Roland JX 3P synth, 1983

Stored indoors for some years and then powering up, missing notes and some top panel sw functions missing. full sm here

formatting link
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.
Reply to
N_Cook
Loading thread data ...

Stored indoors for some years and then powering up, missing notes and some top panel sw functions missing. full sm here

formatting link
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.

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

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.

Reply to
Gareth Magennis

Stored indoors for some years and then powering up, missing notes and some top panel sw functions missing. full sm here

formatting link
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.

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

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.

Reply to
Gareth Magennis

Stored indoors for some years and then powering up, missing notes and some top panel sw functions missing. full sm here

formatting link
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.

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

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.

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

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.

Reply to
Gareth Magennis

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.

Reply to
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.

Reply to
bitrex

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.

Reply to
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

Reply to
N_Cook

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.

Reply to
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.

Reply to
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.

Reply to
N_Cook

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

Reply to
Jon Elson

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.
Reply to
N_Cook

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...

Reply to
bitrex

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.

Reply to
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.

Reply to
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.

Reply to
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

Reply to
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

Reply to
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

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

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.

Reply to
Gareth Magennis

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.