Denon PMA-50 relay to enable amplifier circuit

[mk2]

As a novice learning some electronics, I'm making progress but seem to have ended up at a logical dead-end and would appreciate any advice.

This is my trusty Denon PMA-50 which has failed (service manual is here)

Broadly, the buttons/displays are fine and it clocks onto a Bluetooth, optical or coax sources. Normally I expect an audible relay 'click' a few seconds after power-on but this does not happen and there is no sound (neither speakers or headphones)

The PSU has separate groups of outputs for the "main" PCB and amplifier boards and when bringing the unit out of standby, the "P/S-1" is raised which asks the PSU to enable +/-28V supply to the amplifier board. That seems to test ok with a multimeter.

The relay sound I'm hoping to hear seems to be RLY9000 aka. JZC-42F (datasheet) -- a pair of relays, driven by SPK_ON which seems to be something the MCU decides. Testing that with a multimeter and it's low. I nervously bridged 3V to it, to see if that would connect the relay and it did not. According to page 55, it's exclusively for the speakers so would not account for the lack of audio on the headphones. So I think lack of SPK_ON is a problem, but I think it's a symptom, not the root cause itself.

During inserting the headphones, a small amount of buzz is heard when things short. So I suggest the headphone amp is 'live' but has no audio.

The lack of SPK_ON is a clue, that the MCU is not happy, so I think I need to work out some way it makes that decision based on actual audio. Perhaps it's driving HP_MUTE as well. All of that would seem to require communication in the reverse direction to what I'm expecting, though.

The actual moment of failure was when I powered it on, and there was a loud 'pop' from the speakers. That switched off the device and a red LED was lit, which I assume is protection circuitry. There doesn't seem to be any evidence the protection circuity is active now; operation returned to normal (no red LED) . The computer logs the last protection event which it reports as SP_PROTECT "Speaker short-circuit: Amplifier circuit is abnormal".

I'm very happy to have made this much progress, but don't have a clear next step. Any good advice is appreciated here, many thanks.

 

[RandomEar]

Hi. Since nobody else commented, I'll give it a try. Let it be known that this isn't my field of expertise

• Since a short was documented on the speaker terminals, I would first check whether that still exists (assuming it did at some point). So measure the resistance at the terminals of the amp with the speakers connected. That should be around 3-8 Ohms.
  • If there is a short, disconnect the speakers and measure again.
  • If the short still persists with no speakers connected, check the output transistors for shorts. Use diode mode on your multimeter for this and google how to check transistors.
  • If there is no short, reset the device to factory settings as documented in the service manual, this will also clear the error cache. Power up the amp without any speakers connected and see if it behaves differently now.
• If the above testing detected no shorts, I would start testing voltages on the amplifier board. You checked the 28 V, but there should also be some 5 V or 3.3 V rails for the logic, right? Look for those and see if they are present. Also trace them to the most important chips. If those are missing, the main processor might not get the feedback that the amplifier board is powered and ready, causing it to never signal SPK_ON - just a guess, without studying the manual in detail.
• If all that fails, you could try using a thermal camera to check the board for overheating parts like caps, resistors or chips. If you don't have one, maybe you can borrow it from a friend or from work.

 

[mk2]

Thanks, sounds like great advice. You're right I should be able to test from the speaker terminals.

What I did do is test the speakers with another amp, and they seem to be ok; and I've already tried the factory reset. I was assuming if indeed it was a short and it still existed that the amp would drop into 'protection' mode with the red LED.

The concept of the amplifier board being 'ready' is interesting. Because the PSU has "P/S-1", "P/S-2" and "P/D" which I assume to be gradually enabling different power supplies when it's brought out of standby, followed by power down. It keeps P/S-1 high when out of standby; I don't yet see P/S-2 go high. Perhaps it only gets to P/S-2 if it decides P/S-1 is successful, maybe I'll see that in the schematic.

Thanks for taking the time to read my detailed description, let me see what I can find.

