I have set up an RC circuit and some logic to discharge/charge and measure the time it takes for the cap to pull a TTL input pin high.

The R in RC is a linear pot used as a variable resistor. From the little I can remember from school, the time I measure should be proportional to R (since everything else in the circuit remains unchanged).

However what actually happens is that with bigger R the difference in time gets smaller.

I'd like to understand first if I am correct in thinking that t \(\propto\) R, and if so, why that does not happen on my desk.

Here is my setup:

 

What is the LOGIC_IN signal?

How do you measure R, between what two points?

 

100μF is a bit large. You don't want to use an electrolytic.
You do not need any extra components besides R and C. You can connect this to a single I/O port of an MCU. I have done this many times using a 100nF capacitor.

 

logic_in is connected to a PC parallel port, as is discharge.

The software sets discharge high for a while, then sets it low and then continually polls logic_in until it reads high. (keeping track of time elapsed).

R2 has one end floating, one end to R1 and the middle connected to the cap.

 

I used this technique years ago for a crude ADC.
You need a stable capacitor. 1% Silvered Mica.
Although the spec on a CMOS input is 30-70% for a trigger, they are 50%. I haven't come across any HC gates that aren't.
Use a look up table.
Use a FET to discharge C, then open the FET to allow charging from a stable supply.
Remember
V(t)=Vcc(1-e^t/RC)
V(t)=Vcc/2 - CMOS trip point


Take Logs of both sides to get an expression that involves t in terms of Vcc, R & C. The rest is schoolboy maffs.

 

100μF is a bit large. You don't want to use an electrolytic.
You do not need any extra components besides R and C. You can connect this to a single I/O port of an MCU. I have done this many times using a 100nF capacitor.

I will try that, but I am still curious as to why my thing behaves nonlinearly. Why should I avoid elecrolytics?

 

I can't answer your non-linearity without seeing the whole set up.

Electrolytics have a residual polarization effect and hence the capacitance is not well defined. You want to use a non-electrolytic capacitor. Hence keep the capacitance below 1μF for repeatable results.