Premier PRS-X720 amplifier

[zz28zz]

This is a high-end 2-channel car stereo amplifier made by Pioneer. Bought a couple of them ~15 years ago used. Great amplifiers but now one of them has a scratchy sound coming from right channel that gets worse as volume is increased.
Opened it up and went straight for the electrolyte caps due to it's age. Of the 7 large caps, 3 are tall and thinner and 4 are short and fat. Advertised capacitance is 4700uF for all 7. The 4 short fat caps all had a dome shape on top but the other 3 were flat on top. Removed and tested all 7. The tall thinner ones measured 4976uF, 5050uF and 5310uF. The 4 short fat caps measured 1201uF, 3140uF, 3252uF and 3234uF. Assuming a 20% tolerance, all of the short fat caps are out of limits.

Questions:
1. Where can I download a schematic for this amp?
2. What's the purpose for the rtv looking substance gluing the caps together and to the board? Does it need to be re-applied? See enclosed pic.
3. Do the symptoms described sound like a bad cap? (I'll be replacing all 7 anyway)

I've repaired thru hole PCBs before but always had a schematic. Never tackled an audio amplifier before. Any advice is much appreciated.

Thx!

[coromonadalix]

serv manual 

you have to register, its free, just did

1)  https://www.hifiengine.com/manual_library/pioneer/prs-x720.shtml

choose the CRT2361   2.9 megs,  the other is incomplete.


2) in some case the glue is for mechanical purposes, you do know in a car there is lots of vibrations ...

3)  i would check if you have some dc voltage going out  on the speaker terminals at zero volume,  check the supply lines if they are equals, and the ripple if yo have a scope.


Before jumping to the caps,  did you powered it up and tested the sound ?

[mzacharias]

The big power supply caps are not going to cause your symptom on only that one channel. Perhaps they should be replaced, that is another matter though.

[Twoflower]

Not sure but the temp-rating on the bigger diameter caps looks like 85°C (you can only see the 5°C on another M85) which is for a car amplifier quiet low rated...

If you're replace them, go for 105°C types unless you're living in very cold regions. Of course for the others as well. They should probably low ESR types, which usually are 105°C types anyway.

[zz28zz]

Downloaded the manual. Thx for the link!
Should have known the glue was for mechanical reasons. I'm used to working on lab equip where there's no real vibrations.

I re-installed the caps I pulled out for testing. Powered up amp on the bench with signal generator and scope connected to speaker terminal. Made a makeshift dummy load from a car taillight bulb (~3 ohms static). I've been running freq sweeps from 500hz to 5khz and watching scope. Everything looks good as best I can tell.  Not sure what to try next short of re-assembling and re-installing it in the boat (PITA).

Due to the physical size/pin spacing, finding replacement caps hasn't been so easy. Cannot find anything but 85 deg rated caps in the size I  need.

[zz28zz]

I got some 8-ohm home speakers and a head unit (home stereo CD player)  connected while on the bench. Sure enough, right channel sounds distorted when playing music, even at a low volume level. Swapped speakers and input connections. Right channel of amp still sounds distorted.

Rail voltages on signal side and power side steady and equal (but opposite). +/-15V supply looks good.

I did run across something odd when scope was connected to signal output FET gates. With a 2-channel signal generator outputting a 1khZ sine wave to both channels (at the same drive level), the Pos FET gates on the right side have a P-T-P value of 2.36V while the left side is 1.32V. See attached. Swapped cables at signal gen but it didn't change anything.

When connected to signal output Neg FET gates, the right side doesn't have the same offset as the left side. See attached.

Blue trace is the bad (right) side in both pics.

Only change between screenshots was moving scope probes (didn't touch signal gen or any other adjustments).

Swapped cables at signal gen again but it still didn't change anything. Changing freq on the signal gen didn't have any effect in regards to the different P-T-P value issue or the offset issue.

Hoping someone can make some sense of this.

[coromonadalix]

normally for debugging an  amp, car amp,  shoot a 1khz signal at 0db, and follow its path thru the amp on each channel until you see a discrepancy between the chanels, and work your way.