 

[RandomEar]

It's possible although maybe not very likely that the speakers were shorted by a stray strand of copper while originally connected and switching to a different amp cleard that short. So I would also check the speaker cables and connectors for this problem. It's much more likely that the error originated in the amp and now it clearly persist there but it won't hurt to chek the cables.

Tracing which PSU rail should power up when sounds like a good next step regardless.

 

[mk2]

I do doubt it's stray copper. The wire is properly connected into banana plugs, and not physically moved in years.

The installation is RIAA preamp -> Behringer digital processor -> optical -> Denon.

What does happen is the RIAA preamp normally makes a little pop when I switched it on. I thought maybe I had done this at the exact moment the Denon was coming out of standy. I wondered if such a stray signal could do something bad to the amp during its startup.

 

[mk2]

Testing the speaker terminals, and both pairs have high resistance. The multimeter starts at some lower number and slowly climbs. I don't know anything about amplifier circuits to interpret this.

If the power on happens in stages, I was reminded of the flow diagrams in the manual, specifically page 36 which seems to suggest what voltages to check, in what order.

Stage 1: MAIN PCB
It's suggesting to attach test points to a tiny IC. I couldn't do this, however I was able to find the +5V and +3.3V

Stage 2: AMP PCB
It was much harder to find test points
IC4006: on the back of the board, had to remove it from the chassis. Expecting +3.3V, got 0.174V
IC4004: expecting 1.8V, got 1.8V but had to do this without the MAIN PCB detached
IC4000: This is tiny and on the back of the board; it's infeasible to attach a probe

So perhaps the AMP PCB is abnormal and this prevents the MAIN PCB from ever raising SPK_ON (amongst others)

However I'm not sure how much I trust this guide. The infeasibility to get to the components and probe them these makes me suggest it's more a gesture than something intended realistically for repair. My mind is a bit boggled as to how to trace these through a mutli-layer PCB to anything I can touch with my multimeter. I don't have the experience to know if the missing 3.3V is just another symptom.

 

[RandomEar]

Testing the speaker terminals, and both pairs have high resistance. The multimeter starts at some lower number and slowly climbs. I don't know anything about amplifier circuits to interpret this.

If the value keeps rising, there is a capacitor charging up. This should be fine.

If the power on happens in stages, I was reminded of the flow diagrams in the manual, specifically page 36 which seems to suggest what voltages to check, in what order.

Stage 1: MAIN PCB
It's suggesting to attach test points to a tiny IC. I couldn't do this, however I was able to find the +5V and +3.3V

Stage 2: AMP PCB
It was much harder to find test points
IC4006: on the back of the board, had to remove it from the chassis. Expecting +3.3V, got 0.174V
IC4004: expecting 1.8V, got 1.8V but had to do this without the MAIN PCB detached
IC4000: This is tiny and on the back of the board; it's infeasible to attach a probe

So perhaps the AMP PCB is abnormal and this prevents the MAIN PCB from ever raising SPK_ON (amongst others)

However I'm not sure how much I trust this guide. The infeasibility to get to the components and probe them these makes me suggest it's more a gesture than something intended realistically for repair. My mind is a bit boggled as to how to trace these through a mutli-layer PCB to anything I can touch with my multimeter. I don't have the experience to know if the missing 3.3V is just another symptom.

The missing 3.3V is critical. Without it, at least some and maybe all logic chips on the amp board will not be powered. The fact that it is not zero, but some low value suggests that some power is supplied to the 3.3 V regulator(s). You should look for the 3.3 V regulators on the amp board and see if they are fed with the right voltage or if that is also low already. They will probably be attached to a 5 V line or something in that ballpark if they are linear reglators (most likely based on your photos) or maybe 12-15 V if they are the switching type (less likely).

If the voltage looks good before the 3.3 V regulator(s), they may be faulty themselves or a short will be behind them on the amp board. If they don't get the specced input voltage, the problem lies somewhere upstream of them.

