I am building a module that takes an analog CV and converts it to a midi message. I am connecting the CV via a 1K resistor to an AIN input with a diode clamp to protect against over/under voltage.
I am using Banana cables so when no signal is connected the AIN inputs are floating and will produce randon results. My solution is to connect a 1M resistor to GND on each input. Is this the correct way of doing this?
My other question is with regards to the schematic here… http://www.ucapps.de/midibox_ext/lfo/lfo_schematic.pdf
The diode clamp is using 1n4148 diodes so the PIC pin will get -0.7 to 5.7. But the datasheet quotes -0.3 to 5.3 as being safe. Is it ok to use 1n4148 with a PIC or would it be better to use a BAT48 or similar that would limit the voltage closer to the datasheet range?
First thoughts:
The 1M resistor will create a voltage divider, I guess you chose 1M to minimise the effect of this.
Definitely go for Schottky-type diodes for clamping. You’re correct in that the reverse voltage will kill your IO otherwise.
You might want to check this out:
http://discourse.midibox.org/t/topic/18525
It’s designed for an STM Core but could be easily adapted for PICs. The concept is to use a rail-to-rail opamp as a second inverting amp stage instead of diodes. There’s no chance that a higher voltage gets through unless it manages to raise/lower the power rails (unlikely in a modular setup). The design has input attenuators and switches/jumpers for selecting the voltage range (unipolar 0-5, 0-10 V or bipolar -5 to +5).
I might have some spare boards available, currently I haven’t tested them due to other commitments. Let me know if you want some and I’ll look into it for you.