Trying to fix home subwoofer amplifier. Any help is appreciated.

[jener8tionx]

This is a plate amp from a JBL PS1400 400W subwoofer amplifier. I have tried a few things, as outlined below, but my limited knowledge has been exhausted. Any help is appreciated.

Amp version: Rev - 03D 11/06/2000
All references to components are based on the PS1400 service manual and circuit board.
The service manual (and schematics) can be found here:
http://www.jbl.com/resources/Brands...cuments/en-US/ServiceManual/Ps1400 Sub SM.pdf

1. The first sign of trouble was when I noticed there was no output, the leds were red, and sometimes the leds would turn green for a second and would power the speaker. I checked the driver and installed the working amp from the other PS1400 in where the faulty one was and it worked fine, so I am pretty sure the problem is just within the amp itself.

Action taken:
a. I bench powered the amp and found nothing that would indicate that the amp shouldn't be working. Because transistors are often the culprit, I removed the output transistors. I then bench powered the amp and the relay sounded like it was functioning so I thought I had it figured out.
b. I ordered new transistors and changed the output transistors Q11, Q13, Q15, Q17
c. I then bench powered the amp and it appeared to be working
d. I also noticed visible dark spots on PCB board where R76 and R78 mount. It was the same on the board of my functioning PS1400. See pictures below of functioning board:



2. After reinstalling the plate amp it ran for a few minutes then back to red LEDs. I could hear the relay clicking.

Action taken:
a. I saw that the fuse had blown. I replaced the fuse from the working amp.

3. I powered the amplifier on the bench. PCB board shorted., D3 was melted, Q8 was visibly broken

a. Replaced D3, D?, R76, R78, Q5, Q8
b. Soldered over fried area of PCB board



3. Amplifier LEDs powered straight to green, but no output, burning smell

a. The PCB was fried again in same place; fixed, replaced burnt out U4
b. removed and tested bridge rectifier, diodes tested good.
c. Measured transformer resistance through DMM, both good and questionable had same measurements both VAC powered and Ohms while off.

4. Hooked up board to known working amplifier’s transformer. Blown fuse, no other visible damage.

5. I put in the new capacitors and this time Q17 blew up and then before I could even flip the switch, the fuse blew. See picture below with the offending transistor circled and you can see the new caps as well.

I took all of the power transistors out and fired it back up (with yet another fuse -- I'm getting low!). No pops or blown fuses. I was able to get my DMM in there to measure some voltages.



V+ 50V
V- -49.9V
+15v cycling between ~13v and ~16v
-15v -10v

No clue what the problem is. Maybe to voltages are off because I have the power transistors out and they aren't drawing power, but I suspect that some component is either open or shorted and/or a voltage regulator is bad. I guess I will take a break for a bit and then come back and do a ton of voltage measurements and try to figure out what the problem is.

Other things I may try:

1. Replace the output transistors again
2. There were a couple differences between my board and service manual schematic. Ex: R 76 and R 78 were not 820 Ohms, they were very close though. I wonder if there are any other changes?

Pic of the top of the board:


Note: you can see the replaced resistors R76 and R78. They are both 820 Ohms and 4 watts.

Pic of the top of the board:




Any other ideas?

 

[duke37]

I cannot help with solving the problem but I would use a lamp limiter in the mains input.This would enable you to run without blowing a fuse and give voltages to measure.

 

[KrisBlueNZ]

That's an extremely thorough problem description, and thanks for the link to the service manual! You are very methodical and thorough, and this is a very good thing.

I've had a look at the schematic and it seems pretty straightforward, so I should be able to help you diagnose it. It's a bridge-tied class D output stage with voltage, current and temperature fault detection, controlled by two pulse-width modulators made from op-amps. I'll write up some suggestions in a day or two once I've fully analysed your post.

There's clearly something wrong with one or both of the 15V rails. My guess would be a faulty op-amp but it could be lots of things. It should be possible to diagnose the problem using a methodical approach, and you're the guy to do it!

The first thing you need is to get the thing running without blowing anything up! Duke's suggestion is a good one. Another way would be to remove the four output MOSFETs (it's enough to just disconnect the centre lead, which is the drain). That should prevent any really bad explosions.

I want to mention now that I don't like the look of the area of the board where the track has lost its solder mask and has been repaired. I would clean that area thoroughly with isopropyl alcohol and a cotton swab. Personally I would first remove all the solder mask from a longer area of the track, and solder a piece of thick wire (at least 0.5 mm diameter) along it, tinning the whole track as well, before doing a thorough cleanup. That would look tidier in my opinion.

