Ok, I can take a hint...
I'll read up and do a few calculations later, if possible, and report back... Cheers
Nigel
Strange DC offset problem
Hi Guys,
My amp has developed a strange problem, and I have failed today to find the cause. I'll have another go tomorrow, but I thought I'd post to see if anyone has a good guess as to what might be causing it.
Turning the amp on, everything works just fine (and sounds just fine). After a few minutes the speaker protection circuit kicks in and disconnects the signal from the speakers. After a pause it comes back, only to cut out again, and so on...
So I unhooked it from the system and checked the DC offset. What is happening is that it starts out reasonable (150-200mV or so) and drops
to about 30-40mV, although it is very jumpy, and after a few minutes it jumps to well over 1V, obviously triggering the protection circuit.
It seems to me like it could be a transistor switching off, for some reason, but it's a bit odd that an otherwise stable amp would start doing this... Could it be a failing or failed part? Or maybe I have the trimpots setting the amp at an inherently unstable state? (Somehow??)
It seems unlikely to me that it's the detector circuit causing the problem, but for reference, it's the one from Rod Elliott's ESP site.
There are no other obvious signs that I can see, so I'd be very grateful for any suggestions for things to try or measure tomorrow.
Cheers
Nigel
Hi Guys,
My amp has developed a strange problem, and I have failed today to find the cause. I'll have another go tomorrow, but I thought I'd post to see if anyone has a good guess as to what might be causing it.
Turning the amp on, everything works just fine (and sounds just fine). After a few minutes the speaker protection circuit kicks in and disconnects the signal from the speakers. After a pause it comes back, only to cut out again, and so on...
So I unhooked it from the system and checked the DC offset. What is happening is that it starts out reasonable (150-200mV or so) and drops
to about 30-40mV, although it is very jumpy, and after a few minutes it jumps to well over 1V, obviously triggering the protection circuit.
It seems to me like it could be a transistor switching off, for some reason, but it's a bit odd that an otherwise stable amp would start doing this... Could it be a failing or failed part? Or maybe I have the trimpots setting the amp at an inherently unstable state? (Somehow??)
It seems unlikely to me that it's the detector circuit causing the problem, but for reference, it's the one from Rod Elliott's ESP site.
There are no other obvious signs that I can see, so I'd be very grateful for any suggestions for things to try or measure tomorrow.
Cheers
Nigel
Hi juma,
Just got finished checking a few things. I couldn't repeat exactly what happened yesterday, since the DC offset stayed high all the time and the speaker protection circuit stays active all the time. This leads me to suspect that something was failing and has now failed.
On the left channel the trimpot at R12 is at about 84R (measured in circuit - I didn't take it out to check like I should...) and I measure a little more than 1.4V across it - which is about 16-17mA as it should be. The right channel (which is the one that gave the trouble) shows essentially the same - a little more than 1.4 across 87R. So that looks OK to me.
The problem apears to be in the output stage. On the left channel I measure 0.57V across R7 and R16, which are 0R56, so a bias current of a little more than 1A, which is fine, but there appears to be NO voltage at all across the same resistors on the right channel. This is evidently the problem; or part of it at least.
So, it may be that one or both of the mosfets are blown... Here's a couple of questions.
1. The other time I had a mosfet blow (in another amp) it blew to a short circuit; is it just as likely for one to blow to an open circuit like one of these seems to have done?
2. These are pretty tough things. What might have caused it to go? I realise that there is bound to be a certain failure rate, but they weren't being pushed that hard - might something else blowing have caused it?
3. It's no great problem to switch out the mosfets, although it's pretty irritating and not cheap here. What else can I check to be sure it's really this and not something else causing the problem? I tried switching out the BC557 just in case, but that showed no difference.
Well thanks for any help
Cheers
Nigel
Just got finished checking a few things. I couldn't repeat exactly what happened yesterday, since the DC offset stayed high all the time and the speaker protection circuit stays active all the time. This leads me to suspect that something was failing and has now failed.
