[sdiy] * Piano roll reader project.. using Photo-transistors to read holes.. Sensitivity wise... ??

brianw brianw at audiobanshee.com
Sat Dec 3 00:11:45 CET 2022

I could not find an answer in this thread, or a link to the current schematic, so....

Are the IR LED and sensor circuits active all the time?

... or are they switched in a MIDI synth-style row and column matrix?

I'm asking because there seems to be concern about switching noise in the wiring, so I'm curious how much switching is going on.


On Dec 2, 2022, at 12:10 PM, Jean-Pierre Desrochers <jpdesroc at oricom.ca> wrote:
> Yes.. You are right.. I think I could use 22AWG (2 inches long) twisted pairs for each sensors. 88 pairs overall.
> All grounded at one place on the main pcboard.
> And what about the pullup resistors at each PT's collectors ?
> Again these resistors will be from 8 x ???k SIP networks.
> 11 x 8resistors networks overall on the main PCB, close to the
> 11 x 74HC165 high impedance inputs.
> So far 100k pullups gives the gain I'd like to get on PT's
> but at the expense of noise picking..
> What is your thought about the pullups I should use
> in the actual design condition ? 47k, 68k, 100k ??
> ***********************************
>> -----Message d'origine-----
>> De : Harry Bissell <hbissell at wowway.com>
>> Envoyé : 2 décembre 2022 14:43
>> You might use pieces of 28ga ribbon cable and use two wires, one for signal
>> and one for ground for each sensor. (Yeah x88 but I don’t know
>> mechanically).  Keep all grounds separate and join all at the destination. This
>> way the conductors are guaranteed to lie close to each other and return
>> current flows adjacent to the send. That would minimize crosstalk. Twisted
>> pair would be better. There should be no EMC concerns as the slew rates will
>> be low and not much current flowing, and not often.
>> Harry
>> Sent from my iPhone
>> On Dec 2, 2022, at 3:14 AM, Roman Sowa <modular at go2.pl> wrote:
>>> Frankly I can't see any obvious reason to worry about crosstalk. This is only
>>> 2 inches of wire carying DC signal. Come on.
>>> You may want to add 10n caps across the optosensors to get rid of HF noise
>>> of any origin, that might cause false triggers.
>>> Do you have clean 0 to 5V levels on the optos? I dare to think it's not that
>>> beautiful. With input voltage varying barely above HC165 threshold, you may
>>> force poor HC165 inputs into linear region, making it demand more supply
>>> current. Maybe you could tweak power and GND ails of the sensors separate
>>> from logic supply to find a region where voltage changes almost rail to rail
>>> from HC165 point of view. And BTW, this chip works down to 2V so there's
>>> another way to tweak.
>>> Roman
>>>> *Von:*Synth-diy *Im Auftrag von *Jean-Pierre DesrochersSo far my  Piano roll reader project goes well..
>>>> I’m now doing some experiments using T1 phototransistors (940nm IR
>>>> wavelenght) that are each inserted in the small one inch long copper
>>>> pipes that prolong the rear of each square holes of the main brass reading bar.
>>>> I noticed that rising the PT’s collector resistor will in fact also
>>>> rise the PT gain thus making it more sensitive. That’s good.. but
>>>> there is a problem, rising these resistors too much will create
>>>> possible crosstalk between the IR PT’s wirings (around 88 x 2 inches long wiring pairs) to the pcb that holds the readers IC’s (74HC165).
>>>> There will be 11 x 74HC165 parallel to SPI interfaces mounted on the
>>>> main board To read the 88 holes.. See one of the 11 SPI reader circuit here.
>>>> One thing that encourages me is that the rate of holes changing on
>>>> the scrolling paper Will never be higher than the humain hands that played the original song.. so not very fast for crosstalk..
>>>> I tried 10k’s but this needed a too strong IR beam facing the PT’s..
>>>> 100k’s gave a much more sensitive result..
>>>> I could select a final value from 47k to 100k for all these resistors..
>>>> What do you think about  the possible crosstalk at that read speed, and the value to use ?
>>>> Thanks for your time !
>>>> JP

More information about the Synth-diy mailing list