Adding PC-Link to MDS-JE640 (like MDS-S50)

[sfbp]

hmmm if you can figure out how to add PC-Link to the 640 I for one would be eternally grateful.

[kgallen]

6 hours ago, sfbp said:

hmmm if you can figure out how to add PC-Link to the 640 I for one would be eternally grateful.

Assuming one could get the right parts especially the connectors then the PCB pads are there for the taking. Is this an S50 option? So the hardware is probably do-able in a fashion (connector bodged onto the back panel for instance). The firmware might be the issue. I don’t know if they load all these machines with the same uC firmware. There are pins on the uC labelled ‘model’ but these don’t seem to be tied either way on the 440 or 640 - often 0R links are used to set a binary code. My 440 knows it’s a 440 (from info in Service Mode) but it’s not clear how it knows that unless it has a fixed 440 firmware load. 
(@sfbp are you able to break this out and create a new topic with the last couple of messages here to preserve the OPs thread?)

[NGY]

6 hours ago, sfbp said:

hmmm if you can figure out how to add PC-Link to the 640 I for one would be eternally grateful.

I am actually working on this. Been there a few years ago but dropped the topic for lack of proper time. Picked it up again recently, and I seem to be close, all I need is harden my heart and open my own, mint 640 for the "operation".

27 minutes ago, kgallen said:

Is this an S50 option?

Yes, it is.

And this "universal" JE640 MB does have the printed wiring for the PS2 socket, as the same board is used in the S50 too.

I have just bought a spare S50 some weeks ago -  with a badly broken facia, but intact and fully functional inside. First I want to do the opposite way (make the PS2 connection accept a keyboard instead of PC-Link) for a sort of proof of concept, before I go and butcher my 640.

27 minutes ago, kgallen said:

The firmware might be the issue. I don’t know if they load all these machines with the same uC firmware.

Sadly it is not only the FW, but even a different uP. The S50 uses a version with built in ROM/RAM, while the 640 has an external memory chip.

[sfbp]

27 minutes ago, kgallen said:

(@sfbp are you able to break this out and create a new topic with the last couple of messages here to preserve the OPs thread?)

Man, it's the middle of the night here. But you could do it easy-peasy, if you can decide which ones you want to move. Beyond me right now..... start by checking the box of each one you want to split off.

[kgallen]

9 minutes ago, sfbp said:

Man, it's the middle of the night here. But you could do it easy-peasy, if you can decide which ones you want to move. Beyond me right now..... start by checking the box of each one you want to split off.

It is, so what the heck are you doing up on the Forum - get some shut-eye! :-P

[kgallen]

1 hour ago, NGY said:

Sadly it is not only the FW, but even a different uP. The S50 uses a version with built in ROM/RAM, while the 640 has an external memory chip.

The external Flash chip, IC2, is padded out but is is populated? In my 440 the chip is not populated, but IC2 is shown on both 440 and 640 schematics. I see similar in my two MDS-E10's - one has the external Flash chip populated the other doesn't. I wondered why this is. Is it CYA for when the firmware grows too large for the flash inside the uC? Or maybe a cost option, external flash might have been cheaper at some point than the next uC up the scale with larger internal Flash? What I'm saying, is the population of this Flash chip really feature dependent or just an evolutionary precaution or an option for BOM cost tradeoffs? i.e. having the flash or not, is not necessarily an indicator of functional capability.

In the S50 SM, page 40, the schematic outlines a few components in blue, and some not. The external Flash, IC2, is not outlined. The footnote says "There are a few cases that the part isn't mounted in model is printed on diagrams" (sic). In fact on p42 of the S50 SM, which is the uC sheet, the external flash is not drawn. S50 uC=M30805MG-211GP [144 pin QFP], 440/640=M30805SGP [144 pin QFP]. I'd need to investigate the uC datasheet to see if there is enough information in these part codes to indicate a different variant of the part is selected. S50 timeline suggests it came out a few months after 440/640 so it's possible Mitsubishi revised their uC family slightly and Sony started to order using the new part code - when under the hood they had the same requirements on the uC for internal RAM and flash etc.

[NGY]

10 minutes ago, kgallen said:

The external Flash chip, IC2, is padded out but is is populated?

Yes, it is in my 640 - and in my other two 640-s I already sold it was also the case.

