The Caps in question are 2 6800uf/50V in the Left PS of a Harman Kardon 730.
The Right channel is good and at Power off they drop from 44V to 2mV in about a minute.
The Left channel took 24 Hours to drop from 44V to 36V so I added a couple of 2.2M ohm resistors I had around to Bleed the Caps.
Slight progress, it now takes 4 Hours to from 44V to 36V
Oh and this is with the Main Amp PCB removed from the Chassis. I am not very knowledgeable but that just does not make sense to me.
Any thoughts?
The Right channel is good and at Power off they drop from 44V to 2mV in about a minute.
The Left channel took 24 Hours to drop from 44V to 36V so I added a couple of 2.2M ohm resistors I had around to Bleed the Caps.
Slight progress, it now takes 4 Hours to from 44V to 36V
Oh and this is with the Main Amp PCB removed from the Chassis. I am not very knowledgeable but that just does not make sense to me.
Any thoughts?
Could be the different supplies have different loads on them.
Or the cap discharging first is faulty.
As an example I have hybrid TDA7294 with SRPP valve front end.
The heater works off the negative supply so that one discharges first.
Or the cap discharging first is faulty.
As an example I have hybrid TDA7294 with SRPP valve front end.
The heater works off the negative supply so that one discharges first.
Pass DIY Addict
Joined 2000
Paid Member
with no load on the power supply, I would conclude that the caps that discharge most quickly are the ones at fault. Typically, caps begin to fail after about 15+ years and the fault appears as a short in the cap. The short might be intermittent at first, or will show up as less and less resistance across the poles of the cap.
The cap(s) that retains most of its charge for the better part of a day sound like they are just fine.
Does the amp function properly? Often, an amp with dying power supply caps will begin to blow fuses on a more frequent basis. This happens as the caps form short circuits and the power supply rails short to ground, thus increasing the draw on the transformer and blowing the fuse.
The cap(s) that retains most of its charge for the better part of a day sound like they are just fine.
Does the amp function properly? Often, an amp with dying power supply caps will begin to blow fuses on a more frequent basis. This happens as the caps form short circuits and the power supply rails short to ground, thus increasing the draw on the transformer and blowing the fuse.
The issue is an up to 30V DC spike to the Speaker Terminals in the L Channel at startup.
The Right Channel works as it should and the Caps drain within a min of power off, even with the Main Amp PCB out of the chassis.
I think the L Caps are charging or keeping charged C406 which is part of the Slow Start Circuit so when powered on there is no Slow Start to prevent the Spike.
The HK 730 is a dual mono
The Right Channel works as it should and the Caps drain within a min of power off, even with the Main Amp PCB out of the chassis.
I think the L Caps are charging or keeping charged C406 which is part of the Slow Start Circuit so when powered on there is no Slow Start to prevent the Spike.
The HK 730 is a dual mono
Attachments
Not self discharging is not a problem but a bonus, it only means nothing is draining it.
Maybe you have a problem on that side too, but it´s not the capacitor´s fault.
Sometimes (in fact quite often) , two things happen at the same time, maybe even they are related somehow, yet one is not causing the other.
Typcal example: 90% of people die in a bed ... does that make beds "deadly" ?
Maybe you have a problem on that side too, but it´s not the capacitor´s fault.
Sometimes (in fact quite often) , two things happen at the same time, maybe even they are related somehow, yet one is not causing the other.
Typcal example: 90% of people die in a bed ... does that make beds "deadly" ?
The fact that the left channel supply caps do not discharge probably means that the left channel is not biased properly and this would in fact point to need to replace the electrolytic caps on the left channel PCB first and make sure that there are not other problems as well. Start by replacing C402, C406, C408, and C412.
Thanks for the replies
Thing is, in this case, the R Channel which drains, works perfectly. The L Channel will try to send 30VDC to my Speaker during the time the Slow Start Circuit should be stopping it from doing that. The L Channel works perfectly once the spike has passed
Thing is, in this case, the R Channel which drains, works perfectly. The L Channel will try to send 30VDC to my Speaker during the time the Slow Start Circuit should be stopping it from doing that. The L Channel works perfectly once the spike has passed
C406 provides a time constant that allows Q406 to turn on slowly, if it is bad it will turn on immediately and some asymmetry is possible as a result. Other components associated with C406 like resistors and diodes could be bad as well.
The symptoms don't point directly at the supply caps although it could be possible that they need to be replaced because one of them has insufficient capacitance and the supplies are not coming up symmetrically.
Do you have a scope?
The symptoms don't point directly at the supply caps although it could be possible that they need to be replaced because one of them has insufficient capacitance and the supplies are not coming up symmetrically.
Do you have a scope?
Would have been helpful to know that. Shotgunning is not the way to go, you need to get a scope and carefully compare what is happening in each stage of both amplifier channels, there is a bad part in there somewhere.
Look at the rise times of the supplies on both channels as well.
Not much more I can suggest.
