Hey folks, I hope you are all doing well. I know it's been a long time since I've stopped in. Life has been incredibly busy.
I what I hope will be a quick question, but the past is lengthy and I apologize in advance.
I have a PG elite.5 that a client sent me. When I got it, capacitor C515A had exploded in blaze of glory. I cleaned up the board, replaced the cap, tested all the transistors and their drivers and they all checked out fine. I powered it up and it draws about 3.8a on a current limited supply and input voltage sags to about 8.5V, I'd normally not expect an amp to even power on with the supply voltage dragged down that far, but it does. All channels come up and produce audio. If I start it on a unlimited supply, it draws about 2.5 amps and all channels produce audio.
Here is the weird thing though. And I've never seen an amp do this. If I power it on a current limited supply, and then while it's running, switch it to an unlimited supply, it begins drawing about 12.5A. If I start it on an unlimited supply and briefly cause the input voltage to sag, it starts drawing 12.5a and while in the sink it will eventually thermal.
Things I've found while troubleshooting:
1) With the rectifiers removed from the class d power supply, it doesn't exhibit any of the above issues.
2) All of the output filter caps as well as the final inductor have been removed and tested fine
3) When it's started with a current limit, the class d section oscillates at 67kHz instead of 45kHz. I don't recall for certain, but I believe when I make it exhibit the issue with input voltage dash that it also starts oscillating at 67kHz. I can verify if it's relevant.
4) when started with a current limit, the low side drive supply only produces 6V above VNN, not 12V.
5) PWM drive for the power supply looks completely normal with no shot through under all test conditions, however the drive is roughly half the amplitude when it's having issues. I suspect that's just a factor of the voltage sag from the current limit, but I'm not 100% sure. The PWM seems to be locked at 49% duty cycle regardless of my test conditions, but it's possible I've never raised the supply voltage high enough to get the PWM to roll back duty cycle.
Things I've tried to do:
1) I replaced the class d output fets. The schematic calls for IRF640 and it had been previously repaired with mismatch IRF640Ns.
2) I replaced r class d driver bjts
3) I replaced the two mmbta42 low side driver bjts on the driver card.
I'm really stumped on this thing. I don't do a lot of class d as it is, but I normally can hold my own. I still have no idea what caused the capacitor to explode in the first place. I feel like it shouldn't be powering up at all with the voltage so low. I have schematics for similar designs and they all have undervoltage protection on the PWM board. It appears according to the aamp schematic, that this one does not, however the schematic I have and the amp have several noticable differences.
The tl;dr is as follows:
On a current limited supply, the amp runs fine and draws about 3.8a. On an unlimited supply it runs fine and draws about 2.5a. If it starts on current limited and is switched to unlimited while it's running, it starts drawing 12.5a. If it's started on an unlimited supply and the voltage sags, even very briefly, it starts drawing 12.5a.
Thanks in advance folks!
Jason
I what I hope will be a quick question, but the past is lengthy and I apologize in advance.
I have a PG elite.5 that a client sent me. When I got it, capacitor C515A had exploded in blaze of glory. I cleaned up the board, replaced the cap, tested all the transistors and their drivers and they all checked out fine. I powered it up and it draws about 3.8a on a current limited supply and input voltage sags to about 8.5V, I'd normally not expect an amp to even power on with the supply voltage dragged down that far, but it does. All channels come up and produce audio. If I start it on a unlimited supply, it draws about 2.5 amps and all channels produce audio.
Here is the weird thing though. And I've never seen an amp do this. If I power it on a current limited supply, and then while it's running, switch it to an unlimited supply, it begins drawing about 12.5A. If I start it on an unlimited supply and briefly cause the input voltage to sag, it starts drawing 12.5a and while in the sink it will eventually thermal.
Things I've found while troubleshooting:
1) With the rectifiers removed from the class d power supply, it doesn't exhibit any of the above issues.
2) All of the output filter caps as well as the final inductor have been removed and tested fine
3) When it's started with a current limit, the class d section oscillates at 67kHz instead of 45kHz. I don't recall for certain, but I believe when I make it exhibit the issue with input voltage dash that it also starts oscillating at 67kHz. I can verify if it's relevant.
4) when started with a current limit, the low side drive supply only produces 6V above VNN, not 12V.
5) PWM drive for the power supply looks completely normal with no shot through under all test conditions, however the drive is roughly half the amplitude when it's having issues. I suspect that's just a factor of the voltage sag from the current limit, but I'm not 100% sure. The PWM seems to be locked at 49% duty cycle regardless of my test conditions, but it's possible I've never raised the supply voltage high enough to get the PWM to roll back duty cycle.