On the left channel the trimpot at R12 is at about 84R (measured in circuit - I didn't take it out to check like I should...) and I measure a little more than 1.4V across it - which is about 16-17mA as it should be. The right channel (which is the one that gave the trouble) shows essentially the same - a little more than 1.4 across 87R. So that looks OK to me.
The problem apears to be in the output stage. On the left channel I measure 0.57V across R7 and R16, which are 0R56, so a bias current of a little more than 1A, which is fine, but there appears to be NO voltage at all across the same resistors on the right channel. This is evidently the problem; or part of it at least.
So, it may be that one or both of the mosfets are blown... Here's a couple of questions.
1. The other time I had a mosfet blow (in another amp) it blew to a short circuit; is it just as likely for one to blow to an open circuit like one of these seems to have done?
2. These are pretty tough things. What might have caused it to go? I realise that there is bound to be a certain failure rate, but they weren't being pushed that hard - might something else blowing have caused it?
3. It's no great problem to switch out the mosfets, although it's pretty irritating and not cheap here. What else can I check to be sure it's really this and not something else causing the problem? I tried switching out the BC557 just in case, but that showed no difference.
Well thanks for any help
Cheers
Nigel
Nigel, please direct me to the schematic that you are referring.
Sorry, I really don't have the time right now to browse through the thread.
Obviously the output CCS ("upper" MOSFET with BC557) is not working. Are there any signs of high temp. on the board, parts or wires (changed colour)?
In the meantime check how turning a trimpot on the problematic channel affects the voltage on the JFET's drain resistor.
Sorry, I really don't have the time right now to browse through the thread.
Obviously the output CCS ("upper" MOSFET with BC557) is not working. Are there any signs of high temp. on the board, parts or wires (changed colour)?
In the meantime check how turning a trimpot on the problematic channel affects the voltage on the JFET's drain resistor.
Hi juma,
The schematic I am referring to is posted in #95 above. (Sorry - I should have said so earlier; of course you can't read the whole thread looking for it...)
There is nothing obviously burned or overheated or discoloured.
By the jfet's drain resistor you mean R10 and R11 (the 10R's on the jfets in the LTP)? Or R13 on Q4? Maybe best to check on the good channel also, to have a reference.... (Have to be tomorrow now.. a little late...)
Cheers
Nigel
The schematic I am referring to is posted in #95 above. (Sorry - I should have said so earlier; of course you can't read the whole thread looking for it...)
There is nothing obviously burned or overheated or discoloured.
By the jfet's drain resistor you mean R10 and R11 (the 10R's on the jfets in the LTP)? Or R13 on Q4? Maybe best to check on the good channel also, to have a reference.... (Have to be tomorrow now.. a little late...)
Cheers
Nigel
Ah... Looking at the pcb design posted above in #118 it looks like I have drain and source switched.... You may remember from the discussions above that I previously had one switched and the other not, and in the last version I "fixed" it. Also the datasheet says they are interchangeable, so it seems unlikely this is the problem (especially since the thing worked fine for a long time.) Nonetheless, for reference if other people try to build this, the pcb design above should be adjusted if this is an issue, and the schematic in #46 has an error, which is corrected in #53...
I'll measure things later and post. (I'm at work right now).
Cheers
Nigel
So I've done some measuring. Interestingly, the problem seems to be intermittent again.
SO: I measured on both channels, to have a comparison.
On the left channel (apparently OK): 4.66V across R1 (which is 560R, not 470 as shown in the schematic) and twiddling R12 changes this, although not by very much: Vgs on Q3 and Q6 is 4.0 V in both cases.
On the right channel (apparently not OK): 3.27 V across R1 (either didn't change at all altering R12 or changed too little to see easily. I didn't want to risk things by twiddling too far...): Vgs on Q3 is 3.27 V, but Vgs seems to drop from 3.2 or so (not enough time to really be sure of the measurement) to about 0.5V. Another time when I tried to check this when the relays wouldn't activate at all it was essentially zero.
Voltage across R7: I didn't check it on the left today, but yesterday it was 0.57V across 0R56, which is OK; on the left channel it is zero (across R16 it's the same, not surprisingly.)
Also, the jfets in the LTP are in the right orientation after all.. Obviously the pcb design I looked at above isn't the one I used in the end.