Look at the rise times of the supplies on both channels as well.
Not much more I can suggest.
kevinkr is right. You need to take some measurements.
The fact that the spike is almost at the positive rail suggests the negative is not turning on or the negative half of the amplifier is not turning on at aprox. the same time as the positive half. Check bias. At turn on it is probably way off. Once the amp is running, whatever component is not turning on finally turns on, and at shut off, it is shutting off first, and that channel is not drawing any current to drain the caps.
Look at the diodes in the front end, and possibly Q412.
The fact that the spike is almost at the positive rail suggests the negative is not turning on or the negative half of the amplifier is not turning on at aprox. the same time as the positive half. Check bias. At turn on it is probably way off. Once the amp is running, whatever component is not turning on finally turns on, and at shut off, it is shutting off first, and that channel is not drawing any current to drain the caps.
Look at the diodes in the front end, and possibly Q412.
Pass DIY Addict
Joined 2000
Paid Member
Have you actually measured that the left channel is working perfectly, or is the left channel simply producing music that doesn't sound distorted/bad? There can be a significant difference that only careful measurements with meters and a signal generator/scope will reveal.
My experience is perhaps more limited than that of others here, but my first target in a mis-behaving amp is always to go after caps. The general rule is that if one cap is sufficiently old/tired that it is starting to behave badly, they are all potentially likely to be bad. More generally, what you are describing is the result of either a power supply that cannot provide power as intended, or an amplifier stage that is not drawing power as intended. With the ongoing revelation of details, it appears that the amp circuit is in need of greater attention, careful measurements, and comparisons across channels as Kevin and Dan indicate.
Last edited:
I don't have a Scope, My skills are at the level of if someone on the Internet has had and fixed an Issue I can follow what they did.
Have checked the Idle Current (Page 5 of the SM) in the Left after warming up for 20 mins and it is in Spec 35mV
Replaced Q412 & Q408 at the same time I replaced Q402/4/6 and all the electrolytics. I replaced R406 because I discovered that C406 was not draining after Power Off. When replacing R406 did not resolve C406 draining thats when I started looking at C9/10. I was/am going to replace the Bias Diodes but came across the C406 issue.
Turned it on this morning after checking to make sure C406 had drained (.5V) which matched C405. The Slow Start appeared to work and I got a 1V peak at about 8 secs.
Its been off for 2 Hours now and I read 4.5V on C406 vs 0.3V on C405. On C 9 & 10 I am reading 4.5V and 8V. Powered On after 2 hours I get a 32V peak at 4 secs in the L. In the R I get a 1.2V peak at about 5 secs
Have checked the Idle Current (Page 5 of the SM) in the Left after warming up for 20 mins and it is in Spec 35mV
Replaced Q412 & Q408 at the same time I replaced Q402/4/6 and all the electrolytics. I replaced R406 because I discovered that C406 was not draining after Power Off. When replacing R406 did not resolve C406 draining thats when I started looking at C9/10. I was/am going to replace the Bias Diodes but came across the C406 issue.
Turned it on this morning after checking to make sure C406 had drained (.5V) which matched C405. The Slow Start appeared to work and I got a 1V peak at about 8 secs.
Its been off for 2 Hours now and I read 4.5V on C406 vs 0.3V on C405. On C 9 & 10 I am reading 4.5V and 8V. Powered On after 2 hours I get a 32V peak at 4 secs in the L. In the R I get a 1.2V peak at about 5 secs
C406 shouldn't have more than around 1.2V across it (assuming you're measuring across the cap and not to ground or -B1). The base-emitter junction of Q406 and D404 are in series across it, so that should hold it to two diode junction drops. Check the transistor and diode.
Last edited:
The voltages reading are with it powered off, 1 probe to ground, the other to the + terminal of the capacitor.
I think the reason C406 has been reading up to 4.5V is because its pulling a charge from C9/10.
If I drain C9/10 (with a 10W R), C406 drains as well to the same charge (0.4V) as C405 and when powered on I get a 1 to 1.4V peak at the L speaker terminal after about 8 secs which matches the R Channel. I have repeated this 4 times now with the same result. Attached test speakers the last 2 times and there was no thrumb at all in the L Speaker
I think the reason C406 has been reading up to 4.5V is because its pulling a charge from C9/10.
If I drain C9/10 (with a 10W R), C406 drains as well to the same charge (0.4V) as C405 and when powered on I get a 1 to 1.4V peak at the L speaker terminal after about 8 secs which matches the R Channel. I have repeated this 4 times now with the same result. Attached test speakers the last 2 times and there was no thrumb at all in the L Speaker
Check R406, D402, D404 and D406. Make sure they are all still diodes (i.e. not shorted) I assume D402 and D404 are the diodes you intend to replace. Don't replace the bias diode (D408) in output stage unless you want additional and potentially more serious problems.