10 minutes ago, kgallen said:

In the S50 SM, page 40, the schematic outlines a few components in blue, and some not. The external Flash, IC2, is not outlined. The footnote says "There are a few cases that the part isn't mounted in model is printed on diagrams" (sic). In fact on p42 of the S50 SM, which is the uC sheet, the external flash is not drawn. S50 uC=M30805MG-211GP [144 pin QFP], 440/640=M30805SGP [144 pin QFP]. I'd need to investigate the uC datasheet to see if there is enough information in these part codes to indicate a different variant of the part is selected.

Both my S50-s are with this CPU, and w/o the external memory chip. Had a third one some years ago, I just need to find the photos I had taken, to check it, but I suspect that one had the same.

The datasheet does have reference to which P/N is to use with external flash and which are embedded. But I don't see Sony's preference and reasoning behind chosing this or that in different models of the same family.

[kgallen]

13 minutes ago, NGY said:

But I don't see Sony's preference and reasoning behind chosing this or that in different models of the same family.

It's perfectly possible this comes down to cost and device availability. If Sony have a number of population options in their products they can go with the flow and choose the cheapest option at the time of build, or are more robust against yield issues on a particular part. Silicon memory faults are the most likely cause of device yield dropoff, particularly on-chip flash which will require different processing steps and different silicon implants in the wafer fab compared to SRAM and random logic which are more "standard". Mixed-process devices are more problematic (and it's very impressive that the Sony CXD chips have on-chip DRAM mixed with "normal" VLSI digital logic, SRAM and probably even some analogue interface circuitry - very impressive for the year, and Sony are "fabless", so they don't have the luxury of in-house processing they can tune and tweak like Intel). If all machines of a given model have the same device population then that would indicate an engineering requirement. Where there is population variability it probably is a decision made on BOM cost at the time.

[kgallen]

There seem to be conflicting usage of the I2C bus from the uC. On the 640 this bus is used to implement the keyboard interface. On the S50 it's used via an I2C buffer chip to implement the PC-Link. So from what I can see keyboard vs PC-Link are mutually exclusive and there is likely different firmware in the uC for each variant to account for this.

FWIW the I2C buffer chip, IC800 is a standard NXP part, P82B715TD .

[sfbp]

IIRC the *Japanese* version of the JB940 actually allowed both keyboard and PCLink. However I never got any 940 in my hands, and I never got the Japanese SM for any of these decks.

[BearBoy]

I think you're right sfbp, the page for the MDS-JB940 on MiniDisc.org says that it can be used with PC Link (PCLK-MN10) but there is no mention of this in the manual for my MDS-JB940 (UK model).

Looks like the PCLK-MN10 connects via the optical digital input on the deck. I wonder whether it would work with my UK model but they just didn't mention it in the manual as the PC Link kit was unavailable outside of Japan? Or whether the necessary "innards"* are absent?

 

 

* Technical term ;-)

[kgallen]

35 minutes ago, BearBoy said:

* Technical term ;-)

LMAO!

Looking at the schematics I would summarise as:

- the JE640 keyboard interface is input only for the data. The buffer stage they've implemented here works keyboard -> uC only for data (the clock, SCL, is bidirectional since it's driven out to the keyboard but in the I2C protocol the slave (the keyboard) can hold the clock low as a flow control mechanism).

- the S50 PC-Link interface is pure birectional I2C. I suspect the deck uC is the bus master but data can flow in either direction. The implementation uses a specific I2C buffer chip (listed above) to buffer the I2C bus from the uC to be able to drive out on a "long" cable to the slave device (this MN10 thing that is mentioned).

 

So it's perfectly possible that on the JP model, they've hung both of these buffer circuits from the uC-side I2C bus (since I2C is specifically designed to work like that). Indeed if I look more closely at the S50 vs 640 schematics, maybe if both sets of components are populated on the PCB, then given the firmware support within the uC this support may indeed be possible on the 640 (subject to uC equivalence flagged by @NGY earlier). Indeed looking more closely at the 640 schematic, the discrete buffer stage is between the true I2C bus, which does appear on the KB connector *on the main PCB not necessarily on the deck connector* and a separate set of "KB" I/O on the uC. It looks like the PC-Link also uses this (non-standard I2C) "IIC-BUSY" signal which is not brought out on the 640 KB interface. More investigation needed, leave it with me...

 

 

[kgallen]

Related from the MDS-PC3 SM, captured for future reference...

 

 

[sfbp]

(I'm pretty sure the PC3 does NOT support a keyboard)

The other place that might be interesting to look is the CMT-PX3 (aka DHC595MD). Mine is a Japan model and the PCLink definitely works. However chunks of the (English) SM don't correspond to reality, at least for my machine.

If we're desperate, my aged Pa has one of these now stashed in my sister's garage in Surrey. This is the UK model, of course.

