Roland JX 3P synth, 1983

[N_Cook]

Stored indoors for some years and then powering up, missing notes and
some top panel sw functions missing.
full sm here
http://www.synfo.nl/servicemanuals/Roland/JX-3P_SERVICE_NOTES.pdf
same 2 notes ,not each octave , but as keyboard is scanned as 8x8 matrix
keys 8n and 8n-2 are missing , n= 1 to 8.
Hopefully that means the 74LS138 3 to 8 demultiplexer did not survive
1/3 century,internal metalisation creep or something, rather than main
micro problem.

[Gareth Magennis]

"N_Cook" wrote in message news:o8s467$88f$1...@dont-email.me...

******************************************************



It is more likely dead VCF chips, IR3109.

Scope around this area and see if they have any input/outputs.


It will not be a uP problem.


Gareth.

[Gareth Magennis]

"N_Cook" wrote in message news:o8s467$88f$1...@dont-email.me...

[Gareth Magennis]

"Gareth Magennis" wrote in message news:RSFsA.624849$9Y7.3...@fx27.am4...

***************************************************


Well, I may have been a little hasty there, DMUX chips do blow as regularly
as VCF's.

Point is, this is where you should first be looking.
If the VCF has input but no output, then either the VCF is dead or the
control voltages from the DMux's are bad.

Any other fault is rare IMHO.


You can check whether the keyboard is processing every note press by
monitoring the MIDI out socket.
A simple LED shoved into the two pins either side of central will show all
MIDI activity quite nicely.


Gareth.

[bitrex]

Did you try doing a factory configuration reset? These old Rolands use a
battery backed SRAM to store both patch data and for the system
scratchpad RAM, if it gets corrupted you can get all sorts of problems
like the ones you describe.

[bitrex]

That is to say it stores front-panel settings etc. in the battery backed
RAM during power-down, and if they get corrupted to garbage values the
keyboard stops triggering correctly, etc.

[N_Cook]

no factory reset sequence found for this model. Reseated the socketed
memory JIC, but no change
As some of the function switches are out as well as keyboard, part of
same polling structure, i'll go with a problem on the digital side
rather than VCF/analogue. Continuity back to 40245 ok , but as 2 of the
octal lines out , presumably back farther than that, have to do some
digital probing after finding bench space.
One tip for dealing with these, small cable ties between top and base,
to act as a hinge for probing access to the internals and keep
everything in reasonable registration so no strained cables.

[N_Cook]

Appropriate timing response for "dead " key presses cf working adjascent
keys , in the signal lines, so will transfer my attention to
interpretation / output side of things

[N_Cook]

On 26/02/2017 19:31, Gareth Magennis wrote:
>
>

I'd forgotten about Midi monitoring and your flashing simple LED tip is
much more convenient that an earpiece continually dropping out.
Confirmed keyboard+switches scanning problem, presumably a stuck address
line so next to find which one, then time to remove the mainboard and do
some trace cutting.

[N_Cook]

Cold DVM-D test to gnd of D5 line showed 1.07V "reverse " and D7 1.11V,
other data lines 1.02V. So planned to cut and swap D6 and D7 lines at
the 40245 to see if dead key stayed at position 8 or moved to 7.
Plenty of space to drag back the point of a "diamond" needle file to cut
the traces at 40245. DVM-D test again before swapping over and same
readings.
So cut same D5,D7 traces to IC43 LS373 and retested, LS373 pins both
read 1.2V now and main uP pins up to 1.9v ,so presumably some
metalisation creep/problem interconnecting pin on the LS.
Make good traces, remove LS373 and replace looks a sensible move next.

[N_Cook]

Old removed, tuned pin 20w socket , in place, easier to check continuity
etc, finish that tomorrow and push in NOS LS373.
Loads of black corrossion product on the pins of this IC and only this
one , no other TTL,4000 or analogue, whatever that means, outgassing
something ?. Cracked it open and frame metal was clean on the internal
pins I could see where free of encapsulant.
Texas Instruments, SN74LS373N, Malaysia 8206X if a batch problem.

[Jon Elson]

Well, not clear if the original chip was in a socket.

There was a certain vintage of TI chips that had a silver plating that could
turn absolutely BLACK! These could certainly cause bad contacts in sockets,
but I think I've seen a couple times where it was so bad it got down and
isolated the pin from the through-hole in the board. I thought these were
before 1982, back in the mid-late 70's or so, but must have still been doing
that plating later. I'm pretty sure it didn't come from inside the chip,
but the effect of atmospheric sulfur on the silver plating.