After you clean up the area, can you take a good close-up photo of it please.

I am also a bit concerned about the charring of the board, due to heating of the 820 ohm resistors. Even fibreglass will start to lose its insulating properties if it's charred enough.

You can estimate the load on the 15V rails by measuring the voltage drop across the two 820 ohm dropper resistors.

I'll post in a day or two with some more specific suggestions.

What test equipment do you have? Multimeter obviously; oscilloscope? Anything else?

 

[jener8tionx]

Kris, Thank you for such a detailed response. I will try to do what you suggested.

1. I agree that something is wrong with one or both 15v rails. The V+ (50v) and V- (-50v) appear to be correct and working just fine

2. Right now I have the board running with the output transistors removed. I can also look into a lamp limiter, but the board is running for now.

3. I will try to clean up the area where the track popped. I have big fumble fingers so it is hard to get a pretty result. I can replace the soldered repair with some wire. I have also tested the repairs and there aren't any shorts that I can find in the repair. It isn't pretty, but I don't think it's the current (get it? current?) problem. I will take a photo once I get it cleaned up.

4. The heated area around the 820 ohm resistors bothers me as well. The working amp has the same burned marks as shown in one of my photos. I'm not sure if the circuit created the extra heat or if there was a late stage repair made to the board and a tech over heated the area when it was repaired. That's just an idea that popped into my head.

5. The voltage drop across the 820 ohm resistors is:

Working amp:

R76: -50V to -22.9V
R78: 50.5V to 14.14V in standby and 17.8V when active

Broken amp:

R76: -50V to -7.5V
R78: 50V to cycling between 12V through 16V and back down

 

[sheldonstv]

first off where you have bridged that damaged track is no good-remove the broken track completely and solder a short length of tinned single core wire across it-isolate the supply from the rest of the amp and get that working first-looking at the schematic its an easy supply to repair-with all voltages as they should be ill talk you thro the rest

 

[KrisBlueNZ]

Personally I would leave the track and just add a wire on top of it, running down the middle. And LOL (slightly

Yes I noticed you said that the working amp has the same overheating issue. I would fix that as well. Perhaps move the resistors off-board - mount them against the heatsink, or the front panel, or on individual heatsinks, or at least on standoffs. You may find that thick film resistors are easier to mount onto heatsinks.

I doubt very much that the discolouration would be caused by soldering. It takes sustained high temperature over a period of time to cause it. If there's any sign that the area is becoming conductive, I would remove that section of fibreglass completely, using a drill and some files, and remake any affected circuitry - perhaps on pieces of stripboard or perfboard. That's quite a job though.

It's interesting that the +15V regulator voltages are not steady. Are they cycling in time with any other action? For example, the fault LEDs or the relay?

I'll let sheldonstv be your advisor, since he has offered, and add comments when I think they're necessary. My first comment is that the +/- 15V rail problem is probably due to overloading so it won't show up with the rest of the circuit disconnected.

 

[jener8tionx]

I replaced the solder bridge with a piece of gold plated jumper. I think this will be a better fix. It really looks like a mess in the macro picture, but it's the best I could do. I will get some alcohol to clean up the board a bit.

 

[jener8tionx]

How does the board supply the 15V rails? Is it the LM2574 Step-down voltage regulator (U6)? Could that be something worth replacing? Any other capacitors that I should have a look at maybe?

 

[KrisBlueNZ]

The +15V and -15V rails are supplied by Q6 and Q8, which are supplied from the main +V and -V rails via R78 and R76. There is another rail, "+15VSW", from the LM2574, which is actually +15V relative to the -V rail. It supplies U7 and U8 and associated buffer transistors, but I don't think it supplies anything else. There's no reason to suspect that supply yet.

sheldonstv are you still planning to "talk him through the rest"?

 

[jener8tionx]

Could it be that I replaced Q6 or Q8 with the wrong part?

On the parts sheet Q6 is listed as "NPN Transistor, MPSW06RLRA" and I replaced it with mouser part number "863-MPSW06RLRAG"

On the parts sheet Q8 is listed as "NPN Transistor MPSW56RLRA MPQ" and I replaced it with Mouser part number "863-MPSW56RLRAG"

I think these were the correct part numbers for replacement. I bout 2 of each, so I could swap out the ones that are on there if you guys thought that might work.