Things I've tried to do:
1) I replaced the class d output fets. The schematic calls for IRF640 and it had been previously repaired with mismatch IRF640Ns.
2) I replaced r class d driver bjts
3) I replaced the two mmbta42 low side driver bjts on the driver card.
I'm really stumped on this thing. I don't do a lot of class d as it is, but I normally can hold my own. I still have no idea what caused the capacitor to explode in the first place. I feel like it shouldn't be powering up at all with the voltage so low. I have schematics for similar designs and they all have undervoltage protection on the PWM board. It appears according to the aamp schematic, that this one does not, however the schematic I have and the amp have several noticable differences.
The tl;dr is as follows:
On a current limited supply, the amp runs fine and draws about 3.8a. On an unlimited supply it runs fine and draws about 2.5a. If it starts on current limited and is switched to unlimited while it's running, it starts drawing 12.5a. If it's started on an unlimited supply and the voltage sags, even very briefly, it starts drawing 12.5a.
Thanks in advance folks!
Jason
With the rectifiers out of the circuit, if you reduce the supply voltage to as low as thew amp will remain on, does it start to draw excessive current?
Did you try asking this on the PhoenixPhorum?
Is the supply regulated?
Did you try asking this on the PhoenixPhorum?
Is the supply regulated?
I did not check, but I'll do that Monday when I'm back to the shop.
I haven't asked on the phorum. I'm normally the person answering the questions there. I did reach out to a few other techs from my FB tech group who do a lot more class d stuff than I.
It appears the power supply is is fixed at 49% duty cycle which is 100% as far as a sg3525 is concerned. I've not seen the the duty cycle change regardless of power input fluctuations. The class d and ab supplies are both run from the same pwm.
I'll investigate more on Monday. Thank you for sharing any thoughts you might have. Do you have any suspicion as to why that cap might explode? I just realized the component designation for that cap might be less than helpful. The class d section is pretty much identical to the audio-q aq1200 and the cap is c208 in that amp.
Thank you,
Jason
I haven't asked on the phorum. I'm normally the person answering the questions there. I did reach out to a few other techs from my FB tech group who do a lot more class d stuff than I.
It appears the power supply is is fixed at 49% duty cycle which is 100% as far as a sg3525 is concerned. I've not seen the the duty cycle change regardless of power input fluctuations. The class d and ab supplies are both run from the same pwm.
I'll investigate more on Monday. Thank you for sharing any thoughts you might have. Do you have any suspicion as to why that cap might explode? I just realized the component designation for that cap might be less than helpful. The class d section is pretty much identical to the audio-q aq1200 and the cap is c208 in that amp.
Thank you,
Jason
I'm sorry for the late reply, I did more troubleshooting over the last couple of days. For clarity and to save me typing "in the class d section" repeatedly, all of the troubleshooting below was performed only on the class d side of the amp. I basically ignored the 4 ab channels.
With the rectifiers removed I was able to get the amp to power on all of the way down to 8V5dc. Current draw was in the neighborhood of half an amp while voltage was that low. Any lower and the lights go out, however, given enough time sitting at 7V, the protection relay eventually engages. I did not watch rail voltages while doing this and I kind of wish I had, but I think the current draw was the most important variable.
After that did not yield any results, I went on to rebuild the ±15V and +12 above VNN supplies. The other folks I mentioned both thought it possible it was an issue with the auxiliary rails. The ±15V supply had some seriously baked caps I planned to change anyhow, so I just did them all. All of the caps tested at spec and replacing them didn't help.
I then put the class d driver card in a socket in case I'll have to remove it several more times. I replaced both tl072s and I tested every transistor, diode, and resistor on the card with nothing showing faulty. I think I already mentioned that I replaced the 2 mmbta42s that the guide says fail and sometimes test good.
Then, I removed all the power supply fets and installed 4 known good used parts just for further testing. I also replaced the power supply buffer/driver bjts just in case. I wanted to get some thermal images of the amp and since I don't have it in the sink I swapped out the fets to keep from blowing the original ones.
Thermal images are showing what seems to be random heating of one of the power supply fets and the low side outputs get scalding hot while it's drawing high current. Even when it's not pulling tons of current the low side is noticeably hotter than anything else.
