Building MIDI-Keyboard scanner - circuit layout and lag?

[FuZZ1L0G1C]

Hi. Planning to construct a keyboard scanner / controller

(1): The decoupling caps all appear to be in parallel (pin 16 Vs to Gnd), so does each IC see 2uF (20 x 0.1uF)?

(2): On 16 of the IC's it shows Vs connected to pin 16 and pin 6.
Where should the 0.1uF 16V mini-disc caps be physically close-coupled to at IC body?

(3): As the E510 can scan up to 128 inputs, can I run seperate controllers, about 1-2m apart?
My plan is to have a "Melody" controller, a "Chord/Bass" controller, and one for "percussion".
My sequencer software can allocate any of 128 keys to any channel / voice / patch.

My question is: will running a ribbon from a controller-section 1-2m away affect the midi-sync, or should everyting be wired compactly?
Calculating from 8MHz highest (external) clock frequency, I get wavelengths of:
FWL=37.5m; HWL=18.75m; QWL=9.375m

An old MAPLIN UK catalogue selling the E510 Scanner IC, (touch-sensitive MIDI), shows a simplified schematic of a generic (no brand) keyboard scanner.
This is what got me going on this planned MIDI project.

About the IC (E510) - data extracted from book:
It scans keys then sends MIDI output to MIDI 5-pin DIN socket SKT1.
The clock circuitry shown with crystal produces a 4MHz CLK freq at Clock pin 14.
4MHz CLK produces an output rate of 31250 baud on IC1 pin 9 and SKT1 pin 5 (serial output).
4MHz also allows a minimum debounce-resolution of 256uS.

If using an external CLK, say from your synth, you can use from 4MHz to 8MHz clock-rate.
When Ext.CLK is used, leave pin 14 (Clock) open (disconnect) and feed Ext.CLK to pin 15 only.
When Ext.CLK rate is increasedto 8MHz, serial output rate is doubled to 62500 baud.
8MHz also allows a minimum debounce-resolution of 128uS.

Formulas are:
res(S)=(2^10)/X or 1024/CLK
baud(Hz)=X/(2^7) or CLK/128

The E510 IC is apparently discontinued (WWW/Google research), but hopefully I can still find one here (RSA) or get a listed equivalent.

Microscopic circuit from components catalogue was abbreviated and re-drawn using Circuit Wizard.
Original Hi-res scan too large for posting (13Mb) & book is probably copyrighted anyway.

Thanks in advance, Clive.

 

[Harald Kapp]

The decoupling caps all appear to be in parallel (pin 16 Vs to Gnd), so does each IC see 2uF (20 x 0.1uF)?

For DC and low frequencies: yes.
For high frequencies: no. The board traces act as inductors which separate the decoupling capacitors. For this reason these capacitors need to be placed as close as possible to the IC's pins. See

Where should the 0.1uF 16V mini-disc caps be physically close-coupled to at IC body?

As close as possible, whichever pins are used for GND and Vcc. Maybe use the underside of the PCB to keep wire lenght as short as possible.

As the E510 can scan up to 128 inputs, can I run seperate controllers, about 1-2m apart?

You can try, but you risk interference from teh long cables in between. Use well shielded cable at the least in this setup. Also each controller should have its own electrolytic buffer capacitor in addition to the decoupling capacitors. E.g. use one 100 µF / 16 V electrolytic capacitor from Vcc to GND for each separate unit.

My question is: will running a ribbon from a controller-section 1-2m away affect the midi-sync, or should everyting be wired compactly?

Ribbon wire is not good for this application. Use shielded wire or at least twisted ribbon wire. For mechanical stability (in addition to the electrical parameters) you wil need a cable with a good cable jacket. WOuld shielded twisted pair as used in ethernet patch cables have enough wires for your application?

 

[FuZZ1L0G1C]

Thanks for the info, Harald.
A company near us specializes in data-cables, including multi-core TP.
I'll probably have to use 36-pair so I'll pay them a visit.
Regards, Clive.