So, it looks to me like Q6 on the right channel is either dead or dying, but I have no idea why.....
Cheers
Nigel
SO: I measured on both channels, to have a comparison.
On the left channel (apparently OK): 4.66V across R1 (which is 560R, not 470 as shown in the schematic) and twiddling R12 changes this, although not by very much: Vgs on Q3 and Q6 is 4.0 V in both cases.
On the right channel (apparently not OK): 3.27 V across R1 (either didn't change at all altering R12 or changed too little to see easily. I didn't want to risk things by twiddling too far...): Vgs on Q3 is 3.27 V, but Vgs seems to drop from 3.2 or so (not enough time to really be sure of the measurement) to about 0.5V. Another time when I tried to check this when the relays wouldn't activate at all it was essentially zero.
Voltage across R7: I didn't check it on the left today, but yesterday it was 0.57V across 0R56, which is OK; on the left channel it is zero (across R16 it's the same, not surprisingly.)
Also, the jfets in the LTP are in the right orientation after all.. Obviously the pcb design I looked at above isn't the one I used in the end.
So, it looks to me like Q6 on the right channel is either dead or dying, but I have no idea why.....
Cheers
Nigel
Nigel, please leave the working channel alone, don't mess with it (if you poke it too much...). We are talking about and taking readings only from problematic channel.
Take out off the board Q3 and Q6 of the problematic channel and check them using the matching procedure:
http://passdiy.com/pdf/matching.pdf
That will show the problematic MOSFET.
Front-end section seems OK because CCS is working and JFETs are conducting.
Take out off the board Q3 and Q6 of the problematic channel and check them using the matching procedure:
http://passdiy.com/pdf/matching.pdf
That will show the problematic MOSFET.
Front-end section seems OK because CCS is working and JFETs are conducting.
Hi juma,
I am sure you're giving good advice here, but nonetheless I think it is worth the risk to understand better. I am trying to be very careful about checking things, to minimise the chances of bu****ing things up, but checking things against the working channel is (for me) a little bit like having a control in an experiment... By seeing what is "normal" I hope I understand better what is "altered". Don't worry though, I won't go so far as to pull a mosfet off the working channel...
OK, I'll get on to this later. So you expect the problem to be with one of the mosfets?
I'm guessing that the difference in voltage across R1 (comparing right and left channels) is merely a consequence of the output stage not working. Is this a reasonable assumption?
Here's some observations I'd like input on, if possible.
1. My local parts store probably won't have IRFP9140 (which I'm using right now) but will probably have IRFP9240. If I change them on the right channel my guess is I'd better do it in the left also, and it is probably cheaper (because of miminim orders and postage/packing) to change all four than to order online. Is there anything in the schematic that it would be better to change for the different mosfet?
2. Right now there is no C2 on the boards. You might not remember, but there is a discussion about his above in the thread... I pulled them off after having trouble. If I am going to put the boards off anyway I can easily put this in, or any other small change that seems advisable. Ideas?
3. It would be nice to understand what happened for the mosfet to blow, and arguable necessary to understand to avoid it happening again. (Supposing it's blown, of course...) Doesn't it seems a little odd that it would blow like this? It's not like it was being pushed too hard, unless there is something else on the board that is unstable... Ideas
Thanks for all the help
Nigel
I am sure you're giving good advice here, but nonetheless I think it is worth the risk to understand better. I am trying to be very careful about checking things, to minimise the chances of bu****ing things up, but checking things against the working channel is (for me) a little bit like having a control in an experiment... By seeing what is "normal" I hope I understand better what is "altered". Don't worry though, I won't go so far as to pull a mosfet off the working channel...
OK, I'll get on to this later. So you expect the problem to be with one of the mosfets?
I'm guessing that the difference in voltage across R1 (comparing right and left channels) is merely a consequence of the output stage not working. Is this a reasonable assumption?
Here's some observations I'd like input on, if possible.
1. My local parts store probably won't have IRFP9140 (which I'm using right now) but will probably have IRFP9240. If I change them on the right channel my guess is I'd better do it in the left also, and it is probably cheaper (because of miminim orders and postage/packing) to change all four than to order online. Is there anything in the schematic that it would be better to change for the different mosfet?
