Premier PRS-X720 amplifier

[zz28zz]

Thx for the education.
Here's what I came up with:
https://www.digikey.com/product-detail/en/t-global-technology/H48-6-150-150-0.5-1A/1168-1213-ND/3042042

Thermal conductivity is 3.2W/m-K . I have no idea if that's good or not but it's on the high end of the reasonably priced silicone sheets.

It's a 0.5mm thickness sheet. From what I've found online, 90V can only jump ~.0005" in free air. 
0.5mm = .019" so I believe 0.5mm is plenty thick enough for electrical isolation.

Just tried to order but Digikey reports they are temporarily not taking orders??

[zz28zz]

Digikey is up again. Parts on the way.

[zz28zz]

Got the power FETs and silicon sheets installed. Couldn't find a 1 amp fuse to install but did have a 3 amp. It blew when power was connected just charging up the primary side caps. Installed one of the three 30A fuses and powered-up. No issues at all. Installed the other 2 fuses. Amp sounds better than ever. Adjusted idle current on both channels and played it fairly hard for 30 mins. Probed around with my temp probe. Nothing got over 100 degs F. Couldn't be happier. Thx for all the help.

One final question: I still need to apply the glue that I originally found gluing the large caps together and to the board.
I have some marine sealant (3M 4200) that grips pretty well and is non-conductive. Can I use that or is the glue something special?
Amp is going back in the boat so it needs all the help I can give it in regards to shock protection.

[David_AVD]

Make sure that the silicone is neutral cure (non acetic).  Super common over here, some have heard that it's not as easy to find in the US?

[zz28zz]

Thx for the heads-up on the non-acetic silicone. After digging around on the net, I've found the safest/easiest/quickest product to acquire is GE Silicon II. Heading up to the hardware store today anyway so I'll pick some up there.

I would hate to mess things up at this point of the repair.

[zz28zz]

Well, things went well for ~3monnths. Last Thurs the amp blew out most of it's magic smoke. Got it on the bench and found all 6 fets in the power section were toast. Pulled the board out and started poking around. Just like last time, I find the +45V and -45V rails are shorted together.
Last time it was from the back of the audio fets shorted to the heat sink effectively connecting the two rails. Now, with the heat-sinks not attached to the fets, I still measure 0.4 ohms between the + and - rails, but neither rails is connected to ground. Looked at schematic and the only place I can see where the two rails could possibly connect (w/o connecting to ground) is thru the rectifier diodes. I removed the diodes (that bench test OK) but still have a short across the rails. Visually inspected board looking for solder blobs & etc, but can't find anything wrong. Any ideas on how to isolate the short?

[zz28zz]

Pulled all 6 FETs from one channel and short disappeared. Tested the FETs I had pulled and sure enough, 1 shorted (Drain/Source) FET on the pos side and another on the neg side.  They look fine visually. No evidence of heat damage or arcing at all.  All the failed parts are IR brand and received from Digikey. Amp wasn't being pushed hard at all. Got another set of FETs coming but still scratching my head as to what could have caused this.

[zz28zz]

After visual insp found 1 resistor (R669) in the I-V prot circuit that had burned the board and was cracked but measured OK. It was a carbon 1/8W 2.2k. Replaced with M/F 1/4W of same resistance. Also found 2 resistors in parallel (R566 and R550) that had burned board. They were carbon 24K 1/8W. Didn't have any 24K's avail so installed one 22K and a 27K. Both 1/4W carbons. Measured with the 2 in parallel and got 12.04K ohms, so I think I'm OK there now.

Power section FETs replaced and primary side of power section is good. Re-installed rectifier diodes. Connected scope to +/- 45V rails and applied power to amp. +/- 45V rails both went to abt 20V then started dropping back towards zero within abt 1/2 a second. Removed all the remaining audio section FETs and the +- 45V rails are OK now. Installed 1 new FET for + side and 1 new FET for - side. Did the same for the other audio channel.
Powered up and both rails shorted together thru the audio FETs again. Bench test shows 1 + FET and 1 - FET shorted (source/drain).