Did you check for dc voltage at the speakers terminals ??

[zz28zz]

My cheepo sine wave generator doesn't display output in db. Just shows output voltage and offset which was 0.02V and zero for offset.. Maybe I'm missing something here??

Forgot to measure the voltage at speaker terminals. I'll check that tonight. If I do find voltage at speaker terminals with amp idling, is that a sign of leaky signal output FETs?

[coromonadalix]

not necessarely, you may have an imbalance on the output stage,  the output stage is a push pull type   

Ex:  you may have a transistor / mosfet who drive / leak a little to a supply line, in the end you end up with a dc voltage on the output speaker and you may fry a speaker if a protection doesnt kick, or if its not equiped with this circuit.

The speaker terminal is a cross over between the positive and negative supply, it should read zero, or pretty near zero  in the millivolts range.  Only the ac signal  IE:  audio is amplified

[zz28zz]

Understood, thx! I'll update later this evening when I get a chance.

[zz28zz]

Made voltage measurements on the speaker terminals with input cables and speakers disconnected. Good channel= -17mV. Bad channel= +50mV. Is that a significant amount?

[zz28zz]

Found something interesting this evening while playing with the scope and signal gen.

I connected one scope probe to the gate and one to the output (source) of the "bad channel" negative MOSFETs.
At 80hZ the output follows the gate signal nicely, however as freq is increased, the gate signal continues to look fine but the output signal starts looking less and less like a sine wave. The top half of the sine wave looks pretty good but the bottom half is not so good. At 200hz, it looks really bad. I saved the data starting at 100hZ and stepping up in 100hZ increments up to 1khZ.   See attachments. Blue traces are the gate signals. Disregard the measurement data at bottom of pics. Still trying to figure out how to make that read right.

Above 1 khZ the neg MOSFETs output quality continues to improve.

Negative rail voltage stayed within ~150mV of set-point (41.5V) the whole time.

Repeated same tests on good channel and the outputs followed the gate signals perfectly from 80hz to over 20khZ.

I then moved to the bad channel positive MOSFETs. They behaved ok at lower freqs but had a similar issue at high freqs.

Time for a new set of audio section MOSFETs??

[coromonadalix]

i would say the negative side of the signal is distorted,  that relate to the negative voltage side of the amplifier

[zz28zz]

So, bad MOSFETs on neg side? I don't see much else it could be except maybe the voltage and current protection circuits running amuck.

[zz28zz]

Found one of the bad channel neg-side FETs has a significantly lower mean gate voltage than the other two partner FETs while injecting a 200hZ sine wave.
Got all the audio FETs and 7 caps on the way. It will be interesting to see what the old FETs look like on the curve tracer after ~20 yrs of operation.

[shakalnokturn]

I'd actually be really surprised if one of the power FET's was the trouble.
Assuming that bad solders on the copper side and the +/-50V supplies (TP927/TP928) have also been checked, and given that symptoms seem frequency dependent things I'd be looking at first are: C501, C551, C555. (Or equivalents on the concerned channel if I have the wrong one...

[zz28zz]

Well, things went from promising to down right ugly.

While awaiting delivery of the new audio output FETs, I installed the 4 new large rectifier caps and the 3 new caps on the primary side of the transformer. Then I figured that since there are 3 audio FETs in parallel (for each polarity of each channel), I could test audio output at a low level with one removed. I removed the suspect audio output FET from the circuit and powered up amp. Sine wave signals looked good at all freqs on both channels. Connected CD player and music sounds good. I felt I had this one in the bag.

 New audio output FETs arrived yesterday and I installed them last night. I was fully prepared for the sweet sound of music but instead I got a small fire immediately after applying power.
2 of the power section FETs blew off their source legs off. I pulled all 6 of the power FETs off and cleaned up the soot. Tested all 6 of the FETs with hand-held meter and every one has a dead short between gate and drain. With the FETs removed and amp powered up, checked the gate signals. I see a nice clean square wave with 11.2V P-T-P amplitude and a freq of ~34khZ at all 6 positions.  See attachment. I'm stumped as to why this happened. ESD??

