PSS-A50 sounds stereo

[Lionel N]

Hello,

I focussed on this PSS-A50 product because I read many interresting articles from @pjd, on his site (http://sandsoftwaresound.net/yamaha-pss-a50-look-inside/)

I would like to contribute to the discussions about this, and especially share a piece of information, following my first investigations and prototyping on this PSS-A50 : This keyboard has stereo capabilities (the samples are in stereo), but unfortunately, the engineer only used/wired the left channel of the SWLL YMW830-V. They could save some cents on the cost of a dual OpAmp, using a dual instead of a quad.

I wired the DACLPP/DACLMM and DACRPP/DACRMM outputs of the SWLL to an external active Low Pass filter with a basic LM324 and a few resistors/capacitors (same schematics as for PSS-F50, and... indeed, the sound is now a very audible stereo for all samples. (I'm dissapointed because the Motions Effects the programmed are not stereo).

I uploaded a video with sound. Have a look there : https://youtu.be/cwDBfVab3vc

Enjoy,

Lionel

[pjd]

Hello Lionel --

Sorry for the late reply -- a hotel wifi ate my first response!

Nice hack!!!

I agree. I don't know why Yamaha didn't make the PSS-A50 stereo...

-- pj

[Lionel N]

Hello PJ,

nice article here : http://sandsoftwaresound.net/yamaha-pss-a50-stereo-mod/
Very pedagogic with nice illustrations. I also think the differential approach is relevant in order to lower the noise.

It's funny, because this week, I retro-engineered the Op Amp block, and... I get the same schematics.
Not so a suprise because the global picture is nearly the same as the one we can see in the PSR-F50 service manual.
Unfortunately, I could not find the service manuel for PSS- Remie series; so I wonder if you could find it or if you also re-constructed it from the pcb.
I confirm also the resistors values and the 1µF capacitors, but I could not get a solid value for other small caps; that's why your schematics is interesting, because even if I try to reconstruct using the circuit simulation https://www.circuitlab.com/; it's better to have the good picture.

The only major difference I have on my side is R501 and R502 for which I have 1,5k (instead of 2,2k on your side, which is not dramatic since the divide bridge is still 1/2 in order to add the half direct voltage.

My initial idea with this mod was a kind of proof of concept, in order to identify if its worth going deeper (i.e check if the stereo samples were present and palyed back). Which is the case.
Now, my target is to create a small PCB with the missing op amp, and replicate the mono circuit.
Because of the components shortage, and since I plan to make my PCB via jlcpcb.com, I identified three possible alternatives : TL972, TS462 or MAX4477. For availability and cost reasons, I'll use the TL972.
Now the question is : only replicate a "Right" channel or replicate both in order to have a common design on the same PCB ? Well, I'll replicate only one channel. In case performances is worse (ground loops, noise... it might be easier to fix), and in case performances is better, I will just have to use two PCB (probably stacked, so I have to think about the design) : one for the new right channel and on for the left channel that will go into the headphones amplifier.
The left channel path for the loudspeaker will remain, using the original NJM2740M Op Amp

Most probably, I will unsolder the 4 caps C601 to C604 and wire the outputs of my PCB. TBC.

Regards,
Lionel

[pjd]

Hello Lionel --

Great work!

I was thinking about my own mod and started searching for a small stereo preamp board with differential inputs. I think I would connect the SWLL signals to the preamp board and run the board in parallel with the existing electronics. That would minimize impact to the existing Yamaha DM board in case the experiment didn't work out. 

I'm a little nervous about working with SMT devices mainly because my hands aren't as steady as they once were. Nonetheless, those SMT traces are tiny!

Bonne chance -- pj

[sm7x7]

Hey, Lionel, how is going there? Any progress with PCB? I have two A50s, one already opened and waiting for modding ))

[Lionel N]

Hey,

Still thinking about some details. The main step I went thru was to remove C601/602/603/604 and adjust and solder a 8 pins 1.27 male smd header (2x4) in order to have something clean, easier to prototype in, and, in case I want to go back to the initial setting : plugging a 1.27 female header on it, with 4 caps soldered on the header.
I made some changes in my PCB design in order to add a breakable part for this "back to the past" plug. But have to admit that once you tasted to the stereo, you don't want to go back again in the one dimension audio world.

