Dick,
Is that Mosfet grading box a commercial product? Do you happen to have a schematic?
I did look at Variacs on eBay and they are really not that expensive. The killer is I used to own an electronic repair facility. When I sold it I only kept a solder station and not an oscilloscope or signal generator. We had some nice equipment, but I didn't think I would need it.
I think I am going to see about getting a set of matched N and P channel MOSFETS just in case. I have a strong suspicion that is the problem. I'll post what I find out.
Lindsay
Is that Mosfet grading box a commercial product? Do you happen to have a schematic?
I did look at Variacs on eBay and they are really not that expensive. The killer is I used to own an electronic repair facility. When I sold it I only kept a solder station and not an oscilloscope or signal generator. We had some nice equipment, but I didn't think I would need it.
I think I am going to see about getting a set of matched N and P channel MOSFETS just in case. I have a strong suspicion that is the problem. I'll post what I find out.
Lindsay
tincanear,
Yes I did install the 47k resistor and moved it to the replacement board when I got that from Musical Concepts. I am pretty methodical when it comes to doing this type of work. I have been over the connections multiple times and since there aren't that many connections, I am pretty sure they are right. I have left off the input and output connections so i only have B+, B- and ground from the supply and the three connections to each bank of MOSFETS. They are the same as the board that works.
My thinking is that since two boards from Musical Concepts have the same apparent problem, it is not the Musical Concepts boards. And the only other thing connected are the P and N channel MOSFETS.
The MC boards have the fuses on the board and the only thing fused is the power connection to the MOSFETS. With the fuses out of each board, I am going to compare voltages and see if I can figure out which bank of MOSFETS has a problem. Then I will try removing them one at a time to see if I can identify the bad component(s). With no power to the MOSFETS I am assuming that pulling them out of the circuit will not hurt anything. I will use a grounding strap and great care to keep static electricity at bay.
Lindsay
Yes I did install the 47k resistor and moved it to the replacement board when I got that from Musical Concepts. I am pretty methodical when it comes to doing this type of work. I have been over the connections multiple times and since there aren't that many connections, I am pretty sure they are right. I have left off the input and output connections so i only have B+, B- and ground from the supply and the three connections to each bank of MOSFETS. They are the same as the board that works.
My thinking is that since two boards from Musical Concepts have the same apparent problem, it is not the Musical Concepts boards. And the only other thing connected are the P and N channel MOSFETS.
The MC boards have the fuses on the board and the only thing fused is the power connection to the MOSFETS. With the fuses out of each board, I am going to compare voltages and see if I can figure out which bank of MOSFETS has a problem. Then I will try removing them one at a time to see if I can identify the bad component(s). With no power to the MOSFETS I am assuming that pulling them out of the circuit will not hurt anything. I will use a grounding strap and great care to keep static electricity at bay.
Lindsay
First off, if you have no load on the Mosfets, you do not need a large amperage fuse. You can use a fast blow 500ma fuse in all probability. Just do not exceed the power rating of the fuse! For testing this may give you some peace of mind.
Quiescent current in the amp is likely about 250 ma or less... for two devices.
Put a lightbulb of about 40 watts in series with the primary, and that will drop the B+ by a fair amount and provide series current limiting as well... Check the AC voltage at the input and see where it lands. It will need to be about >75% of AC mains for the amp to work. Need more voltage, bigger bulb.
Mosfet checker:
power supply, 10turn pot, voltmeter and ammeter. You can use a DVM for one or both meters if you like. Harbor Freight and others sell perfectly serviceable DVMs for $3.00... Ammeter should be probably like 0-2 amps or so... you'll need to be able to see about 100-200 ma fairly clearly... so maybe a shunt on a lower current meter is a good idea.
The voltmeter is set to read DC to the gate.
