linear response from rheostat?

Status
Not open for further replies.
J

JohannGambolputtyEtc

New member
Hi all. I just purchased a Teensy 3, primarily to resurrect a prehistoric PC racing wheel. I'm pretty new to electronics (I know Ohm's law and... well, let's leave it at that for now ;)).

The problem is that the steering axle uses a two-terminal (linear) variable resistor (rheostat?) (110k, 55k when centered) so I can't just connect it without anything else. Adding a fixed resistor makes the signal not respond linear anymore (I can calculate that much). Keeping the current constant would solve the problem, but I have no experience with current sources (or any other components for that matter, apart from following the nice tutorials). Any suggestions on how to get this work? Do you recommend using a constant current source or are there other simple alternatives worth considering? I also found this instructable to make a current source from an op-amp. Any thoughts on that?

Another option is of course to replace the rheostat with a pot, but it would need to have a 180° rotation.

Thanks!
 
unclear: is the goal to use the variable resistor stand-alone, its leads connected to your new circuit, and the old circuitry goes unused?
Where you say "I can't just connect it without anything else." says you can't disconnect if from existing circuitry (cut wires, cut traces).
If you can isolate it, the problem is simple, perhaps like:
One lead on the variable resistor-call it R1, goes to ground. R1's second lead goes to a new resistor, R2. The other lead of R2 goes to Vcc.
If you then measure the voltage at the junction of R1 and R2, you'll see it vary from 0 to the ratio of R1:R2 times Vcc. (Ohm's law applies, resistors in series are additive).
This voltage can be sent to the A/D in a microprocessor, then input and process (scaled, whatever) in firmware.

There are more complex schemes and there are some guidelines to watch such as the A/D's input impedance, but this isn't hard.
Assuming you can isolate the R1's leads. If not, we have to 'scope R1 and see if it's a DC voltage w.r.t. ground or if it's pulsed,etc. Too messy.
 
Try this...

rheostat.jpg

If you don't have these specific parts, pretty much any diodes and any small signal PNP transistor should work. Adjust the 20K resistor if the range isn't right.
 
Oh, yes, Stevech raises a good point. If the rheostat is still connected to other circuitry, then that really complicates things.
 
Yes, I should have specified, sorry: the variable resistor is completely removed from the old circuitry. I'll take some time to process your answers. Thanks a lot for the quick replies!
 
Well, I simulated it with Qucs (very neat open source simulator) and it looks like Paul's suggestion is exactly what I need!

rheostat diagram.PNG

I do have some scrap parts at home but I'm at work now so I can't test it anyway. But the store I always order from sells exactly those parts, so that's covered. I did consider linearizing it through coding, like stevech suggested, but I'm afraid it would lead to asymmetrical sensitivity (an important factor for a steering wheel). Though I should have tried that first, it may work out fine. But it's better to have a solution closer to the source.

I may play around with R1 (20k) to adjust the range, but since the error is small and the signal is linear, this might be done with coding. We'll see.

Thanks a lot for the suggestions! When it's ready, I'll post some photos and a schematic for the project page if you're interested.
 
Status
Not open for further replies.
Back
Top