The latest news is that I implemented on a breadboard the missing Symmetric Low Pass filter for the right channel. Surprisingly, even with a basic LM324, the result is very satisfying, no major difference with the existing left audio from a noise point of view.

Meanwhile; I also started to explain again the progression in the blog I created especially for that : http://nollet.blog.free.fr/
I have to add the last details about the implementations I just talk above.

I also wonder if the low-end variants (PSS-E30 and PSS-F30) have stereo sample built-in and if this mod could also fit. Most probably yes.
I would also be curious to check what happens when using a PSS-F30 Flash memory DUMP on a PSS A50 HW (is dynamic touch managed, for example) ?
Same question for the Keytar Yamaha SHS-300 (don't know how the PCB looks like). Especially because the guitar samples on the PSS-A50 sound really great in stereo !

At this time, it seems no one try to DUMP the 16Mbits nor Flash of a PSS-E30/F30 SHS-300 "just to see what's inside"; or at least, could not find any on the internet.

If there's a really huge demand, maybe it would be interresting to crowfund an "easy to adapt" PCB.
But at this time, I don't want to spend too much money in having too many PCB prototypes that might have some design bugs. That's why I take my time to think about the best possible clean integration.

Lionel.

[Lionel N]

Hello everyone,

I uploaded my latest video that explains the stereo mod for the Yamaha PSS-A50.
https://youtu.be/bAUvMFMGzyE

Hope you will enjoy !

Lionel

[Lionel N]

Hello,

for your information, I uploaded a new video :
https://youtu.be/F4OMHS_O1EA

The PSS-A50 now integrates the manufactured PCB mod (Follow the Blog: http://nollet.blog.free.fr/)
In this video, the PSS doesn't play a midi file from the USB midi, but the full sequence is performed using the PSS- features and keyboard only.

Enjoy!
Lionel.

[Lionel N]

Hello,

For your information, Darekasan implemented his own version of the stereo Low Pass Filter integrated in a PSS-A50, using a Raspberry Pi Pico 1bit DAC shield. It provides an additional stereo audio jack output.
The shield seems to be easily available in Japan, and most probably only in Japan.

https://darekasan-net.hatenablog.com/entry/2022/09/19/045703

Lionel.

[pjd]

Hello Lionel --

Thank you for posting the links.

-- pj

[Lionel N]

Hello,

I recently bought a second-hand PSS-F30 in order to integrate my stereo mod PCB and confirm it is also valid for other keyboards of the PSS- series.

This time, I made the integration a bit easier because I did not use 1.27mm headers. I directly soldered some wires. I also just wired the missing DACR+/DACR- channel.

Now, the PSS-F30 also sounds stereo ! For some sounds (as for the PSS-A50) it's not so remarquable, but for others, it's really pleasant
I expected that the (many) rythms/styles integrated in this PSS- variant would sound much more beautiful, but in fact it's not so outstanding (but sounds better anyway).

Since I also bought this PSS-F30 for experimentations, I desoldered the SPI QUAD Flash memory in order to get a DUMP of its content and have a look on "what's inside", if anything remarkable.
I also plan to do the same with the PSS-A50 Flash memory in order to compare them and maybe -replace some voices- or use voices from PSS-F30 on PSS-A50 ?

I'm not sure that someone ever performed a DUMP of the flash for this/these instrument(s), so maybe "I'm the first" !
For copyright reasons, I won't share the content, but you'll get the information on my Blog about how I proceed.

Currently, it's not so clear for me about the architecture, especially if there is part of the SW in the YMW830-V Processor (it has a JTAG, so most probably a code memory). Most propbably, there could be a bootloader in the YMW830 (maybe the same for all PSS-series). Nothing happens if we turn the PSS- ON (while the Flash is not connected).
It looks strange because it's not possible to enter in a "test mode" (explained in the Service Manual), and this test mode is supposed to help to identify (error code) if the Flash is not connected. Anyway.