2. Right now there is no C2 on the boards. You might not remember, but there is a discussion about his above in the thread... I pulled them off after having trouble. If I am going to put the boards off anyway I can easily put this in, or any other small change that seems advisable. Ideas?
3. It would be nice to understand what happened for the mosfet to blow, and arguable necessary to understand to avoid it happening again. (Supposing it's blown, of course...) Doesn't it seems a little odd that it would blow like this? It's not like it was being pushed too hard, unless there is something else on the board that is unstable... Ideas
Thanks for all the help
Nigel
1. No need to change anything.
2. You are fine without C2.
3. MOSFET-killers are, most often, heat and too high Vgs (more than 20V). We can exclude Vgs as a culprit here. So, the first thing that comes to mind is quality of thermal coupling between the MOSFET and the heatsink - check the temp. differrence between them, the MOSFET' case shouldn't be much hotter than the heatsink (say, 10 degrees C at most).
Also, check the connections on the board (especially around power resistors) - bad solder joints may reveal themselves only at higher temp. It might be a good idea to resolder any suspicious looking joint (those who look too gray, porous, without a shine).
2. You are fine without C2.
3. MOSFET-killers are, most often, heat and too high Vgs (more than 20V). We can exclude Vgs as a culprit here. So, the first thing that comes to mind is quality of thermal coupling between the MOSFET and the heatsink - check the temp. differrence between them, the MOSFET' case shouldn't be much hotter than the heatsink (say, 10 degrees C at most).
Also, check the connections on the board (especially around power resistors) - bad solder joints may reveal themselves only at higher temp. It might be a good idea to resolder any suspicious looking joint (those who look too gray, porous, without a shine).
Well, I have my amp back together and working, at least in a temporary way.
Last few days I can't get to the passdiy website and look through the matching pages again, and I can't put it together from memory. (I gather from elsewhere here that there is some kind of problem with the passdiy site, but other people seem to get the pages open somehow - I don't know why I can't, but there you have it...) So rather than spend the weekend without music I decided to switch out the mosfets on the right (problematic) channel and see what happens. I couldn't get IRFP 9140, so I put a pair of IRFP9240s in the right channel, and for the time being have left the IRFP9140s in the left channel. After re-adjusting the DC offset (after about a half-hour this appears more or less stable at 10-15mV, which is fine) I fired it up and am now listening to Stan Getz. I changed nothing else on the boards at all. I'm going to leave it on this evening and see if anything goes wrong again.
I should add that all the voltages we measured before now look fine
When I can get the passdiy pages to open again I'll check the mosfets I took off, but I'm pretty sure I already know what it'll say - the upper one is blown, the other probably fine. This again raises the question of why this happened. There was nothing on the board that appeared bad - no dodgy-looking solder joints, no burned components, nothing visible at all. There must be a certain percentage of components that just fail, I suppose, but it seems much more likely that it was caused by something I did wrong. Can you think of anything else I could look into?
Anyway, I have music again (hooray!) at least for the time being.
Many thanks for the help, juma.
Cheers
Nigel
Last few days I can't get to the passdiy website and look through the matching pages again, and I can't put it together from memory. (I gather from elsewhere here that there is some kind of problem with the passdiy site, but other people seem to get the pages open somehow - I don't know why I can't, but there you have it...) So rather than spend the weekend without music I decided to switch out the mosfets on the right (problematic) channel and see what happens. I couldn't get IRFP 9140, so I put a pair of IRFP9240s in the right channel, and for the time being have left the IRFP9140s in the left channel. After re-adjusting the DC offset (after about a half-hour this appears more or less stable at 10-15mV, which is fine) I fired it up and am now listening to Stan Getz. I changed nothing else on the boards at all. I'm going to leave it on this evening and see if anything goes wrong again.