The 10T pot is a voltage divider off the main supply, to put 0- "X" volts on the gate. The main supply does NOT have to be equal to the usual operating voltage, it can be somewhat less, but 35volts or so is a good target. 24vdc is ok too... nothing fancy, or high current needed, even for power Mosfetters.
You turn the volts up, the ammeter shows the current through the device. With a SET DC on the gate, you look for devices that are close in terms of ammeter reading for a given gate voltage.
Switches reverse the polarity.
This permits matching to the *bias point* of devices. It does NOT assure matched GAINS.
It is usually sufficient to match bias point. The voltage used for testing should permit biasing to some value of current near what the quiesent bias current will be for the amp itself.
_-_-bear
Quiescent current in the amp is likely about 250 ma or less... for two devices.
Put a lightbulb of about 40 watts in series with the primary, and that will drop the B+ by a fair amount and provide series current limiting as well... Check the AC voltage at the input and see where it lands. It will need to be about >75% of AC mains for the amp to work. Need more voltage, bigger bulb.
Mosfet checker:
power supply, 10turn pot, voltmeter and ammeter. You can use a DVM for one or both meters if you like. Harbor Freight and others sell perfectly serviceable DVMs for $3.00... Ammeter should be probably like 0-2 amps or so... you'll need to be able to see about 100-200 ma fairly clearly... so maybe a shunt on a lower current meter is a good idea.
The voltmeter is set to read DC to the gate.
The 10T pot is a voltage divider off the main supply, to put 0- "X" volts on the gate. The main supply does NOT have to be equal to the usual operating voltage, it can be somewhat less, but 35volts or so is a good target. 24vdc is ok too... nothing fancy, or high current needed, even for power Mosfetters.
You turn the volts up, the ammeter shows the current through the device. With a SET DC on the gate, you look for devices that are close in terms of ammeter reading for a given gate voltage.
Switches reverse the polarity.
This permits matching to the *bias point* of devices. It does NOT assure matched GAINS.
It is usually sufficient to match bias point. The voltage used for testing should permit biasing to some value of current near what the quiesent bias current will be for the amp itself.
_-_-bear
Bear,
Thanks for the explanation on how to set up a process to match the MOSFETS. The bias per the instructions for the new boards is 330ma. So I will assume that means the target for an individual MOSFET is 110ma since there are three devices in parallel.
What range of gate voltage would be appropriate? I assume just a few volts would be sufficient. Based on mmerig's comments, the maximum should be less than 15 volts. Could the gate voltage be supplied by one or more batteries without the pot or does it need to be variable? I have a wall wart rated at 24 VDC and 500ma. Would that be sufficient to supply power to the MOSFET under test?
Thanks for your help.
Lindsay
Thanks for the explanation on how to set up a process to match the MOSFETS. The bias per the instructions for the new boards is 330ma. So I will assume that means the target for an individual MOSFET is 110ma since there are three devices in parallel.
What range of gate voltage would be appropriate? I assume just a few volts would be sufficient. Based on mmerig's comments, the maximum should be less than 15 volts. Could the gate voltage be supplied by one or more batteries without the pot or does it need to be variable? I have a wall wart rated at 24 VDC and 500ma. Would that be sufficient to supply power to the MOSFET under test?
Thanks for your help.
Lindsay
Update:
I replaced the drive board with the original Hafler board and got more normal results. I did have DC offset - the voltages were as follows:
Current Draw: 300ma
DC offset: Initially -2.3 VDC after 10 minutes -2.1 VDC
B-: -94.7
B+: 94.7
Negative drive: -2.3 VDC
Positive drive: -1.4 VDC
This was with the DC offset adjustment set fully clockwise. Fully counter clockwise was -7.15 VDC and mid point was -4.65 VDC.
I noticed that the MOSFETS on the good channel and the negative rail of the bad channel were all alike with the Hafler part number on them. The MOSFETS on the positive rail of the bad channel are marked differently with K176. I am the original owner and have not replaced the outputs. Could there be some slight imbalance between the MOSFETS that the original board could handle, but the Musical Concepts board cannot? All of the MOSFETS have the Hafler grading number on them.