On a side note, I was staring at the board and noticed that only one of the anode pins on D3 was properly soldered and my DMM was reading about .7v through the diode. I soldered the other anode and now it reads 1.8v through the diode. I haven't fired it up after doing this.

Again, thank you for all the help!

 

[KrisBlueNZ]

Yes, you've used the right transistors. Q8 is a PNP. The parts list says it's an NPN; that's an error.

D3 only has one anode pin. Looking at the photo in post #7 on this thread, the anode is the pin at the bottom left of the device. The pin at bottom right is not used. If that's the pin that wasn't properly soldered (and it doesn't look like it's properly soldered in that photo), then soldering it won't make any difference.

As for the voltage measurements, I assume you're measuring across the diode with your multimeter set to the diode test range? If so, you should measure about 0.7V across the diode, with the positive lead to the anode (bottom left) and the negative lead to the cathode (top middle). If you're measuring 1.8V across the diode with the probes that way round, the diode is probably damaged. But I don't think that's the case.

Edit: You didn't answer my earlier question. What test equipment do you have, apart from the multimeter? Do you have an oscilloscope?

 

[jener8tionx]

Yes, you've used the right transistors. Q8 is a PNP. The parts list says it's an NPN; that's an error.

D3 only has one anode pin. Looking at the photo in post #7 on this thread, the anode is the pin at the bottom left of the device. The pin at bottom right is not used. If that's the pin that wasn't properly soldered (and it doesn't look like it's properly soldered in that photo), then soldering it won't make any difference.

As for the voltage measurements, I assume you're measuring across the diode with your multimeter set to the diode test range? If so, you should measure about 0.7V across the diode, with the positive lead to the anode (bottom left) and the negative lead to the cathode (top middle). If you're measuring 1.8V across the diode with the probes that way round, the diode is probably damaged. But I don't think that's the case.

Edit: You didn't answer my earlier question. What test equipment do you have, apart from the multimeter? Do you have an oscilloscope?

You were right as far as the difference it would make. No change by re-soldering. My test equipment is limited to a DMM. I don't know anyone around here with an oscilloscope.

 

[KrisBlueNZ]

What's happening with this project?
It looks like sheldonstv has lost interest.
Have you cleaned up the repairs? Can you upload a new photo?
Do you still want to get this thing going?

 

[jener8tionx]

I'm still trying. I just lack the electronics background to get any farther. Not much is different from the last pic, but I will see what I can do. Isopropyl alcohol did very little to remove the flux.

My theory: either the power section has a bad component, or something is shorted and drawing too much current from the 15v rails. Thoughts?

Also, I'm trying not to bump this thing everyday because I don't want to be a bugger! Thanks for all the help so far Kris.

 

[quantumtangles]

You are excellent as always Kris. Interesting thread. Amusing as well.

How would you use an oscilloscope to diagnose this?

 

[KrisBlueNZ]

I don't think an oscilloscope is needed - yet. But this is a Class D amplifier so it will be very useful later to check the PWM drive signals. This is significant because all the PWM stuff is done with op-amps, not with a dedicated control chip that could just be replaced on suspicion.

Isopropyl alcohol should be able to clean up that board nicely. You need to use a cotton swab (stick with cotton wrapped around each end) and a fair amount of pressure and back-and-forth rubbing, but it should work. I've never had a situation where it didn't.

Yes I agree, probably something is drawing too much current from the 15V rails. I'm not home at the moment; I'll be back on Friday. I need to print out the diagrams and have a good look at the design, and go over the posts so far on this thread. Then I'll suggest some things to try.

In the meantime, see if you can get access to an oscilloscope. It will be very valuable later in the repair process, for checking various signals in the PWM circuitry.

 

[KrisBlueNZ]

Please respond to each of these points.

1. Can you confirm that on both the good amp and the faulty amp, the discolouration of the PCB around R78 is much worse than the discolouration around R76? This is interesting because it indicates that the positive rail of the low-voltage circuitry draws significantly more current than the negative rail does.

2. FYI at the moment you need to leave the output MOSFETs disconnected, to prevent fuse-blowing. Only once the drive waveforms are checked should you connect them back as originally designed.

3. In your first post, you say "3. I powered the amplifier on the bench. PCB board shorted., D3 was melted, Q8 was visibly broken". 3 (a) What do you mean by "PCB shorted"? Do you mean that the board accidentally touched onto something conductive, and that's what caused all the damage? 3 (b) If so, do you have any idea what shorted to what?

