the attachment and post any comments.
Thanks in advance,
Here are the measurements for the faulty & good channels.
Once the bias is increased on the faulty channel from 0.1mV to 4.5mV the DC Offset decreases from 58VDC to 45VDC.
The voltages in the input change very slightly if I increase Bias from 0.1mV to 4.5mV but the output decreases.
Please review the attachment and post any comments.
Thanks in advance,
Observed Changes from 1st Set to 2nd Set of Measurements
Between the 1st & 2nd set of measurements you will notice an improvement in the input stage.
I changed 3 resistors and resoldered all the jumpers for peace of mind.
So looking at the 2nd set of measurements we have + DC voltage at the output only and the input is reasonably good for now.
So you guys are saying that the 3 transistors in the pre-driver stage (Q613 619 & 635) that are on the + V Rail may be the cultprit? So if we are getting a Positive Offset then I should be looking at these 3 as per the measurents?
If it was a Negative offset I would be looking at the transistors being fed from the Negative V Rail?
Is this a general rule or am I misinterpreting your comments?
Let me know.
Thanks again for all your input!
Between the 1st & 2nd set of measurements you will notice an improvement in the input stage.
I changed 3 resistors and resoldered all the jumpers for peace of mind.
So looking at the 2nd set of measurements we have + DC voltage at the output only and the input is reasonably good for now.
So you guys are saying that the 3 transistors in the pre-driver stage (Q613 619 & 635) that are on the + V Rail may be the cultprit? So if we are getting a Positive Offset then I should be looking at these 3 as per the measurents?
If it was a Negative offset I would be looking at the transistors being fed from the Negative V Rail?
Is this a general rule or am I misinterpreting your comments?
Let me know.
Thanks again for all your input!
Parts Replacement - No Change
Replaced Q613, Q615, Q617 with new spare replacements..... no change.
Then replaced Q619 & Q621 (with used units- Hfe & Diode measurements are good) and no change.
Note Is this a clue but : Q615 (2SC2682) on heatsink does get hot but there is +57VDC on collector , -63.6VDC on Emiiter & -62.7VDC on the Base while its partner Q613 (2SA1142) also on the heatsink is barely warm as their is 64.5VDC on Emitter, 62.3VDC on Collector & 62VDC on Base... probably not conducting much current.
The Input Stage voltages are still reasonable on the +/- ~0.5V ranges on their transmitter emitters (diff pairs & current sources).
I have reach a crossroad as I don't know what to do next..... lol!
The only parts that haven't been changed out are the 160V Polystyrene pico range capacitors and the 2.2kohm pot.
The board has no cracks as I was going to desolder all connections to look for separated/disconnected solder terminal rings. But the Rotel rings are pretty beefy..
Any ideas.
I will buy another pair of Q619 & Q621 as the ones I just pulled out seem to measure good via Hfe & diode test as well as the pair I put back in its place.
Thanks in advance,
Replaced Q613, Q615, Q617 with new spare replacements..... no change.
Then replaced Q619 & Q621 (with used units- Hfe & Diode measurements are good) and no change.
Note Is this a clue but : Q615 (2SC2682) on heatsink does get hot but there is +57VDC on collector , -63.6VDC on Emiiter & -62.7VDC on the Base while its partner Q613 (2SA1142) also on the heatsink is barely warm as their is 64.5VDC on Emitter, 62.3VDC on Collector & 62VDC on Base... probably not conducting much current.
The Input Stage voltages are still reasonable on the +/- ~0.5V ranges on their transmitter emitters (diff pairs & current sources).
I have reach a crossroad as I don't know what to do next..... lol!
The only parts that haven't been changed out are the 160V Polystyrene pico range capacitors and the 2.2kohm pot.
The board has no cracks as I was going to desolder all connections to look for separated/disconnected solder terminal rings. But the Rotel rings are pretty beefy..
Any ideas.
I will buy another pair of Q619 & Q621 as the ones I just pulled out seem to measure good via Hfe & diode test as well as the pair I put back in its place.
Thanks in advance,
In my experience in over 12 years in industrial electronics manufacturing, cracks in boards can be so small that a microscope is needed to see them. Failing that, you might check continuity of traces with your DMM.
Do you have access to a scope? I ask this because oscillations could register as DC voltage on a DMM while they're actually AC. A faulty cap could be the cause of oscillations. Failing a scope, replace those you haven't yet too.
VR601 is among the resistors in the negative leg of the circuit that drives Q617. A wrong voltage on its base will affect the circuit to the right of it, I'd replace VR601 to make sure it's OK.