Thanks for any suggestions.
Lindsay
I replaced the drive board with the original Hafler board and got more normal results. I did have DC offset - the voltages were as follows:
Current Draw: 300ma
DC offset: Initially -2.3 VDC after 10 minutes -2.1 VDC
B-: -94.7
B+: 94.7
Negative drive: -2.3 VDC
Positive drive: -1.4 VDC
This was with the DC offset adjustment set fully clockwise. Fully counter clockwise was -7.15 VDC and mid point was -4.65 VDC.
I noticed that the MOSFETS on the good channel and the negative rail of the bad channel were all alike with the Hafler part number on them. The MOSFETS on the positive rail of the bad channel are marked differently with K176. I am the original owner and have not replaced the outputs. Could there be some slight imbalance between the MOSFETS that the original board could handle, but the Musical Concepts board cannot? All of the MOSFETS have the Hafler grading number on them.
Thanks for any suggestions.
Lindsay
Lindsay,
The current draw of 333ma would be divided by the number of Mosfets, as you wrote.
The wallwart would work ok, but you'd likely have to add some outboard filter caps...
Fixed bias voltage is probably not a good idea, since you want to find Mosfets that all fall into the range that you are looking for. Eg. 111ma. with some fixed gate voltage. If you supply 1.5vdc only, you might find the Mosfets will want to draw way too much current or not enough current. Best to use a pot, and maybe a 3.0 or 4.5 battery set up if you wish to use batts.
There are two drive boards, one for each channel.
Do both channels act the same?
Sounds like something else is amiss in your board. The "drive" as you call it is not equal, which suggests that there is an offset being caused by some other sort of problem...
Also, does the thing amplify without distortion? Notwithstanding the offset? A scope is what you want/need to see... although a soundcard and appropriate freeware program will let you "see" enough to make this determination too...
The K176 is a good part - if they all have Hafler "numbers" coding the bins for the parts, then ur fine.
_-_-bear
The current draw of 333ma would be divided by the number of Mosfets, as you wrote.
The wallwart would work ok, but you'd likely have to add some outboard filter caps...
Fixed bias voltage is probably not a good idea, since you want to find Mosfets that all fall into the range that you are looking for. Eg. 111ma. with some fixed gate voltage. If you supply 1.5vdc only, you might find the Mosfets will want to draw way too much current or not enough current. Best to use a pot, and maybe a 3.0 or 4.5 battery set up if you wish to use batts.
There are two drive boards, one for each channel.
Do both channels act the same?
Sounds like something else is amiss in your board. The "drive" as you call it is not equal, which suggests that there is an offset being caused by some other sort of problem...
Also, does the thing amplify without distortion? Notwithstanding the offset? A scope is what you want/need to see... although a soundcard and appropriate freeware program will let you "see" enough to make this determination too...
The K176 is a good part - if they all have Hafler "numbers" coding the bins for the parts, then ur fine.
_-_-bear
Bear,
Thanks for your reply, I appreciate your taking time to help me.
Before starting to do the upgrade, the amplifier was used on an almost daily basis driving Martin Logan SL3s without problem or audible distortion. However, the left channel (now the bad channel) had DC offset with the speaker connected of about 0.3 VDC when adjusted to the minimum.
I installed two Musical Concepts drive boards, one in each channel. The right channel is working fine. It can be set to 330ma bias per the instructions. I did not measure DC offset yet, but the protection circuit pulls in as though the right channel DC offset is within spec. So I believe it is working correctly. The problem is only with the left channel.
I am not sure what else could be wrong. The MOSFETS have resistors to their gates, which all check out. There are new caps from the rail to the output of one each of the P and N channel MOSFETS and a film cap from the positive rail connection on one each of the P & N MOSFETS to ground. These caps were supplied with the new board and seem to be OK. I believe they are to prevent oscillation/stability problems. There is nothing else connected except the drive board.