Checked Digikey and found the original power section FETs are obsolete. Found an option that looks promising. Here's links to both. Does this look like a good match?

https://www.digikey.com/products/en?keywords=P50NE

https://www.digikey.com/products/en/discrete-semiconductor-products/transistors-fets-mosfets-single/278?k=&pkeyword=&sv=0&pv606=1&pv570=73&umin2068=80&umax2068=300&rfu2068=V&pv608=1970&pv1993=4&pv1236=26&pv2262=u110nC+%40+10V&sf=1&FV=ffe00116&quantity=&ColumnSort=0&page=1&pageSize=25

[Twoflower]

Not sure, but could that be that this was caused by a short over the heat-sink? Have you forgot to insulate the FETs from the heat-sink? Usually they need to be electrically-insulated.

[zz28zz]

All backs of all the TO-220 FETs are sandwiched up against some unknown material abt 1/4" thick that's non-conductive. That material is heat sunk into the shell of the amp. Never touched the power side FETs. Also verified the heat sink compound isn't conductive.

[zz28zz]

Started checking audio side of amp and discovered the +45V and -45V rails were shorted together. Started loosening the clamps for the audio MOSFETs and the short went away. Cleaned up the "unknown" material where the FETs source heat to. It appears to be alum with a non-conductive coating. After cleaning it up, found a few small places where I nicked the coating and created a path for current (reminder to self: don't use a knife to clean the heatsinks). Since both polarities of FETs are on the same chunk of alum, it shorted out but not until the very last tightening of the clamps. Looks like I will need some sort of insulating film between the FETs and the heatsink. Good call Twoflower. Thanks.

[David_AVD]

Did the original FETs have mica (very thin clear) insulators under them?

[zz28zz]

No insulators. Just the coating on the heat-sink.

A couple of the "nicks" turned out to be gouges. You could run your fingernail over it and feel the metal sticking up just a tiny bit. They just happen to line up with the back-side of a neg FET and pos FET. Rotten luck there. I dressed up the damage with some crocus cloth wrapped around a tiny file so that I had a slight depression then used a toothpick dipped in enamel paint to fill the depressions. Next step is to re-assemble and test for shorts.

Still need to find replacement power FETs and get them on order. May need to start another thread for that.

[Twoflower]

You should go for insulation sheets. Rely on the already damaged Aluminium-oxide or coating isn't the thing I would go for. Also how even is the surface now with the applied paint and how good is the thermal performance of the used paint?

Look for a 0.5 or 1mm silicone sheets to mount the FETs. You can get some 10x10cm sheets instead of individual TO220 sized ones. Double check the datasheet for the pads if they're electrically insulating. If you use mica sheets don't forget the thermal grease.

But as you wrote you measured on the fried FETs a short to the gate. That might have created some other problems at the gate-driver.

[zz28zz]

I don't have a way to accurately measure the flatness of the heatsink at the painted spots but they look flat using the naked eye. I'd estimate the painted spot makes up less than 10% of the surface area of the FET.  Also have no idea of the thermal conductivity of the paint. I assumed the heat sink compound would help transfer the heat around the painted areas.

The silicon pads would be a good insurance policy. I was concerned about adding any additional layers between the FET and the alum. If I use insulating sheets, there would be two layers, the oxide layer and the silicon sheets. Maybe I'm over thinking this. I could grind off the oxide coating but it wouldn't be nearly as flat as it is now.

Ive seen where folks mention the mica sheets tend to crack over time so the silicon sounds better in that regard. Is the thermal grease the same as heat sink compound? Is heatsink compound needed when using silicon sheets or is it used dry?

The gate signals look OK with no FETs to drive. Hoping they remain so when FETs are installed.

[Twoflower]

Thermal grease or thermal compound is not needed for the silicon sheets. Only fuse the grease for mica or no insulation material. For the silicon pads, I would use "single side sticky" ones. that way you can remove it when needed.

If you don't get the surface as smooth as it is supposed to be you're make it probably worse. So no grinding.

I've nearly killed my scope because the TO-3 mica-sheet in my lab-power supply got defective over time. That caused a nice DC short-circuit over PE.