Add configuration option - extended no. of pins

[sdmeijer]

• I2C
• SoftwareSerial
• ShiftIn registers

[sdmeijer]

Tried 74HC165 ShiftIn-registers (2 daisy chained). No luck so far.
http://dduino.blogspot.nl/2012/04/arduino-74hc165-psio-shift-regsiter.html
http://playground.arduino.cc/Code/ShiftRegSN74HC165N
http://wardyprojects.blogspot.nl/2011/05/74hc165-piso-shift-register-arduino.html

[cefn]

No luck so far? What did you experience.

Presumably they can read digital inputs with regular buttons and 10K pull-ups (that's what they're geared for) but you've had issues with the heavy-duty pull-ups of the MakeyMakey circuit?

Be good to understand a little more of where the experiment got to.

[sdmeijer]

I didn't spend much time for it and didn't read the specifications (I think).
I was testing without the pull-up resistors. Will try again with the resistors attached and see what happens.
I got wrong readings on pins (it look like random values).

[sdmeijer]

Did some proper testing, use the code from this (http://dduino.blogspot.nl/2012/04/arduino-74hc165-psio-shift-regsiter.html) link. With 10k resistors the output is as you expect (0 - 1 - 0 - 1, etc. when holding the pin to GND).
But with 22M resistors it doesn't detect a change of state most of the time.
Couldn't find any info on the max. pull-up resistor value.
According to this I think that ShiftIn-registers aren't an option.
Mayby I2C or SoftwareSerial? (I'm not familiar with those)

[sdmeijer]

After some reading on I2C and SoftwareSerial I'm in doubt if that would be a great addition to the ShrimpKey.
From what I've read you need two Shrimp(Key)s to make it to work.
I think that's too much for a simple device like the ShrimpKey (I like the simplicity of it).

Another posibility is using multiple ShrimpKeys at the same time with different keys. A computer doesn't have a (real) limitation on the number of connected devices (OK, it's 127, but who's weird enough to do this...). The only thing you have to do is use a different USB_CFG_DEVICE_NAME in usbconfig.h, to make sure your computer doesn't understand all the same devices (although it probably will work with the same name).

So, change of plan????
Don't implement this and go for the multi-connection-option?
[Please shoot at this if you're not agreeing with me.]

[cefn]

Sounds quite practical, though quite a bit of extra wiring to have a whole 'nother set of USB cabling, pull-ups, zeners. Adding a (parasitic) ATMEGA328 via Serial or SPI isn't too hairy to do, and I need to prove I2C anyway for another project. Anything timing critical like Serial, SPI (and probably) I2C can also interfere with certain timers though, which we're already stressing with the USB interrupts, so this needs to be thought through.

[sdmeijer]

If you can get it to work with an extra (parasitic) ATMEGA328, please do!
If that fails for some reason, I think we should go for the multi-option.
Can we classify this issue as a nice-to-have? And focus first on the important issues, like issue #4, #9 and #10.