How does this inductive musical-keyboard function?

[yinyue]

It's bedtime, but I have a question that make me insomnia. I've been looking to interface with a keyboard that I've removed from a non-functional system, specifically an Ensoniq EPS-16. As luck would have it, I managed to find schematics for the controller board, which handles the output of inductive sensors underneath the keys, and hands it on to the main controller.

I have reason to believe the controller board may be faulty (or at least, I am having a lot of trouble getting input or output from the MC68HC11 http://www.kynix.com/Detail/425510/MC68HC11.html micro controller on the board) and I can not find any documentation about the proprietary IC (Motorola branded S38BC010PS01) used, processes the key output signals.

My plan is therefore to read the data directly from the keyboard signals which is where I'm having trouble. I can so far deduce that there are 4 input select lines and 4 data output lines, the input lines, CS0 to 3, drive a set of 4 to 8 demux ICs (74LS138) which select the key/bank of keys to poll data from.

From there, the output of the demux lines is fed through a strange network of transistors and the individual key inductors, and then into a bizarre use of inverter gates which are then directed to 4 output lines F0 to 3. I have no idea what kind of output signal I should be expecting. At rest, all of the output lines sit at 2.5 volts does not change if I press anything.

Everything I've described is shown in full in the schematics KBD-LOWER with the rest of the keys in KBD-UPPER. The main input/output header leads into the KPC Controller board where the micro controller and the other proprietary chip process the information.

Someone else with a similar mission to mine posted here that the output signal is a 3MHz rectangle that adjusts with key presses but I'm not seeing how it could be produced, and my only analog scope is low frequency and doesn't detect any worthwhile changes in the 2.5 volt signal.
If you have any idea, I'll feel happy. Thanks all.

[MintyTheCat]

Hi,

My plan is therefore to read the data directly from the keyboard signals which is where I'm having trouble. I can so far deduce that there are 4 input select lines and 4 data output lines, the input lines, CS0 to 3, drive a set of 4 to 8 demux ICs (74LS138) which select the key/bank of keys to poll data from.

138 is a "3 to 8" mux with an chip-select line on S1. 2A and 2B split the selection lines into two sets of four - purely from memory but I can check if this sounds iffy.
I forget if 138 has active high or low pins for input but assuming active high to select Y2 you input CBA as: LHL which is "2" in binary.

What type of test equipment do you have? I would get a scope and check along the path the levels with a key press and without a key press. Try a few different keys too just in case there is an issue with single keys. I can see a couple of capacitors on that schematic and they may have degraded. I doubt it though.
The inverter trick is most likely to maintain the signal, but we have signal->not-signal->n-n-signal=signal->n-signal so the output is indeed inverted at point F3 (is that F3 - I cannot read it) but after R16 that point.

Someone else with a similar mission to mine posted here that the output signal is a 3MHz rectangle that adjusts with key presses but I'm not seeing how it could be produced, and my only analog scope is low frequency and doesn't detect any worthwhile changes in the 2.5 volt signal.

Could be, as a guess a Pulse width modulation scheme is being used to determine which key was pressed.

https://learn.sparkfun.com/tutorials/pu ... modulation

Really for these digital situations you need a Logic Analyser or a Scope with one built in. If you can get along to some kind of Maker group, amateur electronics group of even a University or local College they may let you take your PCB in and test it out - worth a try.

If not, then I shall recommend the Salea Logic here:

https://www.saleae.com/

It's a fairly decent Logic Analyser and I use it the most out of the four that I have.

I hope that you can make some progress with it but let us know how you get on.

Cheers.

[Ixmucane2]

The easiest way to measure a variable inductance is measuring the duration of a transient in which it is involved.
Plausibly, the control lines connect C5/C7 to ground through each key inductance in turn, and the three inverters form some kind of monostable circuit. I'd expect the output out of C3 to meet the clock driving J2, U1 and U2 in some sort of comparator. Do you have an uncensored schematic?

[MintyTheCat]

The easiest way to measure a variable inductance is measuring the duration of a transient in which it is involved.
Plausibly, the control lines connect C5/C7 to ground through each key inductance in turn, and the three inverters form some kind of monostable circuit. I'd expect the output out of C3 to meet the clock driving J2, U1 and U2 in some sort of comparator. Do you have an uncensored schematic?

It is not easy to read but we can calculate the expected rates for the transcients:

Tau_C3R16 = C3*R16 = 2*10^2 * 22*10^-12 = 44*10^-10 := 4.4nS.

He has a scope so he could check that out. Also, testing the reactance.
Voltage at p3 (think it says that but cannot read it easily): X_C3/(X_C3 + R16) * V.

I am interested in the reactance of C3 as we can easily tell another way if the capacitor is healthy - most likely it will be but nice to check it out.

Good point, yes: each of the inverters contributes a lag due to latching. I'd expect the 4.4nS to be a decade or so 'in time' for the signal.