If I really couldn't find the fault, I'd probably resort to removing Q617, Q619 and Q621. That way any connection to the circuit to the right of Q613 and Q615 is broken. In between VR601 and R625 should be ~0 V. If it isn't, there must be something wrong there. If it there is ~0 V there, then reinsert Q617 and measure again.
Edit: did you check the caps that have not been replaced for short-circuit?
Do you have access to a scope? I ask this because oscillations could register as DC voltage on a DMM while they're actually AC. A faulty cap could be the cause of oscillations. Failing a scope, replace those you haven't yet too.
VR601 is among the resistors in the negative leg of the circuit that drives Q617. A wrong voltage on its base will affect the circuit to the right of it, I'd replace VR601 to make sure it's OK.
If I really couldn't find the fault, I'd probably resort to removing Q617, Q619 and Q621. That way any connection to the circuit to the right of Q613 and Q615 is broken. In between VR601 and R625 should be ~0 V. If it isn't, there must be something wrong there. If it there is ~0 V there, then reinsert Q617 and measure again.
Edit: did you check the caps that have not been replaced for short-circuit?
Last edited:
I don't mean this to sound critical so please don't take this the wrong way but you have to be much much more methodical in this. Replacing components in hope is a recipe for disaster.
All the evidence is there if you can find it.
For example,
"Is this a clue but : Q615 (2SC2682) on heatsink does get hot but there is +57VDC on collector , -63.6VDC on Emiiter & -62.7VDC on the Base
That gives a B-E drop (and this is where the main clues will lie) of 0.9 volts which should turn the transistor fully on. (In fact it shouldn't really be possible to reach 0.9v across a forward biased junction at low currents). So if we say "fully on" then why is the C-E differential 120 volts or so ?
I'm sure all these semiconductors aren't faulty, its just you are not measuring in the correct way. You MUST do a measurement across the B-E junction to get a true reading and determine from that if there is a problem. ALL the B-E drops irrespective of NPN or PNP type should be around 0.7 if forward biased. If any are reverse biased due to the fault then that will limit out at around 8 or 9 volts (of the opposite polarity) give or take. Rmemember for the NPN's the base is the higher voltage with respect to the emitter and vice versa for PNP.
If needed you can break this amp into stages and with 10 minutes work get a good idea of where the fault is.
Did you short out the vbe multiplier as I suggested earlier ?
If you remove Q619 and Q621 you isolate the output stage completely.
Do you see component marked "C657" which is a 22.1K RESISTOR that is the main feedback return. Its to the right of the output transistors and connects to the amp output. If that is removed and the right hand end reconnected so that it now connects to our "shorted out vbe multiplier" i.e. collector of Q614 or Q616 and the amp powered up then the voltage at Q614/616 should settle to near zero. If it does then you are looking at a problem to the right of this. If the offset (at this point in the circuit) is still present then you have a problem in the front end.
All the evidence is there if you can find it.
For example,
"Is this a clue but : Q615 (2SC2682) on heatsink does get hot but there is +57VDC on collector , -63.6VDC on Emiiter & -62.7VDC on the Base
That gives a B-E drop (and this is where the main clues will lie) of 0.9 volts which should turn the transistor fully on. (In fact it shouldn't really be possible to reach 0.9v across a forward biased junction at low currents). So if we say "fully on" then why is the C-E differential 120 volts or so ?
I'm sure all these semiconductors aren't faulty, its just you are not measuring in the correct way. You MUST do a measurement across the B-E junction to get a true reading and determine from that if there is a problem. ALL the B-E drops irrespective of NPN or PNP type should be around 0.7 if forward biased. If any are reverse biased due to the fault then that will limit out at around 8 or 9 volts (of the opposite polarity) give or take. Rmemember for the NPN's the base is the higher voltage with respect to the emitter and vice versa for PNP.
If needed you can break this amp into stages and with 10 minutes work get a good idea of where the fault is.
Did you short out the vbe multiplier as I suggested earlier ?
If you remove Q619 and Q621 you isolate the output stage completely.
Do you see component marked "C657" which is a 22.1K RESISTOR that is the main feedback return. Its to the right of the output transistors and connects to the amp output. If that is removed and the right hand end reconnected so that it now connects to our "shorted out vbe multiplier" i.e. collector of Q614 or Q616 and the amp powered up then the voltage at Q614/616 should settle to near zero. If it does then you are looking at a problem to the right of this. If the offset (at this point in the circuit) is still present then you have a problem in the front end.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.