I'm going to do a bit more testing this morning, but I'm coming close to cutting my losses and moving on. I normally turn down this stuff from aamp because every one I've worked on has been ridiculous. I only took it on because a friend in the Phoenix Gold community asked me to help.
Thanks,
Jason
With the rectifiers removed I was able to get the amp to power on all of the way down to 8V5dc. Current draw was in the neighborhood of half an amp while voltage was that low. Any lower and the lights go out, however, given enough time sitting at 7V, the protection relay eventually engages. I did not watch rail voltages while doing this and I kind of wish I had, but I think the current draw was the most important variable.
After that did not yield any results, I went on to rebuild the ±15V and +12 above VNN supplies. The other folks I mentioned both thought it possible it was an issue with the auxiliary rails. The ±15V supply had some seriously baked caps I planned to change anyhow, so I just did them all. All of the caps tested at spec and replacing them didn't help.
I then put the class d driver card in a socket in case I'll have to remove it several more times. I replaced both tl072s and I tested every transistor, diode, and resistor on the card with nothing showing faulty. I think I already mentioned that I replaced the 2 mmbta42s that the guide says fail and sometimes test good.
Then, I removed all the power supply fets and installed 4 known good used parts just for further testing. I also replaced the power supply buffer/driver bjts just in case. I wanted to get some thermal images of the amp and since I don't have it in the sink I swapped out the fets to keep from blowing the original ones.
Thermal images are showing what seems to be random heating of one of the power supply fets and the low side outputs get scalding hot while it's drawing high current. Even when it's not pulling tons of current the low side is noticeably hotter than anything else.
I'm going to do a bit more testing this morning, but I'm coming close to cutting my losses and moving on. I normally turn down this stuff from aamp because every one I've worked on has been ridiculous. I only took it on because a friend in the Phoenix Gold community asked me to help.
Thanks,
Jason
What's the amplitude of the high and low-side drive signals at the output transistors?
Did you use 640s or 640Ns as outputs? The diagrams I have call for 640Ns.
What's the oscillation frequency of the output stage?
Did you use 640s or 640Ns as outputs? The diagrams I have call for 640Ns.
What's the oscillation frequency of the output stage?
Perry, I sent you an email, but I'll reply with answers to the above as soon as I get a free.
Thank you
Thank you
Ok, I'll do my best here.
With a 2R current limiter the high side drive is 3V3 referenced to VPP. Low side drive is 6V5 referenced to VNN. Oscillation frequency is 69kHz. Rails are ±29Vdc and symmetrical.
With no current limit the high side drive is 10V5 referenced to VPP. Low side drive is 9V9 referenced to VNN. Oscillation frequency is 48kHz. Rails are +60Vdc and -59Vdc.
If I force the amp to start drawing high current, it behaves pretty much like the 2R limit readings, however all the drives and rails show a lot of ringing, and obviously it starts drawing high current. If I switch it back to the current limiter, it goes back to behaving somewhat normal, meaning it goes to the first case readings. If I try to get it to keep running on an unlimited source, I have to restart the amp by cycling the remote.
I do have 640s in it as that is what my schematic shows. However, it initially had 640n's, and it was behaving the same way.
I'm wondering if something is wrong with the lm7812/15. I've measured the voltage if those supplies and they are roughly half what they should be when on a current limit or during the fault. The rail voltage is staying plenty high enough that it seems they should be rock solid.
I have the parts in stock, I may try that's in a moment. If I had hair, is be pulling it out.
Thanks,
Jason
With a 2R current limiter the high side drive is 3V3 referenced to VPP. Low side drive is 6V5 referenced to VNN. Oscillation frequency is 69kHz. Rails are ±29Vdc and symmetrical.
With no current limit the high side drive is 10V5 referenced to VPP. Low side drive is 9V9 referenced to VNN. Oscillation frequency is 48kHz. Rails are +60Vdc and -59Vdc.
If I force the amp to start drawing high current, it behaves pretty much like the 2R limit readings, however all the drives and rails show a lot of ringing, and obviously it starts drawing high current. If I switch it back to the current limiter, it goes back to behaving somewhat normal, meaning it goes to the first case readings. If I try to get it to keep running on an unlimited source, I have to restart the amp by cycling the remote.
I do have 640s in it as that is what my schematic shows. However, it initially had 640n's, and it was behaving the same way.
I'm wondering if something is wrong with the lm7812/15. I've measured the voltage if those supplies and they are roughly half what they should be when on a current limit or during the fault. The rail voltage is staying plenty high enough that it seems they should be rock solid.