I'll also identify where the non-volatile data is saved (volume setting for example) inside YMW830 or in the external Flash Memory. Just needed to perform a DUMP with the default settings, and another DUMP with the value changed. If nothing changes in the external Flash, it means that the data is stored in the processor.

I'll do the same with the PSS-A50 in order to try to locate the value (if any for the PSS-F30) that could be used to define the "Touch sensitivity" : Currently, the PSS-F30 HW doesn't implement dynamic touch. The chance is that the keyboard PCB and the rubber keys/contact is nearly the same as the one of the PSS-A50 : dual keys (press and release) are present. Just, a diode is missing for each key. Also, just one wire is also missing between the Keyboard PCB and the MainBoard PCB. So maybe, it might be possible to active a dynamic touch in the PSS-F30, if the SW implements the function and if it can be activated dynamically by changing a value in the memory.

I also noticed that the midi messages are sent out on the Tx pin of the YMW830, and that could help to implement a wire(less BT) midi to record what you play !
Talking about midi, I could easily extract (from the flash DUMP) the midi files for the 30 songs stored in the PSS-F30. I did it for the first song. In the DUMP, the songs are stored exactly like any .mid file on your PC.

To be continued...

As usual, some details here : http://nollet.blog.free.fr/

[pjd]

Hi Lionel --

Just saw your post -- Nice work!

As to Yamaha's use of Standard MIDI File (SMF) format, Yamaha use that representation/design approach quite frequently. That's why style files, VCE files, etc. are SMF in disguise. 

I will read your message more carefully tomorrow. It's almost rehearsal time...

Good work, thank you -- pj

[pjd]

Just got a minute while cooking dinner. 

The 12-wire version (velocity sense plus odd-numbered diodes) adds an extra sense line to each six-key group. The group drive lines pass through the 7-pin connector.

Velocity sensitivity needs all of the diodes as the software measures the closure time between the two wires (switches, diodes) associated with a struck key. Each key belongs to a six-key group and keys in a group share a scan drive line. The drive line is one side of the switch pair for each key. Then, the other side of the switch pair sends the switch state through the diodes into the two sense lines. The software measures the time between closure of the first switch and closure of the second switch. (And doesn't confuse.)

I posted the Reface switch matrix schematic at:

http://sandsoftwaresound.net/random-answer-day-1/

The PSS switch matrix is the same modulo missing diodes and sense wire.

Hope this helps -- pj

[Lionel N]

Hello,

I'm back again with another mod. Firmware mod this time.

What if PSS-A50 had a built-in BossaNova rhythm ?!
PSS-A50 has internal arpeggios (part of them are rhythms), while the PSS-F30 has a rhythm+bass accompaniment.
PSS-F30 has a BossaNova rhythm but PSS-A50 unfortunately does not have it.

I own both PSS-F30 and PSS-A50. I spent a long time to deconstruct the Flash ROM of both keyboards and retro-engineer them.
I also modified them to bring back the stereo sound.

Now, I'm happy to share I own the first PSS-A50 with a built-in BossaNova arpeggio. To do this, I converted the data (BossaNova section) from the F30 Rom to a format compatible with the A50, and I wrote these data in the A50 flash (I overwrite the last-in-the-list "African" arpeggio). Now, Arpeggio 138 plays BossaNova (only rhythm, no bass accompaniment) !

Have a look on my Youtube channel : https://www.youtube.com/watch?v=W0INlS42ggM&t=188s

Lionel

[pjd]

Now, I'm happy to share I own the first PSS-A50 with a built-in BossaNova arpeggio.

Have a look on my Youtube channel : https://www.youtube.com/watch?v=W0INlS42ggM&t=188s

Lionel

Lionel -- This is positively the most amazing hack!!!!! 

-- pj