[NGY]

58 minutes ago, kgallen said:

More investigation needed, leave it with me...

Seems like you are going down the same route I did during past few months. I did not feel like sharing all my (wrong and good) hypoteses and conclusions of my investigation on the go, only post here the end result, but to speed up your work I can share with you my findings, that led to the solution (I believe would work) to be implemented in practice, as the last remaining step. My benefit then would be saving time and leave my 640 intact :-) .

[kgallen]

Is there a machine with KB and PC-Link? I'd like to look at the schematic...

[NGY]

13 minutes ago, sfbp said:

(I'm pretty sure the PC3 does NOT support a keyboard)

No, it does not - nor does the S50 either. For the very same reason.

6 minutes ago, kgallen said:

Is there a machine with KB and PC-Link? I'd like to look at the schematic...

Yes, there is, the Japanese version of the 940. Nobody seems to have an SM for that, and I for one had been searching for it on the net extensively, but no luck so far ... that would have this made very easy.

[kgallen]

43 minutes ago, NGY said:

Yes, there is, the Japanese version of the 940. Nobody seems to have an SM for that, and I for one had been searching for it on the net extensively, but no luck so far ...

Hardware wise I would expect the 640 MB populated with the I2C buffer associated SMD resistors and DIN socket would be the hardware solution, because I2C can work like this (maybe also need to check for a single pullup resistor on each of SCL and SDA). The uC and firmware support is a harder question to answer and possibly only answerable by "just trying it" and the 640 seems the only viable candidate. My 440 is open to experimentation but it is at least two steps away from where we want to get to since it has neither KB nor PC-Link interface. If we did come up with an answer, I'm not sure Stephen's 50W plumbing iron is up to reflowing 0803 SMD resistors or TSOP ICs :-P

44 minutes ago, NGY said:

the last remaining step

How can I help answer your "last remaining step"? (The link took me to an earlier post in our conversation).

[NGY]

5 hours ago, kgallen said:

How can I help answer your "last remaining step"?

If you can open a 640, and do the mods needed, and check the result - this is all what's waiting. I was kinda hesitating to do it on my saved piece, but a poor 640 still could not pop up amongst so many ads flying around here, I could only get an S50 that I could sacrifice.

(The I2C buffer is really not an issue. I ended up believing that the whole thing can work without it, actually. And I do recall I had found some proof for that, but let me check my notes for it).

[kgallen]

10 minutes ago, NGY said:

I did not feel like sharing all my (wrong and good) hypotheses and conclusions of my investigation on the go, only post here the end result

There is collective nous here that I like to share my ramblings, since one of you fine chaps will either spot the error of my ways, bring some extra wisdom, or promote a new idea by the mechanism of "cardboard engineer" - that of just writing down one's own thoughts bringing clarity or a new insight into a problem. And long may it continue, fellow MD-ers!

[sfbp]

FWIW, I seem to recall that my Japanese MXD-D5C, which works with PCLink, does NOT support the keyboard.

[kgallen]

3 minutes ago, NGY said:

If you can open a 640, and do the mods needed, and check the result - this is all what's waiting. I was kinda hesitating to do it on my saved piece, but a poor 640 still could not pop up amongst so many ads flying around here, I could only get an S50 that I could sacrifice.

I truly understand not wanting to open a prized machine - I still hold that for my 520 which is in truth nothing special in this world, but held dear to me for no tangible reason other than sentiment.

However @NGY it sounds that, much to your emoted chagrin earlier, you are in the ideal place to acquire such an experimental beast since your eBay world seems to be much more tempered than our own!

[NGY]

2 minutes ago, kgallen said:

There is collective nous here that I like to share my ramblings

Nothing wrong with that, I just normally feel a bit "shy" to do so :-) . See my LPM thing ...

[sfbp]

https://www.ebay.com/itm/SONY-MDS-JE640-MD-RECORDER-ATRACK-DSP-TYPE-R-ATRACK3-TESTED-WORKING-TGJ/392947208881

Haha, this is a Japanese version of the 640 and it says (around) the keyboard port "PC-LINK (keyboard)"

How mysterious is that?

[kgallen]

4 minutes ago, NGY said:

See my LPM thing ...

Which is a crazy case in point. Several of us were most impressed by this piece of work - and one of the first truly inspirational posts I found when I joined this forum. And we would still like more details! Myself, I got as far as breadboarding the photodiode and transimpedance amp circuit, but got overwhelmed by the mechanical challenge of getting it inside an MD case - something you had achieved with grace.