Note that if Q406 shuts off prematurely it will also shut off Q412 and Q408 before the power supply caps are discharged, and with the output stage in cutoff they will not discharge. Note that if they Q412 and Q408 are not turned on at the same time and at the same rate you will have large offsets on the outputs until things settle.
Also make sure there is no DC pulse on the base of Q402 as you power up the amp.
The 1.2V on the right channel actually isn't a sign that all is well on that channel either. Find the problem on the left channel and go back and do the capacitor replacements on the other. Don't change semiconductors.
Note that if Q406 shuts off prematurely it will also shut off Q412 and Q408 before the power supply caps are discharged, and with the output stage in cutoff they will not discharge. Note that if they Q412 and Q408 are not turned on at the same time and at the same rate you will have large offsets on the outputs until things settle.
Also make sure there is no DC pulse on the base of Q402 as you power up the amp.
The 1.2V on the right channel actually isn't a sign that all is well on that channel either. Find the problem on the left channel and go back and do the capacitor replacements on the other. Don't change semiconductors.
Some of the capacitor values may be somewhat critical to getting a pop free unmute, I assume you are using the same value and tolerance parts as originally fitted. (I cannot imagine they selected any parts at all for this amplifier, likely straight unselected bogey parts)
20% tolerance would be appropriate. In the low voltage locations use a 6.3V part as shown.
20% tolerance would be appropriate. In the low voltage locations use a 6.3V part as shown.
Kevin, thanks
"Note that if Q406 shuts off prematurely it will also shut off Q412 and Q408 before the power supply caps are discharged, and with the output stage in cutoff they will not discharge. Note that if they Q412 and Q408 are not turned on at the same time and at the same rate you will have large offsets on the outputs until things settle."
That sounds like what has been happening but if C9/10 & C406 are fully discharged before Power On then the Slow Start works.
At the Speaker Terminal, it will read 0V for about 8 secs, jump to 1.2V and rapidly drop to about 15mV.
If I don't discharge C9/10/406, at the Speaker Terminal it will read 0V for 1 - 4 secs then jump to anywhere from 4 - 30V and settle to about 15mV after 1 min.
If C406 is still charged at Power On because its drawing a charge from C9/10, the Slow Start can't work. If I discharge C9/10 then C406 drains within seconds and the Slow Start works.
All the Caps I replaced on the Main Amp were the same value but slightly higher voltage 16V vs 6.3 for C406
If I can find a way to check for a DC Pulse without shorting something I will. With the Main Amp PCB in the Chassis its hard to get at.
"Note that if Q406 shuts off prematurely it will also shut off Q412 and Q408 before the power supply caps are discharged, and with the output stage in cutoff they will not discharge. Note that if they Q412 and Q408 are not turned on at the same time and at the same rate you will have large offsets on the outputs until things settle."
That sounds like what has been happening but if C9/10 & C406 are fully discharged before Power On then the Slow Start works.
At the Speaker Terminal, it will read 0V for about 8 secs, jump to 1.2V and rapidly drop to about 15mV.
If I don't discharge C9/10/406, at the Speaker Terminal it will read 0V for 1 - 4 secs then jump to anywhere from 4 - 30V and settle to about 15mV after 1 min.
If C406 is still charged at Power On because its drawing a charge from C9/10, the Slow Start can't work. If I discharge C9/10 then C406 drains within seconds and the Slow Start works.
All the Caps I replaced on the Main Amp were the same value but slightly higher voltage 16V vs 6.3 for C406
If I can find a way to check for a DC Pulse without shorting something I will. With the Main Amp PCB in the Chassis its hard to get at.
Do you have a good ground connection from the power supply to the PA board? Are you getting both +45V and -45V supplies to the PA?
You should establish that at the very start. I haven't seen you mention a negative voltage yet. C406 positive terminal would normally show a negative voltage to ground (I figure around -40V), so your reading of 4.5V makes me suspicious.
You should establish that at the very start. I haven't seen you mention a negative voltage yet. C406 positive terminal would normally show a negative voltage to ground (I figure around -40V), so your reading of 4.5V makes me suspicious.
Sorry,
I've been lazy when it comes to describing +-.
B+/- is 44V
The V readings on C406 have been up to -6V which drops to -0.5V when C406 is fully discharged.
When C406 has been fully discharged, which doesn't happen until C9/10 have discharged, the Slow Start works as it should
I'm going to order new Caps for C9/10 as they are the only component that behaves differently from the R Channel and see what, if any, difference that makes.
Oh, and I did try reflowing the solder on the PS PCB
I've been lazy when it comes to describing +-.
B+/- is 44V
The V readings on C406 have been up to -6V which drops to -0.5V when C406 is fully discharged.
When C406 has been fully discharged, which doesn't happen until C9/10 have discharged, the Slow Start works as it should
I'm going to order new Caps for C9/10 as they are the only component that behaves differently from the R Channel and see what, if any, difference that makes.
Oh, and I did try reflowing the solder on the PS PCB
- Status
- Not open for further replies.