I assumed the DC offset problem was in the old drive board and would be corrected with the new board. I did not put any load on the output when measuring the DC offset, so I don't know if it has actually increased from the 0.3 VDC to the current 2.1 VDC. I do not want to fix the old board and only put it in to see if something had changed and to confirm the MOSFETS were still good.
Is there any failure mode of the MOSFETS that would cause DC offset? For example can the parameters change such that the MOSFET works but requires a different bias voltage or has a much different gain?
I can probably set up an oscilloscope with sound card (are special probes needed?). What would I look for? I did not get a schematic with the Musical Concept boards, so I would not know where test points are other than the power and MOSFET connections. I am hoping to hear back from Musical Concepts on Monday, but I think they are going to want me to ship my amp to them, and I would rather not do that. It isn't DIY if they do it.
In any case thanks for the info. I think my next step will be to build the MOSFET matching circuit you described and see if the devices appear to be matched.
Lindsay
Thanks for your reply, I appreciate your taking time to help me.
Before starting to do the upgrade, the amplifier was used on an almost daily basis driving Martin Logan SL3s without problem or audible distortion. However, the left channel (now the bad channel) had DC offset with the speaker connected of about 0.3 VDC when adjusted to the minimum.
I installed two Musical Concepts drive boards, one in each channel. The right channel is working fine. It can be set to 330ma bias per the instructions. I did not measure DC offset yet, but the protection circuit pulls in as though the right channel DC offset is within spec. So I believe it is working correctly. The problem is only with the left channel.
I am not sure what else could be wrong. The MOSFETS have resistors to their gates, which all check out. There are new caps from the rail to the output of one each of the P and N channel MOSFETS and a film cap from the positive rail connection on one each of the P & N MOSFETS to ground. These caps were supplied with the new board and seem to be OK. I believe they are to prevent oscillation/stability problems. There is nothing else connected except the drive board.
I assumed the DC offset problem was in the old drive board and would be corrected with the new board. I did not put any load on the output when measuring the DC offset, so I don't know if it has actually increased from the 0.3 VDC to the current 2.1 VDC. I do not want to fix the old board and only put it in to see if something had changed and to confirm the MOSFETS were still good.
Is there any failure mode of the MOSFETS that would cause DC offset? For example can the parameters change such that the MOSFET works but requires a different bias voltage or has a much different gain?
I can probably set up an oscilloscope with sound card (are special probes needed?). What would I look for? I did not get a schematic with the Musical Concept boards, so I would not know where test points are other than the power and MOSFET connections. I am hoping to hear back from Musical Concepts on Monday, but I think they are going to want me to ship my amp to them, and I would rather not do that. It isn't DIY if they do it.
In any case thanks for the info. I think my next step will be to build the MOSFET matching circuit you described and see if the devices appear to be matched.
Lindsay
LMH,
Have you tried swapping MOSFETS from the good channel to the bad one to verify whether the problem is with the MOSFETs on the bad channel?
I have swapped MOSFETs around to do just this. Just remember to pull all the fuses to the good channel so it is not working. Also, a few extra mica insulating washers might be needed as sometimes they get cracked when removing MOSFETs that have not been taken out for several years.
Have you tried swapping MOSFETS from the good channel to the bad one to verify whether the problem is with the MOSFETs on the bad channel?
I have swapped MOSFETs around to do just this. Just remember to pull all the fuses to the good channel so it is not working. Also, a few extra mica insulating washers might be needed as sometimes they get cracked when removing MOSFETs that have not been taken out for several years.
Each channel only needs ONE mosfet per rail to actually function.
The other Mosfets serve to decrease the output Z and to increase the current capability of the amp.
As I said before, if the side with the DC offset produces undistorted ouput, the problem is different than if it produces distorted output - you'd look for different things.