I have the parts in stock, I may try that's in a moment. If I had hair, is be pulling it out.
Thanks,
Jason
Well, it's not the regulators.
I hate feeling like I'm grasping at straws. I'm 99% certain the issue is on the class d driver but I've tested every component I can and they test fine.
I hate feeling like I'm grasping at straws. I'm 99% certain the issue is on the class d driver but I've tested every component I can and they test fine.
48k oscillation is lower that I'd expect.
Are you using your scope to get the drive amplitude voltages?
Is there one group of FETs that heats up more quickly than others when it's drawing high current?
Are you using your scope to get the drive amplitude voltages?
Is there one group of FETs that heats up more quickly than others when it's drawing high current?
Yes, I used the math function and two channels to get the high side. I also verified with a inexpensive battery powered scope.
The low side gets are getting far hotter than the high side, even when current limited.
The guide says that this driver oscillates at 47kHz. I too thought it was low, but assumed normal after reading the section in the guide.
Thank you,
Jason
The low side gets are getting far hotter than the high side, even when current limited.
The guide says that this driver oscillates at 47kHz. I too thought it was low, but assumed normal after reading the section in the guide.
Thank you,
Jason
I've since changed that to 75k. The first reading didn't hold on later amps. 75k was more common.
Does the SMD cap that's connected to pin 6 of the top op-amp connect to ground on its other terminal?
If so, confirm that it's 220pf.
Do you see a lot of ripple on the primary filter caps?
Does the SMD cap that's connected to pin 6 of the top op-amp connect to ground on its other terminal?
If so, confirm that it's 220pf.
Do you see a lot of ripple on the primary filter caps?
Well, it was 220pf, but since I kept seeing the oscillation shift I pulled and tested it. It was 220pf, but had a fairly low dc resistance so I assumed the ceramic substrate had some defects. I replaced it with 330pf since I was following along with old troubleshooting steps.
Anyhow, long story short, I just put a new 220pf in and it's oscillating at 79kHz. It's also behaving much better. I can allow input voltage to sag much longer and it recovers without the high current draw. If I let it go maybe a second simulating an input voltage sag, it still does return to the high current situation, but I really doubt that is realistic in normal operation. I really think it should have some undervoltage protection. The power supply card does have an 8V5 zener in the remote (not shown in the schematic, but the card has a jumper and appears to allow you to select the zener or no threshold for the remote). I may change the zener to a higher value for rudimentary undervoltage protection.
When you ask about the primary filter caps are you referring to the power supply input, or the output filter? If the power supply input, I'm powering it with a 50a linear supply so there should be very little ripple but I haven't actually checked. I'm causing the "ripple" which causes voltage to sag by switching a series 2R power resistor in and out of the B+ supply. If we are talking about the output filter well they have a rail to rail square wave on them.
I'm probably going to put just a pair of the old 640N's back in for more testing. I imagine that the better switching characteristics of the N suffix parts will lead to even better behavior.
Thanks!
Jason
Anyhow, long story short, I just put a new 220pf in and it's oscillating at 79kHz. It's also behaving much better. I can allow input voltage to sag much longer and it recovers without the high current draw. If I let it go maybe a second simulating an input voltage sag, it still does return to the high current situation, but I really doubt that is realistic in normal operation. I really think it should have some undervoltage protection. The power supply card does have an 8V5 zener in the remote (not shown in the schematic, but the card has a jumper and appears to allow you to select the zener or no threshold for the remote). I may change the zener to a higher value for rudimentary undervoltage protection.
When you ask about the primary filter caps are you referring to the power supply input, or the output filter? If the power supply input, I'm powering it with a 50a linear supply so there should be very little ripple but I haven't actually checked. I'm causing the "ripple" which causes voltage to sag by switching a series 2R power resistor in and out of the B+ supply. If we are talking about the output filter well they have a rail to rail square wave on them.
I'm probably going to put just a pair of the old 640N's back in for more testing. I imagine that the better switching characteristics of the N suffix parts will lead to even better behavior.
Thanks!
Jason
I've seen some of the primary filter caps (across the B+ and ground terminals) open and cause excessive noise everywhere on the amp which caused some strange problems. It's easy to see. Watching the B+ terminal with the scope will show a LOT of high frequency noise.
So I've done a bit more testing. I removed the 640s and put a pair of 640n's back in. It did not make a difference.