4. (a) Can you confirm that until the PCB track was blown up, you hadn't measured the 15V rails and you had no suspicion that there might be a problem with that part of the board, apart from the discoloration problem? 4 (b) is it a complete coincidence that the track that was blown up is in that part of the circuit?

5. I am quite concerned about the condition of the PCB around the R78, and to a lesser extent, R76. At the moment I'm giving them the benefit of the doubt but at some stage I may ask you to remove R78, R79, D2 and Q5 and check for resistance between tracks in that area, to see whether the insulating property of the fibreglass PCB has been compromised. If this is the case, you will need to use a drill and files to remove all the discoloured fibreglass, and remake the required parts of the circuit.

6. You said that R76 and R78 are not 820 ohms but were very close. Do you mean you measured them and they weren't exactly that resistance? That's to be expected; resistors, like all components, have a tolerance. Were the originals marked as 820 ohms (either with printed markings or with colour bands), or were they marked with some other value?

7. I was serious when I suggested moving R76 and R78 off the board. The discolouration on both boards is quite bad; if it isn't causing problems now, it will in the future.

8. I think it's time to check the loads on the +15V and -15V rails. A good way to do this is to insert a one ohm shunt resistor in series with the emitter of Q6, and another in series with the emitter of Q8. That will allow you to measure the load on those rails at any time, just by measuring the voltage across those resistors, without disturbing the circuit and without affecting the voltages on those rails much.

For Q6, then for Q8, you need to disconnect the emitter wire from the pad it was soldered to, and connect a one ohm resistor (5%, 1/4 watt is fine) from the emitter wire to the pad. Ideally these resistors should be easily accessible so you can get your meter probes on them, and protected so that the meter probe will not slip off and short onto something nearby! You may want to run wires from the resistors to a multi-terminal connector that is glued or taped to some stable component (e.g. the big electrolytics) to make it easy and safe to connect to them.

Then measure the voltages across the resistors. It should be a low voltage, less than 40 mV (0.04 volts). If it's more, there is a fault in the circuitry that is overloading the rail.

9. As I said earlier, I'll look into the design and suggest a procedure. You may want to buy some more TL072 dual op-amps. If U4 was damaged when the short circuit occurred, it's likely that other op-amp ICs were damaged then too.

 

[jener8tionx]

Kris: Here are a few more pictures

1. Here is a picture after some cleaning:



2. Here is a problem I noticed while cleaning. C135 appears to be fried. Also the lead next to it on the left was not soldered correctly, It was loose. I resoldered it, but it needs to be resoldered again to clean it up.



This it the top view of the capacitor that wasn't soldered correctly. It looks like it is bulging a bit. The green one on the bottom. It is C113 for reference.

 

[jener8tionx]

1. Can you confirm that on both the good amp and the faulty amp, the discolouration of the PCB around R78 is much worse than the discolouration around R76? This is interesting because it indicates that the positive rail of the low-voltage circuitry draws significantly more current than the negative rail does.

Yes, the discoloration on R78 is much worse on both boards

2. FYI at the moment you need to leave the output MOSFETs disconnected, to prevent fuse-blowing. Only once the drive waveforms are checked should you connect them back as originally designed.

OK

3. In your first post, you say "3. I powered the amplifier on the bench. PCB board shorted., D3 was melted, Q8 was visibly broken". 3 (a) What do you mean by "PCB shorted"? Do you mean that the board accidentally touched onto something conductive, and that's what caused all the damage? 3 (b) If so, do you have any idea what shorted to what?

The PCB short is where the pcb track shorted apart. I could see a flash of light through the board (the track was on the underside) as is shorted.

4. (a) Can you confirm that until the PCB track was blown up, you hadn't measured the 15V rails and you had no suspicion that there might be a problem with that part of the board, apart from the discoloration problem? 4 (b) is it a complete coincidence that the track that was blown up is in that part of the circuit?

Before the track blew up I did measure the 15v rails, but I didn't write down the measurements. I remember that the +15v rail was low, maybe +10 to 13v, but it was steady. I think the -15v was OK, but I can't remember exactly.

5. I am quite concerned about the condition of the PCB around the R78, and to a lesser extent, R76. At the moment I'm giving them the benefit of the doubt but at some stage I may ask you to remove R78, R79, D2 and Q5 and check for resistance between tracks in that area, to see whether the insulating property of the fibreglass PCB has been compromised. If this is the case, you will need to use a drill and files to remove all the discoloured fibreglass, and remake the required parts of the circuit.