With all audio FETs removed, applied 1K sine-wave signal to both channels and connected scope to audio FET gate pins (no FETs installed, just a test point pins). I got a signal at the gate pins but it looked more like a square wave. Not sure if that's right or not. Doesn't sound right. After a couple of mins, the signal at the gates disappeared. I brushed my arm against the -15V supply BJT and noticed it was hot. Measured -15 and it's at ~-1V now. Pulled the output leg (emitter) of the BJT (Q908) that supplies -15V and BJT output still measured ~-1V. Checked zener (D909) that controls base voltage/current and it was shorted (to ground).

Attached is a pic of the schematic with the over heated (and replaced) components circled.
Next 2 posts are expanded views of same areas of schematic..

Questions:
1. Any issues with first paragraph?
2. Paragraph 3, should sine wave injected at input of amp look like a square wave (like it was clipping) with input gain turned way down and input from sine wave gen at 0.45V? (amp should handle up to 6.5V on input.)
3. At first I thought one of the audio amps was shorting the -15V, then found the zener D909 shorted. Any thoughts as to why the zener shorted?
4. Any thoughts why R669 over-heated? (I checked Q655 and Q661 for shorts)
5. Any thoughts why R566 and R550 over heated?

Thx!

[zz28zz]

I-V prot circuit.

[zz28zz]

Right channel.

[zz28zz]

Back at it again this weekend. Grabbed another amp which is the baby brother of the broken amp. Circuits are almost identical, just fewer FETS and 45v rails are 37V rails on the little one.

Replaced 15V zener on the base leg of the BJT that makes up the -15V regulator. Did the same for the +15V BJT. Now have 14.7V and -14.7V so that's appears to be OK now. Noticed the 15V BJTs get hot after abt 1 min or so but they are not heat sunk with the board on the bench. I'm forced to power on, take measurements and power off to keep them from getting too hot. Amp seems happy with all the audio FETS removed and nothing is getting hot.

Installed one of the 3 audio output FETs at each location, R-channel+, R-channel-, L-channel + and L-channel-. Connected scope to monitor +45 and -45V rails and applied power. Rails went to abt 20V for abt 1 second, then started dropping again. 30A fuse blew. All 4 audio FETs shorted again. This is with the amps RCA audio inputs shorted with a 1k ohm resistor (used to set idle current) and gain set to almost nothing.

Measured gate signals to the audio output FETs. With no input signal, both the + and - FETs are close to the positive rail voltage. After abt 15 seconds, they both drop to close to - rail voltage. Both channels are doing the same thing except one channel takes a few more seconds before gate voltage dives towards other rail.  On the good smaller amp, the audio output FETs gate voltage is abt 3-4V and steady.

Defiantly appears the gate voltage to waay to high for the audio output FETs and that's turning them on hard causing them to short out.
Looking at the schematic, I see a couple of suspects for each channel, but can't find anything that would cause both channels to do this. I've measured the outputs of every op-amp and measured the voltages at base, collector and emitter of every BJT in the audio path. Compared to good amp and all matches until I get to the portion where +50V and -50V are introduced. At this point both channels are isolated (I believe) so both channels would have to have the same failure mode, which seems unlikely.

 I've also monitored the +/-50V rails, the +/- 45V rails and the +/-15V rails. They are all where they should be and stable when the audio FET gate voltages dive from ~+40 to ~-40V.

Attached are the scope traces when connected at either the right channel audio output FET gates or left channel audio output FET gates and power is applied. Note the time scale is larger for the Left channel since things happen a little slower on that channel.

Does anybody have an idea what's causing this? I'm stumped.

[zz28zz]

Left channel at power up.

[shakalnokturn]

I'll also be referring to that other post at times.