I have the Sanwa LP-1. And whilst this is a small device, using it on my 440 is still a clumsy affair, getting the head near to the OP. I would still be wary using this with my other "more valuable" machines. So your own solution is still a desirable project....

We can but ask nicely and hope you overcome your shyness :-)

[kgallen]

21 minutes ago, NGY said:

(The I2C buffer is really not an issue. I ended up believing that the whole thing can work without it, actually. And do I recall I had found some proof for that, but let me check my notes for it).

It probably would work but not be robust. The uC I2C ports are not designed to drive a cable. Having a read of the datasheet for that I2C buffer gives the background of why that chip is required in such I2C topologies wrt capacitive load and current drive etc. If the USB interface inside the MN10 thing were mounted inside the 640 and wired directly in, I suspect the buffer chip would not be necessary.

[NGY]

8 minutes ago, sfbp said:

this is a Japanese version of the 640 and it says (around) the keyboard port "PC-LINK (keyboard)"

Great find! (Almost jumped on it, but the price ...)

And I am 99.99% sure it uses the same main board pcb as above - see the TOSLINK in/out connectors!

[NGY]

42 minutes ago, kgallen said:

The uC I2C ports are not designed to drive a cable

No, certainly are not, however, the I2C spec. does allow a certain length of cable, at given cable capacitance and shielding, and at a max. data rate - I am still trying to find the exact quote I noted previously, but until then: "the maximum bus length of an I2C link is about 1 meter at 100 Kbaud, or 10 meters at 10 Kbaud".

So my point is (and I did drop the ball here already), that in order to get this PC-Link thing working on an otherwise not-supported deck does not depend on having this buffer or not (just keep the cable "short", and the PCLK's PS2 cable is indeed very short, I believe Sony's engineers had done their math for a reason.)

[kgallen]

6 minutes ago, NGY said:

however, the I2C spec. does allow a certain length of cable, at given cable capacitance and shielding, and at a max. data rate

Well I guess we could argue details, but I2C is "Inter-IC" so not really designed/intended for off-board driving over a traditional cable - you'd use a spec like RS232 and it's brethren for that.

[NGY]

Yeah, see my addition above.

So much I hate these messenger/twitter/whatsoever style posting - you cannot really think your words over, or you missed the train ... I go and take a brake break instead ... :-)

[kgallen]

1 minute ago, NGY said:

So much I hate these messenger/twitter/whatsoever style posting - you cannot really think your words over, or you missed the train ... 

Sorry, my bad. Here I go again... :-(

[kgallen]

I2C spec - used to be quite hard to come by, but easily found on't web these days here.

There are three traditional modes, standard-mode @100kbps, fast-mode @400kpbs and Hs-mode @3.4Mbps (not really ever termed baud on I2C).

 

[kgallen]

Sorry for the cut-and-shut picture:

I wonder if you chaps have a 4-pin or 6-pin connector for CN820 on your 640 main board. Then if R826 and R822 are fitted on your KB board (probably). If so, then I think you could rightly label your KB connector as "Keyboard (PC-Link)" with no changes...

[sfbp]

Well, I absolutely know that my 640 won't work with PCLink. But I will go and look at CN820....

Why does it say "jumper" on each of those two components?

[kgallen]

26 minutes ago, sfbp said:

Why does it say "jumper" on each of those two components?

Zero-ohm link. Commonly used to option in and out circuit elements within a product range. See them all over the place. Could be a wire link but more usually these (and those) days a chip "resistor", 0R.

[kgallen]

13 hours ago, sfbp said:

But I will go and look at CN820....

On the 640 it's next to the rear panel Control-A1(II) connector. This picture is flipped. For you with the lid off, looking from the rear and down it will be to the right just above the rear panel just down from the power supply section (big black capacitors)/just across from the S-shaped heatsink.

[sfbp]

This is from my last visit to this particular quagmire (I had blocked it out because I got exactly nowhere). I think you will find I've done the obvious connections.

[kgallen]

6 minutes ago, sfbp said:

This is from my last visit to this particular quagmire (I had blocked it out because I got exactly nowhere). I think you will find I've done the obvious connections.

Well there you go, you've been there, done that! No point in me interfering here any more, you've done it!

Hardware is fixed. That's me done. Firmware then. Over to you :-P

[sfbp]

gaaaah,,,,, no bloomin' clue, nor how to debug it either.

[M1JWR]

not that i know what you lot are on about,i will just add that the 470 has pc link,dont know about keyboard function though, just my 2p worth to confuse matters even further !