BUT, given that the problem existed before swapping boards, then the problem is clearly NOT in the board!
So, that leaves us with things like the Mosfets and the connections/resistors therein.
Perhaps an emitter resistor is OPEN or High on one or more Mosfets?
That would cause 1/2 of the channel to bias on, while only 1 or 2 mosfets on the other rail would bias on, therefore you'd get an offset that you could not overcome with the adjustment...
etc...
Scope using sound card? Divider so that you can not exceed the max input voltage to the sound card - or else some sort of diode clipper/zener circuit for protection...
_-_-bear
The other Mosfets serve to decrease the output Z and to increase the current capability of the amp.
As I said before, if the side with the DC offset produces undistorted ouput, the problem is different than if it produces distorted output - you'd look for different things.
BUT, given that the problem existed before swapping boards, then the problem is clearly NOT in the board!
So, that leaves us with things like the Mosfets and the connections/resistors therein.
Perhaps an emitter resistor is OPEN or High on one or more Mosfets?
That would cause 1/2 of the channel to bias on, while only 1 or 2 mosfets on the other rail would bias on, therefore you'd get an offset that you could not overcome with the adjustment...
etc...
Scope using sound card? Divider so that you can not exceed the max input voltage to the sound card - or else some sort of diode clipper/zener circuit for protection...
_-_-bear
LMHAudio,
you can also test the driver boards by temporarily connecting the "output" terminal to each of the FET drive signals with 47K resistors (at least on halfer PC-19 boards, this closes the feedback loop) and measuring the "output" voltage. if the boards are working, then the offset pot should allow you to adjust the output voltage to zero (plus minus a few mV of noise).
also, in an earlier post, you mentioned that only two of the three LED's were lit (specific on which two may be helpful to John H at MC to diagnose the problem without having to send your amp in).
you can also test the driver boards by temporarily connecting the "output" terminal to each of the FET drive signals with 47K resistors (at least on halfer PC-19 boards, this closes the feedback loop) and measuring the "output" voltage. if the boards are working, then the offset pot should allow you to adjust the output voltage to zero (plus minus a few mV of noise).
also, in an earlier post, you mentioned that only two of the three LED's were lit (specific on which two may be helpful to John H at MC to diagnose the problem without having to send your amp in).
PA3 boards, LHMAudio
LHMAudio,
if you haven't already began dinking with the Mosfets (i don't think its a bad mosfet, see below), i recommending testing the driver boards per my previous post first.
also, on the PA3's for DH200,220,500 make sure both small jumpers tying E3A to E4A and E11A to E12A are present (no jumper if used with XL600, XL280 that have separate rail for driver board).
regarding the Mosfets:
if a fet shorted, the rail current would run up and blow the rail fuse (or the fuse on the other rail)
if one fet was open or disconnected, causing a large difference in transconductance between N and P Mosfets, then the driver board would see a large voltage on FB inverting input, and immediately shift the Mosfet operating points to balance the drain currents , and null out the output DC offset to within a volt or so of zero.
if they worked when retested with original hafler driver boards, then they are probably OK.
Let us know what the results are.
thanks,
Bruce
LHMAudio,
if you haven't already began dinking with the Mosfets (i don't think its a bad mosfet, see below), i recommending testing the driver boards per my previous post first.
also, on the PA3's for DH200,220,500 make sure both small jumpers tying E3A to E4A and E11A to E12A are present (no jumper if used with XL600, XL280 that have separate rail for driver board).
regarding the Mosfets:
if a fet shorted, the rail current would run up and blow the rail fuse (or the fuse on the other rail)
if one fet was open or disconnected, causing a large difference in transconductance between N and P Mosfets, then the driver board would see a large voltage on FB inverting input, and immediately shift the Mosfet operating points to balance the drain currents , and null out the output DC offset to within a volt or so of zero.
if they worked when retested with original hafler driver boards, then they are probably OK.