The output FET's are source followers, so having +/-40V on the gates is not a problem as long as the source is able to follow, this is what happens all the time when you play music at full power.
What you do want to worry about is the differential voltage between gates as too much differential means both sides are conducting too much, current, heat... destruction.
Where you do have a problem though is if your gate voltages settle at the shown voltages when idle, this means a lot of DC on the outputs and you shouldn't connect a load until you have solved this.

You have double checked FET drains are insulated from the heatsink?

I'm not sure adding a 1k resistor to the sources is the best way to troubleshoot. I'd go for an adequate lab supply if you have one that can output +/-45V or inserting a 12V 20W  bulb on each rail before the output transistor drains. (Edit: this may be a little optimistic, maybe higher voltage & power bulbs is safer.)
Turn the idle current trimpots right down towards R505, R506, if layout is as on schematic.

I'd be checking on: Q555, Q557, Q559, Q573, D551, (open D501), R501, R563, R567, R569, trimpot (never ever neglect the trimpot in the bias circuit...), gate resistors.

[zz28zz]

I'll also be referring to that other post at times.

The output FET's are source followers, so having +/-40V on the gates is not a problem as long as the source is able to follow, this is what happens all the time when you play music at full power.
Understood


What you do want to worry about is the differential voltage between gates as too much differential means both sides are conducting too much, current, heat... destruction.
Just looking at one channel, the pos and neg FETs gates seem to follow each other. I.E. when all gates go positive, only the pos output FETs turn on, and vice versa for the neg FETs.


Where you do have a problem though is if your gate voltages settle at the shown voltages when idle, this means a lot of DC on the outputs and you shouldn't connect a load until you have solved this.

You have double checked FET drains are insulated from the heatsink?
Been down that road already. No heat sinks are connected with board out of chassis.

I'm not sure adding a 1k resistor to the sources is the best way to troubleshoot.
I'm sure you're right, just trying to get some feedback going without blowing up more FETs.

I'd go for an adequate lab supply if you have one that can output +/-45V or inserting a 12V 20W  bulb on each rail before the output transistor drains. (Edit: this may be a little optimistic, maybe higher voltage & power bulbs is safer.)
No lab supply here. Just a 12V jump starter battery pack for now. Not sure where I could find bulbs >12V and > 20W.


Turn the idle current trimpots right down towards R505, R506, if layout is as on schematic.
I can do that. I set them the last time I repaired this amp so I was thinking they should be close, maybe not

I'd be checking on: Q555, Q557, Q559, Q573, D551, (open D501), R501, R563, R567, R569, trimpot (never ever neglect the trimpot in the bias circuit...), gate resistors.
I've checked all those BJTs in circuit with a diode checker. Just picked up a TR/LCR/ESR tester. I'm planning on removing each BJT you mentioned and bench test. Tested all the resistors you mentioned in circuit. Measured ~ 5V drop across D501 when voltage was staying positive, so I assumed it was OK.
D551 had abt 0.5V drop across it, so I assumed it was OK.

Last week while troubleshooting a decided to try bypassing the 1K ohm resistors I had placed on FET source legs and turn amp on just long enough for scope to trigger and see if I get any signal at base of right side transistor of Q552. (I believe this is the what's supposed to keep the output FET gate signals from getting too powerful). Got away with it twice, but the third time I fried the FETs again.  So gotta order more parts,, again.

When I look at base of Q574, I see the 1khZ signal I injected going 95V PTP even with input signal barely on (~70mV PTP) and gain turned waay down. When I check the other amp I have, voltage stays much lower even with a higher voltage input signal (200mV PTP).

With both channels doing the same thing, it makes me think there may be something wrong with V/I protection circuits. I did have a burnt resistor at R669 that I replaced already.

There's also another circuit made up of Q203, Q204 and Q 205. Appears it's capable of sending the input signal to ground on both channels but not sure why you would want to do that. Emergency idle-down signal from IC651??

[shakalnokturn]

Q203-4-5, muting.

[zz28zz]

Makes sense, but there's no muting button. Maybe it automatically activates at start-up to save speakers??