I really don't want to do this I did measure resistance with the parts removed and I got infinity measuring in ohms.

6. You said that R76 and R78 are not 820 ohms but were very close. Do you mean you measured them and they weren't exactly that resistance? That's to be expected; resistors, like all components, have a tolerance. Were the originals marked as 820 ohms (either with printed markings or with colour bands), or were they marked with some other value?

They were marked as some other value. I couldn't tell what the colors were, but here's a picture of the good board and the resistors on it.


7. I was serious when I suggested moving R76 and R78 off the board. The discolouration on both boards is quite bad; if it isn't causing problems now, it will in the future.

Do you think increasing the wattage of these resistors helps? Would longer leads help?

8. I think it's time to check the loads on the +15V and -15V rails. A good way to do this is to insert a one ohm shunt resistor in series with the emitter of Q6, and another in series with the emitter of Q8. That will allow you to measure the load on those rails at any time, just by measuring the voltage across those resistors, without disturbing the circuit and without affecting the voltages on those rails much.

For Q6, then for Q8, you need to disconnect the emitter wire from the pad it was soldered to, and connect a one ohm resistor (5%, 1/4 watt is fine) from the emitter wire to the pad. Ideally these resistors should be easily accessible so you can get your meter probes on them, and protected so that the meter probe will not slip off and short onto something nearby! You may want to run wires from the resistors to a multi-terminal connector that is glued or taped to some stable component (e.g. the big electrolytics) to make it easy and safe to connect to them.

Then measure the voltages across the resistors. It should be a low voltage, less than 40 mV (0.04 volts). If it's more, there is a fault in the circuitry that is overloading the rail.

We just got a heavy snow so it will be a few weeks until I can get to my shop where I have extra 1ohm resistors. I could order a couple, but unless I need other parts, shipping is expensive. On that note, I live in the middle of nowhere. People around here don't know what and oscilloscope even looks like so I don't know if I will be able to get access to one..

9. As I said earlier, I'll look into the design and suggest a procedure. You may want to buy some more TL072 dual op-amps. If U4 was damaged when the short circuit occurred, it's likely that other op-amp ICs were damaged then too.

Good idea. I have one extra TL072 on hand.

 

[KrisBlueNZ]

Thanks for the updated images. The reworked area is starting to look a bit better now.

Yes, C135 looks like it needs to be replaced. The joint nearby looks OK now, I think, but I think you should carefully inspect the whole board for dry joints. Run this Google images search to get an idea of what a dry joint looks like: http://www.google.com/search?site=imghp&tbm=isch&q=dry+joints

Yes, that electrolytic is definitely bulging. Is it C113 or C136? It's a non-polarised electrolytic capacitor, rated at 100V. Neither Digikey nor Mouser have any suitable replacement. Actually I don't think the designer should have used that type of capacitor. A polyester or polypropylene capacitor would be a lot more appropriate in that position. See http://www.digikey.com/product-detail/en/QXK2E685KTP/493-3526-ND/2117905. I would also replace the other one.

That is not the only dodgy component choice in that design. I've noticed a few poor design decisions and I think the designer may have been inexperienced.

The failure of that non-polarised electrolytic may be related to the failure of C135.

Thanks for the detailed answers to my questions.

Point 3. "The PCB shorted" is not a meaningful thing to say. I would describe it as a section of track vaporising.

So do we assume that there was some problem internal to the board that caused the track to blow up and damaged D3 and Q8? Is it possible that you made a mistake or something accidentally touched something else, to cause the problem?

Point 6. Can you measure the resistance of R76 and R78 on the undamaged board. They should measure properly in-circuit, i.e. you don't have to desolder them. Try the multimeter one way then the other and post the two values you get.

Do you have the original R76 and R78 from the faulty board? Do the colours look similar to the ones on the good board?

The colours will have changed due to long-term heating of the resistors. I think the ones in the picture _could_ have been 820 ohms... The closest one, which is the most discoloured, reading from the right end, could be grey, red, brown, gold (the brown stripe could have turned orange) and the back one, reading from the left end, could also be grey, red, brown, gold. Do you agree?

Pont 7. Yes, higher wattage and longer leads will both help. If you use the rectangular type, you could attach them both to a single small heatsink.

OK, well, it sounds as though the project is on hold for a while due to weather.