Let us know what the results are.
thanks,
Bruce
tincanear,
Thanks for your reply. I have not done anything since my last post, but I did speak with John at Musical Concepts. He is sending me 3 each of the P and N channel MOSFETS on my assumption that there is something flaky with at least one of the MOSFETS.
I will try your test on the Hafler board tonight. If I can adjust it to zero DC offset that will probably mean there is some kind of problem with one of the MOSFETS. In that case I will remove all but one each P and N channel MOSFET and check to see if I can eliminate the DC offset. If I can find a pair where the DC offset goes away, I think I can be pretty sure that there is a problem with one or more MOSFETS.
I am at work so I cannot look for jumpers from E3A to E4A and from E11A to E12A. However, I would assume they are there as the instructions only say to remove it for the XL-280/600 and the right channel board worked without adding a jumper. But I will check when I get home tonight.
I suspect that both Musical Concepts boards are not working correctly and may have been damaged by whatever is causing the DC offset. I don't want to install new MOSFETS and have the damaged MUSICAL Concepts board cause a new problem with the MOSFETS. So if I can find a pair of MOSFETS that appear to be good then I will try the MC boards and see if they work.
Lindsay
Thanks for your reply. I have not done anything since my last post, but I did speak with John at Musical Concepts. He is sending me 3 each of the P and N channel MOSFETS on my assumption that there is something flaky with at least one of the MOSFETS.
I will try your test on the Hafler board tonight. If I can adjust it to zero DC offset that will probably mean there is some kind of problem with one of the MOSFETS. In that case I will remove all but one each P and N channel MOSFET and check to see if I can eliminate the DC offset. If I can find a pair where the DC offset goes away, I think I can be pretty sure that there is a problem with one or more MOSFETS.
I am at work so I cannot look for jumpers from E3A to E4A and from E11A to E12A. However, I would assume they are there as the instructions only say to remove it for the XL-280/600 and the right channel board worked without adding a jumper. But I will check when I get home tonight.
I suspect that both Musical Concepts boards are not working correctly and may have been damaged by whatever is causing the DC offset. I don't want to install new MOSFETS and have the damaged MUSICAL Concepts board cause a new problem with the MOSFETS. So if I can find a pair of MOSFETS that appear to be good then I will try the MC boards and see if they work.
Lindsay
LHMAudio
LHMAudio,
after re-reading the thread, i think i originally misunderstood the problem--
i'm assuming that you have at least one pair of MC boards that work, except that one channel has a volt or two of DC offset that cannot be nulled out by the adjustment pot.
Make sure the input ground (E2) or E3 depending on the setup is tied to the star ground, or this can cause some DC offset.
On the PA3's that i have (they're not the latest PA3D, but i think most of the circuit is the same), all 3 LED's are lit when the board is powered up. Gate voltages (E6 and E8) should be as shown in the MC modification manual plus-minus about 500mV-- if not, then its probably a bad Mosfet as others on this thread have indicated.
If you wind up changing the mosfets, replace all N's as a set, and all P's as a set-- otherwise, the Vgs variation between your new fets and the old ones will cause uneven biasing.
LHMAudio,
after re-reading the thread, i think i originally misunderstood the problem--
i'm assuming that you have at least one pair of MC boards that work, except that one channel has a volt or two of DC offset that cannot be nulled out by the adjustment pot.
Make sure the input ground (E2) or E3 depending on the setup is tied to the star ground, or this can cause some DC offset.
On the PA3's that i have (they're not the latest PA3D, but i think most of the circuit is the same), all 3 LED's are lit when the board is powered up. Gate voltages (E6 and E8) should be as shown in the MC modification manual plus-minus about 500mV-- if not, then its probably a bad Mosfet as others on this thread have indicated.
If you wind up changing the mosfets, replace all N's as a set, and all P's as a set-- otherwise, the Vgs variation between your new fets and the old ones will cause uneven biasing.