Jon

[N_Cook]

All that pallaver of swapping corroded 74LS373 (but good) and I'm back
to loss of Data5 and Data7 keys but now D6 keys gone west also,
continuity checks out fine and should be there as before.
D0 to D7, correspond to keys 1 to 8 of each octet of keys, not octave.
So went back to original plan, but now with no D6 to swap to D7.
So cut the trace to D3 input of the 40245 keyboard scan IC and fed the
D7 line to that pin. Powering up, the 4th key of each batch of 8 now
sounded at the pitch of "7" not "3" and midi LED flashed.
So looks like swapping out the 40245, I think, but perhaps wrong, as it
seems functioning correct signal on D7 line but not getting through the
40245 nor D6 nor D5, input to outputs , for keybd scan.
Pitch number 8 of each octet would sound (and 6 and now 5) , if the
keyboard scanning funtion was working correctly, I hope.

[bitrex]

My Alpha Juno 2 from around the same timeframe had a problem with the
membrane switches on the front panel not activating correctly; I traced
it down to a bad pullup resistor in one of the SIP packages attached to
the bus headed from the switch board to the demultiplexer IIRC.

Might be worth looking at too...

[N_Cook]

That makes sense, for losing a third set of keys, as in the process of
probing the 40245 I bent the data bus SIP R more vertical slightly, to
give a gap, as was leaning against that side of the IC.
But otherwise I don't remember having come across a failed SIP R pack.

[N_Cook]

Replaced TC40H245P with 74HC245 and all keys now working.
Cold DVM-D test of the old 245, to either supply pin, showed no forward
drop in any direction, to the gates used by D5,D6,D7 unlike to the other
pins.
Some of the front panel switches came back to life but switches 1 to 8
of 16 bank, no function or Midi LED flash. Hopefully just failed
"ribbon" 33 yearold age hardened copper wire that gave up the ghost with
all this unaccostomed flexing of connecting and disconnecting the main pcb.

[N_Cook]

Must have been sat on or similar, to squash area near sw16, down onto
the ps h/s and cracked the pcb.
There is a H&S factor with these, bad enough that main pcb is secured to
chipboard base by wood screws, so loosish screws that could fall away.
But the mains carrying pcb is also just woodscrewed to the chipboard as
well.

[N_Cook]

Before I got to this, an electrician had removed all the keys without
some sort of sequence marking, so I ended up with a carrir bag of them.
I assumed out there, via google bunging in
roland keyboard keys, CF1,EB2,CF, D, A ,G , C ,23
someone would have put out there the sequence required to lay up a
Roland keyboard moulded with those cryptic alphanumerics, but apparently
not. Any pics I've found is ambiguous as the degree of left and right
throws on the white keys, let alone the unseen pivot slot widths that vary.
C D E F G A B C is not very helpful

[N_Cook]

Looking at an old Casio keyboard, for white keys the direction of throws
at the leading edge is, per octave
CDEFGAB
right,double equal,left,right, double biased R, double biased L, L

The Roland moulding number sequence of each octave is
EB2,CF1,D,EB2,CF1,G,A,EB2
top C is CF and black notes 23

Lay out the white keys , just laying on the top, mark the pivots that
take the black keys, remove the white keys, set all the black keys in ,
then all the white keys, finally set the springs on with a hook. holding
the fixed static end with a finger to avoid the spring flying off

[Gareth Magennis]

"N_Cook" wrote in message news:o9p9u4$61i$1...@dont-email.me...

***************************************************************


I can't make head nor tail of the above, but putting keys back is not
usually a problem.

Put all the black keys in first.
White keys marked "CF" are both C and F keys, they are identical.
Similarly "EB" keys are both E's and B's.
D keys are "D".
G keys are "G".
A keys are "A".

Top C key is unique.




Gareth.

[N_Cook]

I've only ever had to repair an odd broken key or 2.
The sort of job where you could do with an apprentice. Show him how to
fix one key and let him get on with the rest.
What we were given 5 fingers on each hand for. One finger to hold the
spring against the frame, so it doesn't fly off, one finger to hold the
key back against the pivot and another finger to push the other loop of
the spring over the spigot, while pulling the spring with a dental hook
in the other hand , and then a satisfying clunk.