I should add that all the voltages we measured before now look fine
When I can get the passdiy pages to open again I'll check the mosfets I took off, but I'm pretty sure I already know what it'll say - the upper one is blown, the other probably fine. This again raises the question of why this happened. There was nothing on the board that appeared bad - no dodgy-looking solder joints, no burned components, nothing visible at all. There must be a certain percentage of components that just fail, I suppose, but it seems much more likely that it was caused by something I did wrong. Can you think of anything else I could look into?
Anyway, I have music again (hooray!) at least for the time being.
Many thanks for the help, juma.
Cheers
Nigel
Nigel, I'm glad it worked all right 
Please check the temperature difference between the MOSFET's case and the heatsink. It will show whether you did a good job on mounting MOSFETs to heatsink or not (it is an important reliability factor).
Here is the sch. that will help you check your old MOSFETs. Never mind the current through them (the 150R resistors sets it at about 75mA) or measurement voltage (it doesn't have to be exactly +15V, anything from +10V upward will do). If you get a reading on the voltmeter of about 4V - MOSFET is OK.
Take care about polarity (N-channel vs. P-channel).
Please check the temperature difference between the MOSFET's case and the heatsink. It will show whether you did a good job on mounting MOSFETs to heatsink or not (it is an important reliability factor).
Here is the sch. that will help you check your old MOSFETs. Never mind the current through them (the 150R resistors sets it at about 75mA) or measurement voltage (it doesn't have to be exactly +15V, anything from +10V upward will do). If you get a reading on the voltmeter of about 4V - MOSFET is OK.
Take care about polarity (N-channel vs. P-channel).
Attachments
Last edited:
Happened again...!
Hi Guys,
The same problem that happened two months ago happened again - Q6 in the right channel blew. I didn't "hear" it go - the thing just wouldn't work when I turned it on yesterday evening. (Muting circuits kept voltage from the speakers, happily...)
This evening I switched it out, and everything is A-OK. The trouble is that I still don't know what's causing it. There is no apparent difference between the heatsinking for this mosfet and the other three (which haven't blown - but this one has, twice...). It doesn't seem to get particularly hotter than the others - a few degrees difference between the mosfet case and the heatsink, although I haven't tried to measure it very carefully.
The only thing I can think of which may be different between the channels is the trimpot I put in for R15, at ZM's suggestion (post #98 above). The original value in the circuit is 47k, and the 100k trimpot I put in seems to have been set high - about 70k or so. (This probably happened when fiddling about with DC offset at some point...)
So what would the effect have been? Might this have had the effect of "stressing" Q6? any other ideas?
The amp is working just fine now, but it would be nice not to have to switch out a IRFP9240 every couple of months...
Cheers
Nigel
Hi Guys,
The same problem that happened two months ago happened again - Q6 in the right channel blew. I didn't "hear" it go - the thing just wouldn't work when I turned it on yesterday evening. (Muting circuits kept voltage from the speakers, happily...)
This evening I switched it out, and everything is A-OK. The trouble is that I still don't know what's causing it. There is no apparent difference between the heatsinking for this mosfet and the other three (which haven't blown - but this one has, twice...). It doesn't seem to get particularly hotter than the others - a few degrees difference between the mosfet case and the heatsink, although I haven't tried to measure it very carefully.
The only thing I can think of which may be different between the channels is the trimpot I put in for R15, at ZM's suggestion (post #98 above). The original value in the circuit is 47k, and the 100k trimpot I put in seems to have been set high - about 70k or so. (This probably happened when fiddling about with DC offset at some point...)
So what would the effect have been? Might this have had the effect of "stressing" Q6? any other ideas?
The amp is working just fine now, but it would be nice not to have to switch out a IRFP9240 every couple of months...
Cheers
Nigel
Hi Guys,
Sorry to bump this, but I'm still scratching my head over this repeated mosfet failure. I can't measure any real difference between the temperature of the case on any of the four mosfets. Here are some numbers. (Ambient temp tonight is 28 degrees.)
1. Putting the probe in one of the little semi-circular notches on each side (so it touches the metal part that's just visible) I measure 56-57 degrees on all four of them.