Bruce,
I have one MC board that works. When I initially put the MC board in the left channel, only two LEDs were lit. The gate voltages (N and P) and output voltage were all equal to the negative rail less about 1.5 V. That is E6, E8 and E9 were all at -95.7 VDC and the rails were at +-97.2 VDC. There was no current draw through the fuse to the MOSFETS.
MC sent me a second board which I connected, but left the fuses out of. The result was the same. So the second board did not work even though there was no power to the MOSFETS.
My MC instructions give no information about the gate voltage, only bias current. But the gate voltage is -95.7 VDC so it really is not an issue yet.
I also connected the first board to power with no connections to the MOSFETS and only two LEDs light up. That's why I think I now have two bad MC boards and bad MOSFET(S). John is sending matched sets of MOSFETS so I will change them out as a complete set. But since the original Hafler board seems to work, I thought I might use it to test the MOSFETS I have one at a time.
Lindsay
I have one MC board that works. When I initially put the MC board in the left channel, only two LEDs were lit. The gate voltages (N and P) and output voltage were all equal to the negative rail less about 1.5 V. That is E6, E8 and E9 were all at -95.7 VDC and the rails were at +-97.2 VDC. There was no current draw through the fuse to the MOSFETS.
MC sent me a second board which I connected, but left the fuses out of. The result was the same. So the second board did not work even though there was no power to the MOSFETS.
My MC instructions give no information about the gate voltage, only bias current. But the gate voltage is -95.7 VDC so it really is not an issue yet.
I also connected the first board to power with no connections to the MOSFETS and only two LEDs light up. That's why I think I now have two bad MC boards and bad MOSFET(S). John is sending matched sets of MOSFETS so I will change them out as a complete set. But since the original Hafler board seems to work, I thought I might use it to test the MOSFETS I have one at a time.
Lindsay
LHM,
near the back of the MC modifications manual for PA3C, there's a picture of the PCB layout and it has some nominal voltages listed. if both e6 and e8 are -95v, then theres a prob with the PA3. from looking at the pcb layout in the MC manual, the two leds near center of board (next to drivers on heatsinks) are for cascode, and the third led near the top edge of the board is for the current source for front end. i would check to make sure the positive rail is getting to the PA3 by carefully measuring the voltage at the on-board fuse holder.
like you said, i would hesitate to connect the new fets to the MC board until its problems can be resolved (otherwise, the new mosfets may be damaged by excessive gate voltage (spec max vgs is 20v, if i remember)
near the back of the MC modifications manual for PA3C, there's a picture of the PCB layout and it has some nominal voltages listed. if both e6 and e8 are -95v, then theres a prob with the PA3. from looking at the pcb layout in the MC manual, the two leds near center of board (next to drivers on heatsinks) are for cascode, and the third led near the top edge of the board is for the current source for front end. i would check to make sure the positive rail is getting to the PA3 by carefully measuring the voltage at the on-board fuse holder.
like you said, i would hesitate to connect the new fets to the MC board until its problems can be resolved (otherwise, the new mosfets may be damaged by excessive gate voltage (spec max vgs is 20v, if i remember)
Bruce,
The PA3D manual I received with the MC boards did not include the board layout. Fortunately I had previously downloaded the PA3C manual so I have the layout.
When you say the maximum gate voltage is 20 V do you mean with respect to the output? So when the gate voltages were both -95.2 and the output voltage was also -95.2, this would not damage the MOSFETS, correct?
Otherwise, how could the old board seem to be working, but with a -2.1 volt DC offset? Or does that offset mean that the MOSFETS are bad?
On the PA3D board, should the current source LED light if there were no connections to the MOSFETS? That is there is only power to the PA3D and no other connections. Thanks.
Lindsay
The PA3D manual I received with the MC boards did not include the board layout. Fortunately I had previously downloaded the PA3C manual so I have the layout.
