Line out distortion

[GregB]

Hi all,

Yesterday, I noticed my PSR-S950 was distorting the audio on the line outs.  Not at the headphone jack, not at the speakers, just at the line outs on the back of the keyboard.

Booting to the test program and using an RTA on my phone, I was able to verify the distortion, as well as to find that there's a little bit of left/right channel crosstalk too.  A glance at the service manual leads me to believe the NE5532 op amp for the line outs is blown; maybe I zapped it with static electricity (I'm in very dry air at the moment) when I was hooking the cables up?

Just curious if anyone else has encountered this.  From the service manual schematic, there appears to be some basic ESD protection on the line outs.

- Greg

[GregB]

Hi all,

After a bit more diagnostic work, I determined that IC5, a NE5532D, on the AJACK board was the likely culprit, so I replaced that IC this evening, which involved some surface-mount soldering work.  That has mostly resolved my problem; at least the audio is clear now and I'm getting 3.6 dBu output instead of -1 dBu on the line outs during the L/R channel test.  Note that the regular NE5532D is hard to get, but the NE5532AD is very common.  The only difference appears to be that the latter has a guaranteed maximum noise level, whereas the former only has a typical level.

One note about the AJACK board...  The schematic shows series chokes on the line outs and AUX in, but none of those three connections actually have them.  They're just plain ol' jumpers on the actual AJACK board.  I'm guessing they were omitted because Yamaha determined they weren't needed for EMI compliance, but this may have limited the ESD resilience of the AJACK board.

I modified my keyboard's AJACK board to add ESD protection (TVS silicon avalanche diodes) to the mic in, aux in, and line outs, and I added MOV protection on the DC 16V in.  The DC in already has reverse polarity and overcurrent protection, which effectively protects the MOV from overcurrent, too.  (note - I make no guarantee of the suitability or reliability of the modifications I made - my mods are rather hackish)

I may still have an issue with the mute circuit as well causing some remaining channel crosstalk, but that's a project for another day.   The mute circuit consists of three bipolar transistors.

(attached images: AJACK board with missing chokes L15 through L20; and my modified AJACK board with MOV and TVS protection)

- Greg

[attachment deleted by admin]

[overover]

Hi Greg,

Thank you for the detailed report regarding the repair and modification of your S950!


Best regard,
Chris

[Toril S]

I can only say WOW!

[GregB]

Ok, I've just placed the order for the mute circuit transistors (and some other supplies).  There's still definitely an issue with the outputs, even though they're now operating cleaner and closer to spec (now in-spec for output level, still out of spec for channel crosstalk).  Apparently mute circuit transistors commonly succumb to ESD.

At this point it would have been almost as cheap for me to just order an entire new S950 AJACK board, if it were easily available.  Yamaha 24x7 Parts does not show it in stock, unfortunately, and I didn't find any other US sources that had it.  Sometimes I have to remind myself that my S950 is 7 years old now (I've been eyeing the SX900 as it has some features that'd be useful to me, but we're not ready to make that investment yet).

- Greg

[overover]

Ok, I've just placed the order for the mute circuit transistors (and some other supplies).  There's still definitely an issue with the outputs, even though they're now operating cleaner and closer to spec (now in-spec for output level, still out of spec for channel crosstalk).  Apparently mute circuit transistors commonly succumb to ESD.

At this point it would have been almost as cheap for me to just order an entire new S950 AJACK board, if it were easily available.  Yamaha 24x7 Parts does not show it in stock, unfortunately, and I didn't find any other US sources that had it.  Sometimes I have to remind myself that my S950 is 7 years old now (I've been eyeing the SX900 as it has some features that'd be useful to me, but we're not ready to make that investment yet).

- Greg

Hi Greg,

Yes, the S950 AJACK board is relatively inexpensive. At the German company "Audio-Service Schierbecker, Hamburg" (www.audio-service.com) it costs around 27 euros. According to the information on the website, it is available, but they say "delivery time possible". (See also the attached picture.)


Best regards,
Chris

[attachment deleted by admin]

[GregB]

Huh, that's really odd.  The AJACK board, as they picture it, would not fit in the S950, as the board extends well beyond the connectors.  That part of the board would have to be cut or snapped off.

- Greg

[overover]

Huh, that's really odd.  The AJACK board, as they picture it, would not fit in the S950, as the board extends well beyond the connectors.  That part of the board would have to be cut or snapped off.

- Greg

Hi Greg,

maybe this is only a "symbolic" picture. As far as I know, this company sells only original Yamaha spare parts.That's why the board should actually fit if it's the right Part No.


Best regards,
Chris

[GregB]

Yay!  Replacing the mute circuit transistors fixed the remaining issues.

A few thoughts about this project:

• The repair involved surface mount soldering, which is not for the uninitiated.  I don't have nice surface mount soldering tools, but a set of very sharp dental picks and a headlamp certainly helped.
• I went ahead and replaced all three transistors in the mute circuit.
• The parts needed for the repair were one NE5532AD op amp (IC5), one 2SA1037 PNP transistor (TR7), and two 2SC3326 NPN muting transistors (TR8 and TR9).  (exact transistor part numbers I bought from Digi-Key: 2SA1037AKT146RCT-ND and 2SC3326-ALFCT-ND).  The muting transistors need to be that exact 2SC3326 part; they are a special device with unique characteristics.
• The overall characteristics of the failure:  1) distortion and L/R crosstalk on the line outs but not at the headphone jack or in the internal speakers; 2) when checking DC and AC voltages for the L and R channel tests in the S950's test program, there was an unusual DC offset on the outputs, the active channel voltage was too low, and the inactive channel voltage was too high; 3) when viewing the output spectrum on a spectrum analyzer on my phone, there was a lot of harmonic distortion when the volume control was above about 12 o'clock at the start of the channel test, when working correctly there is still a little harmonic distortion at full volume but it is about -30dB in amplitude.
• The cause of the crosstalk, DC offset, and some of the distortion was the mute circuit partially conducting when it should not.  Since it was partially conductive, the 0.6V forward bias voltage for a transistor was the main cause of the harmonic distortion, and it caused the DC offset by clipping the signal mostly in just one direction, causing the positive and negative halves of the waveform to become asymmetrical.
• The failed NE5532AD was reducing the output amplitude by about 5dB and also introducing some distortion.
• Overall, all of this appeared to be caused by ESD damage.  Touch something grounded before (and frequently during) setup and teardown.   The damage can be subtle at first, but doing the L and R channel tests and the overall mute test (a multimeter and some connections are required) should be diagnostic for this problem.  Using a spectrum analyzer (RTA) on the line outs during those channel tests should also be helpful in confirming what your ear is telling you.
• If you're not handy with surface mount repair, replacing the AJACK board is the easiest fix, if you can find the AJACK board in stock somewhere.

I hope this helps someone someday!

- Greg