 

[mk2]

Thanks for this information. I took another look methodically now I understand a bit more.

I think I have a step forward, here are the relevant tests:

• With the amplifier board fully disconnected
  • -9V reads +0.01V
  • +9V reads -0.25V
  • ±28V both fine
• With the amplifier board connected by comms only (no ribbon to the PSU)
  • -9V reads +0.01V
  • +9V reads +1.57V
  • ±28V both fine

Therefore I conclude

• the PSU is generating +28V when requested (probably by P/S-1)
• the 9V supply there is not enough information to know if this is broken, or just not being requested

Open page 44 of the service manual and do some more test:

• With the amplifier board fully connected (comms and PSU ribbon)
  • +28V is now 0.31V
  • -28V is now 0.25V
  • manual suggests to check IC4006 fo +5.15 and +3.3V
    • actual reading +1V and +0.8V

It's no surprise that IC4006 doesn't have the right voltage if the +28V supply is being pulled close to zero. But it looks healthy when the amplifier board is not connected.

Looking at page 44, the only other thing using +28V is "AMPLIFIER" units Q90xx.

Is it a sensible conclusion that literally the amplifier circuit has blown and is pulling the PSU line low?

I had assumed these were some kind of chip, but there is a full diagram on page 54. The +28V is the one critical for all the other voltages, but -28V is also affected, which is telling me something.

If so, I would guess that the board is not fully initialising, and the main board doesn't go any further; and one symptom is never activating P/S-2 to give the +9V that ultimately is needed to click relay RLY9000.

 

[LTig]

Is it a sensible conclusion that literally the amplifier circuit has blown and is pulling the PSU line low?

Could be. Disconnect the amplifier board and measure the resistance between +28V and -28V. Try normal mode and diode mode. If resistance is low (a few Ohms) the FETs could be blown.

 

[mk2]

I'll look up "diode mode" (that's new to me, but I'm learning)

However a "normal" test on my multimeter:

The resistance between +28V and -28V is 13.3ohm.
Between GND and +28V is around 250kOhm and falling
Between GND and -28V is around 250kOhm and falling

What's I did find after writing that last post is that there's a clean short across the pins of Q9023 and the same for Q9027. Whereas its "cousins" Q9015 and Q9019 do not show this. I tested these because they are the chips with a heatsink to the chassis. I'm assuming these are pairs of op-amps required for each channel.

I was hazarding a guess that I could de-solder Q9023 and Q9027 to try and remove the short and see if it is able to initialise.

 

[LTig]

I'll look up "diode mode" (that's new to me, but I'm learning)

The measuring voltage in diode mode is somewhat higher, so that you can measure a diode (requires 0.6V minimum). But ...

However a "normal" test on my multimeter:

The resistance between +28V and -28V is 13.3ohm.

... this is sufficient - it's too low.

Between GND and +28V is around 250kOhm and falling
Between GND and -28V is around 250kOhm and falling

What's I did find after writing that last post is that there's a clean short across the pins of Q9023 and the same for Q9027. Whereas its "cousins" Q9015 and Q9019 do not show this. I tested these because they are the chips with a heatsink to the chassis. I'm assuming these are pairs of op-amps required for each channel.

No, these are power transistors (FET type). So Q2013 and Q9027 (left channel) are blown and need to be replaced. It may not be sufficient though as it is possible that more transistors in the driving stage have suffered.

I was hazarding a guess that I could de-solder Q9023 and Q9027 to try and remove the short and see if it is able to initialise.

I think it's safe to try.

 

[RandomEar]

Those are probably be the power Mosfets for one channel. You should Google how to measure them in diode mode: It's sometimes possible to switch them to ON using just the test voltage of the multimeter, but not always. This would also cause them to measure "shorted" for a couple of seconds to minutes. However, in this case and with both of them being shorted and the other two being fine, it definitely looks like they're toast.

Desoldering them will be hard. They are connected to thick ground planes on the board. Those suck all the heat away from a soldering iron. You can try it with an iron and maybe some low melt solder, but it will likely require a hot air station. An additional pre-heater which is big enough for the PCB would make the job much easier.