When you say the maximum gate voltage is 20 V do you mean with respect to the output? So when the gate voltages were both -95.2 and the output voltage was also -95.2, this would not damage the MOSFETS, correct?
Otherwise, how could the old board seem to be working, but with a -2.1 volt DC offset? Or does that offset mean that the MOSFETS are bad?
On the PA3D board, should the current source LED light if there were no connections to the MOSFETS? That is there is only power to the PA3D and no other connections. Thanks.
Lindsay
Lindsay,
with only power and ground (both E2 and E7) connected to the board, and tying both E6 and E8 to e9 with separate 47k resistors to ensure that the feedback loop is closed, the voltages on e6, e8, and e9 should be somewhere close to zero and not near the negative supply rail as you're seeing.
i wonder if there was a pcb stuffing error (missing parts????) on the two bad boards that you have? can you post a pic of one of the boards, both front and back sides?
regarding the mosfets, if all 3 signals E6, E8, E9, were at -95V, then i think to mosfets are OK.
the Vgs 20V spec refers to the difference between the gate drives and the output (to speaker) terminal (connected to source pin of both n and p mosfets.
if you like, you can test them a pair at a time with the hafler boards to test--- if you can null the offset on hafler boards, then the fet pair under test are OK. other pairs should have about the same offset for a given pot setting since they're matched to each other.
-Bruce
with only power and ground (both E2 and E7) connected to the board, and tying both E6 and E8 to e9 with separate 47k resistors to ensure that the feedback loop is closed, the voltages on e6, e8, and e9 should be somewhere close to zero and not near the negative supply rail as you're seeing.
i wonder if there was a pcb stuffing error (missing parts????) on the two bad boards that you have? can you post a pic of one of the boards, both front and back sides?
regarding the mosfets, if all 3 signals E6, E8, E9, were at -95V, then i think to mosfets are OK.
the Vgs 20V spec refers to the difference between the gate drives and the output (to speaker) terminal (connected to source pin of both n and p mosfets.
if you like, you can test them a pair at a time with the hafler boards to test--- if you can null the offset on hafler boards, then the fet pair under test are OK. other pairs should have about the same offset for a given pot setting since they're matched to each other.
-Bruce
Bruce,
The test with only power was done on the MC boards with no resistors between E6 and E8 or E9 and E8. It was just to see if with power and nothing else connected the current source LED would light up. It may not have been indicative of anything.
I can take a picture of the boards tomorrow and post them.
I had not thought about the fact that all pairs of single P and N MOSFETS should have the same DC offset. That should help me find out if there is a problem pretty quickly.
I have done a lot of trouble shooting of computer systems in my life and my gut tells me that there is a bad MOSFET and that both MC boards are bad. It is hard to believe that MC would send three boards, two with missing parts or otherwise not working. Since there was DC offset in the first place I suspect that the Hafler board is possibly more robust than the MC board and could deal with the MOSFET problem. I guess I'll find out this weekend.
I will post pictures and what I learn over the weekend. Thanks for all your suggestions and help.
Lindsay
The test with only power was done on the MC boards with no resistors between E6 and E8 or E9 and E8. It was just to see if with power and nothing else connected the current source LED would light up. It may not have been indicative of anything.
I can take a picture of the boards tomorrow and post them.
I had not thought about the fact that all pairs of single P and N MOSFETS should have the same DC offset. That should help me find out if there is a problem pretty quickly.
I have done a lot of trouble shooting of computer systems in my life and my gut tells me that there is a bad MOSFET and that both MC boards are bad. It is hard to believe that MC would send three boards, two with missing parts or otherwise not working. Since there was DC offset in the first place I suspect that the Hafler board is possibly more robust than the MC board and could deal with the MOSFET problem. I guess I'll find out this weekend.
I will post pictures and what I learn over the weekend. Thanks for all your suggestions and help.
Lindsay