Then I removed the primary filter caps (CapXon garbage) and replaced them with Nichicon UPW series 5600uF caps. I had tested those caps in circuit, and the bulk capacitance if all of them on parallel was close to what it should be. I tested them out of circuit and they are all still in spec. I'm using a real ESR meter not a generic "transistor tester". ESR was below the limits of the meter and dispassion was also very low. As far as I can tell, they are in spec.
I thought going to 5600uF might be enough of a bandaid to make the issue unlikely to occur in any operating conditions. However, with them replaced I can no longer get it to reproduce the issue no matter how long I make the supply sag. Clearly something is up with these caps, or maybe they were on the edge of being to little in the first place.
I'm going to get it ready to go back into the sink on Monday. With any luck burn-in will go well and I'll be able to report success.
Thank you a million times over!
Jason
Then I removed the primary filter caps (CapXon garbage) and replaced them with Nichicon UPW series 5600uF caps. I had tested those caps in circuit, and the bulk capacitance if all of them on parallel was close to what it should be. I tested them out of circuit and they are all still in spec. I'm using a real ESR meter not a generic "transistor tester". ESR was below the limits of the meter and dispassion was also very low. As far as I can tell, they are in spec.
I thought going to 5600uF might be enough of a bandaid to make the issue unlikely to occur in any operating conditions. However, with them replaced I can no longer get it to reproduce the issue no matter how long I make the supply sag. Clearly something is up with these caps, or maybe they were on the edge of being to little in the first place.
I'm going to get it ready to go back into the sink on Monday. With any luck burn-in will go well and I'll be able to report success.
Thank you a million times over!
Jason
So I just wanted to update you all about this elite.5. As it turns out it still wasn't fixed. I had several through hole parts bodged in on the driver board for testing, and once I had it working I ordered the correct smd parts that I didn't have in stock. When I replaced them it was right back to the same old ****.
As I was troubleshooting more I remembered that when I reverse engineered what I call the "generic Asian clone driver" (schematics posted in this forum section) and I modified it to use more robust driver transistors I had a similar issue to this with the modified boards. That problem was slightly different, but not much. The modified drivers would run fine in the amp if they were current limited but it would draw enormous current on an unlimited supply.
The solution to that was to adjust the driver current source. The driver was very sensitive to the actual driver transistors on the board, not so much the final outputs.
Since the class d design in this amp was very similar, I decided to wire in a potentiometer in place of the current source emitter resistor. With that in place I adjusted the current source. If I tweaked it for a hair more current the amp would behave exactly like my modified GAC amps. It would run current limited but not unlimited. When I reduced the current source output slightly the amp started running perfectly stable. In the end I adjusted until I found where it was reliable but with the highest possible current from the current source. The final value was only 300R higher than the stock 1K2.
I'm pretty sure this was a zero day fault. It could be component drift, but the behavior was there before and after nearly all new components in the driver. I've read numerous reports of early elite amps running ridiculously hot and a high attrition and return rate.
After adjusting the current source, the amp passed all bench testing and I ran it for almost a week on burn-in and it never even got remotely warm. The amp finally went back to the client yesterday. I thought you guys might want to know in case you ever run into it.
Thank you all for your help in sorting it out.
Jason
As I was troubleshooting more I remembered that when I reverse engineered what I call the "generic Asian clone driver" (schematics posted in this forum section) and I modified it to use more robust driver transistors I had a similar issue to this with the modified boards. That problem was slightly different, but not much. The modified drivers would run fine in the amp if they were current limited but it would draw enormous current on an unlimited supply.
The solution to that was to adjust the driver current source. The driver was very sensitive to the actual driver transistors on the board, not so much the final outputs.
Since the class d design in this amp was very similar, I decided to wire in a potentiometer in place of the current source emitter resistor. With that in place I adjusted the current source. If I tweaked it for a hair more current the amp would behave exactly like my modified GAC amps. It would run current limited but not unlimited. When I reduced the current source output slightly the amp started running perfectly stable. In the end I adjusted until I found where it was reliable but with the highest possible current from the current source. The final value was only 300R higher than the stock 1K2.
I'm pretty sure this was a zero day fault. It could be component drift, but the behavior was there before and after nearly all new components in the driver. I've read numerous reports of early elite amps running ridiculously hot and a high attrition and return rate.
After adjusting the current source, the amp passed all bench testing and I ran it for almost a week on burn-in and it never even got remotely warm. The amp finally went back to the client yesterday. I thought you guys might want to know in case you ever run into it.
Thank you all for your help in sorting it out.
Jason