2. The plastic part of the mosfet cases is a little cooler (about 53-54 degrees), again, close to the same on all four.
3. Pushing the probe between the fins on the heatsinks so it touches right on the other side of where the mosfet is mounted I get 49 degrees on the right channel (which is the problem one) and 51 on the left. The difference doesn't seem significant to me, since the right channel side faces the window, so it gets the slight breeze we have tonight, but the left channel doesn't.
There doesn't seem to me to be any apparent reason for heat to be the culprit here, firstly because nothing seems *that* hot, but also because the problem mosfet doesn't show any different temp from the others.
Can I ask a couple of questions?
1. Can anyone suggest a temp to measure that might shed some light on this?
2. I asked above if a high setting on the trimpot in place of R15 might have had an effect, but no-one responded, so I'm afraid I'm going to repeat the question What difference would a higher value of R15 have? (Say, 70k...)
3. Is there any way the speaker protection circuit could be affecting things? (It's the P33 project from Rod Elliott's site, which seems to have been built by plenty of people). I can't imagine how, but I'm out of ideas at this point...
4. Any other ideas? It's possible the problem won't happen again, and I'll never figure it out, I suppose, but it's unlikely the same mosfet would blow twice just through chance...
Well thanks for any help
Cheers
Nigel
Sorry to bump this, but I'm still scratching my head over this repeated mosfet failure. I can't measure any real difference between the temperature of the case on any of the four mosfets. Here are some numbers. (Ambient temp tonight is 28 degrees.)
1. Putting the probe in one of the little semi-circular notches on each side (so it touches the metal part that's just visible) I measure 56-57 degrees on all four of them.
2. The plastic part of the mosfet cases is a little cooler (about 53-54 degrees), again, close to the same on all four.
3. Pushing the probe between the fins on the heatsinks so it touches right on the other side of where the mosfet is mounted I get 49 degrees on the right channel (which is the problem one) and 51 on the left. The difference doesn't seem significant to me, since the right channel side faces the window, so it gets the slight breeze we have tonight, but the left channel doesn't.
There doesn't seem to me to be any apparent reason for heat to be the culprit here, firstly because nothing seems *that* hot, but also because the problem mosfet doesn't show any different temp from the others.
Can I ask a couple of questions?
1. Can anyone suggest a temp to measure that might shed some light on this?
2. I asked above if a high setting on the trimpot in place of R15 might have had an effect, but no-one responded, so I'm afraid I'm going to repeat the question
What difference would a higher value of R15 have? (Say, 70k...) 3. Is there any way the speaker protection circuit could be affecting things? (It's the P33 project from Rod Elliott's site, which seems to have been built by plenty of people). I can't imagine how, but I'm out of ideas at this point...
4. Any other ideas? It's possible the problem won't happen again, and I'll never figure it out, I suppose, but it's unlikely the same mosfet would blow twice just through chance...
Well thanks for any help
Cheers
Nigel
Hi Nigel,
your temp. measurements show that you did nothing wrong in the proccess of mounting MOSFETs on the heatsink.
You should not expect from manufacturing procces to be perfect. You had two MOSFETs out of specs (probably belonging to same production series) and that's it - snit happens
your temp. measurements show that you did nothing wrong in the proccess of mounting MOSFETs on the heatsink.
You should not expect from manufacturing procces to be perfect. You had two MOSFETs out of specs (probably belonging to same production series) and that's it - snit happens
Hi Juma,
So you don't think the trimpot on R15 had anything to do with it? (It's the only thing that is really different between the two channels).
You may very well be right about the mosfets being out of spec and it being no more than that - They weren't from the same production series, though (one was a 9140, and the other a 9240, bought from different cities)
You are CERTAINLY right that s**t happens, however...
Thanks!
Nigel
So you don't think the trimpot on R15 had anything to do with it? (It's the only thing that is really different between the two channels).
You should not expect from manufacturing procces to be perfect. You had two MOSFETs out of specs (probably belonging to same production series) and that's it - snit happens
You may very well be right about the mosfets being out of spec and it being no more than that - They weren't from the same production series, though (one was a 9140, and the other a 9240, bought from different cities)
You are CERTAINLY right that s**t happens, however...
Thanks!
Nigel
It seems that you have the same standing current, same heatsinks and same temperature on both channels, so everything being the same ....
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.