 

[mk2]

You're right, I'm trying to desolder but it's not trivial. This is with my little USB soldering iron because it has a temperature control. I can soften the solder on the legs individually.

I suppose it depends if I'm confident they are bad and that this is definitely my next test I could just snip the legs.

 

[LTig]

You're right, I'm trying to desolder but it's not trivial. This is with my little USB soldering iron because it has a temperature control. I can soften the solder on the legs individually.

You need at solder iron with least 50W power, better more, and a bigger tip.

I suppose it depends if I'm confident they are bad and that this is definitely my next test I could just snip the legs.

Yep.

 

[mk2]

It was actually possible to detach the legs with my 20W at 450 degrees C, heating it quickly to limit the distance it could conduct. I think they were just more stiff than I thought. A little dab of flux helped, as I tried to avoid adding more solder. One of them was covered in glue, which didn't help.

With the legs detached, there does not appear to be a short on the board. Powering it on and the ±28V supply is now restored!

However all the rest of the behaviour is the same (including missing 9V supply and P/S-2 is low, no audio even through headphones)

So that's a small step, I suppose. It's rather late here, so I must stop for today.

I could replace these MOSFETs? But it's probably premature to do that without a proper understanding if anything else is broken and why. Without the MOSFETs, the main PCB is deciding not to initialise -- does it sound plausible it could detect that condition, and replacing them might just be seen in the same way?

 

[RandomEar]

I don't think the board detects the missing Mosfets now. As suggested by @LTig, there are likely some more defective components close to them on their power rail. I would check the transistors and diodes close by (looks like they are named Qxxxx and Dxxxx) for open circuits or shorts as a next step. After that, the question is what the other power rails like 5 V and 3.3 V are doing now.

 

[mk2]

Re-checking IC4006 and it reads 2V, 3.3V and 5.2V. That all seems correct now.

I can continue to re-check some of the diodes and things (though I'll be up against the limit of my knowledge here). Also it might be worth returning to the flow diagram on page 36 to see if that guides me to another failure.

 

[mk2]

Ok I will have to finish up again for tonight. There doesn't seem to be any obvious short among the 'nearby' components to the MOSFETs, and their behavior with a meter also seems to (broadly) mimic the circuitry for the other channel (though keep in mind I don't really know what I'm looking for) It's hard to methodically test them against their cousins, as I have removed broken MOSFETs; layout is different; and also everything is so tiny. Nothing looks visibly faulted (eg. no bulging capacitors)

If the amp won't initialise with missing MOSFETs in this one channel, how much chance is there that replacing them:

• blows them again; or
• still doesn't allow the amp to initialise?

When I am looking for parts cost seems to add up for such tiny components and, also, AI suggests I need "matched pairs" of MOSFETs but this doesn't seem to feature in any catalogue.

Page 36 didn't seem to have any useful wisdom; and it mainly seems concerned with testing chips on the main PCB, which seem ok to me as they drive the screen, inputs etc. and all required voltages are there.

I wonder if I am running out of road here. How does the main PCB detect this amp board as 'not working'?

 

[LTig]

You can replace the MOSFETs, and I expect them not to blow as long as you do not connect a speaker. I'm not very familiar with class D amps and switched power supplies, though.

I also expect that this is not sufficient to let the amp initialize, Could you do all the measurements (voltage) as stated in TROUBLE SHOOTING (page 34 and following) and report back which are wrong? This may give a hint.

 

[mk2]

Thanks for the good advice. I'll see how far I can get, as the problem with the troubleshooting is that it's mainly asking to probe pins of ICs which are too small to connect to without shorting, and mostly inaccessible when the PCBs are assembled anyway.

So far I have been trying to test all the ones I can. Page 36 asks for DIR_RESET which seemed like the next port of call but I was not able to find it at all (though checking now it might also be known as DIR_RST)