I wouldn't give that much chance of success. As I understand it, the laser in a DVD player is reading patterns encoded on the disc surface whereas your shaft has only random surface imperfections and tool marks.
if the other sensor marks a peak of red and a little blue, my twist angle is not 6 degrees, but a >value less than 6.
Why are you trying to use an analog analysis (colour) , when a digital analysis (Black/White) will be faster in response. Not just sensor wise but computation wise as well.
What position resolution do you need?
What speed is the shaft rotating?
More accuracy, finer Black/White markings.
OR.
Google;
rotary position sensor hall effect
Tom.. 
What do you do with the return flexing of the shaft? AKA vibration.
Sure seems like there ought to be a better way. Why did you choose this solution? What other methods did you consider? What is the big picture?
There are other ways, but when I researched measuring torque on rotating shafts earlier this year, this seemed to be a very common way of doing it.
I also recall building a simple arduino system for someone a few years ago that was doing something similar. Not for a torque measurement per se, but to detect excessive shaft torsion.
Thanks. Interesting.
I found this article with a thorough explanation of the method.
They used Optek OPB739RWZ sensors with black and white tape. Claim quite good results.
Reading the OPB739RWZ datasheet I saw that in one of the tests they demonstrate exactly what I need.
As in the picture:
See how the signal from the sensor rises, peaks and then falls.
Forming a bell shape.
That way, I could calculate the angle much more accurately.
Thank you I will try to implement something like this.
I will ask for the second time, how do you intend to filter out the torsional vibrations to get the applied torque?
Well, actually, I hadn't thought about it that much.
That would result in sporadic increases in amplitude right?
Could a moving average filter handle the noise caused by vibrations?
Together with this sensor, I also equipped the engine with an accelerator, maybe it will be possible to cross the data and create a filter.
And wouldn't any sort of high resolution torque monitor have to deal with such fluctuations? In that regard, does this method have drawbacks that others don't?
"This paper presents a novel, contactless torque measurement system"
I'm not sure how novel it is when I came up with the idea after thinking about it for half an hour...
No. It would be an increase in vibration after every change in applied torque. The vibration may or may not be in sync with the torque change.
44 RPM, which you have said is the speed of the shaft, is 44/60 revs per second or 1 revolution in 1363 ms.
For a twist in the shaft of 1 degree between the two sensors, there will be timing difference of 3.78ms between corresponding marks. If the speed of the speed of the shaft and load is reasonably constant, you need only one mark on the shaft for each sensor. More marks mean a better resolution during acceleration or deceleration or change in load on the shaft. However, if there are too many marks on the shaft, relative to the maximum twist, you will have difficulty to match up corresponding marks.
The justification for using coloured marks in this application is not very clearly made. A photo diode (at a reasonable price) can have a response (rise) time of a few nano seconds, however, this doesn't discriminate between colours.
On an Arduino, you can measure timing differences down to around 1/16us using the input capture feature.
I guess if it was my project I'd probably replace the coloured bands with black tape with some holes cut in it at regular intervals and shine a light (maybe infrared) into it on the opposite side of the sensor, since the shaft is transparent. The sensor would then be a photo diode or photo transistor.
1 Like
It depends on what the material is used as the "Torque Shaft".
You can digitally filter.
I had to get a make shift torque sensor going when a customer was building a new processing machine, the one he needed was on 6 months back order.
The customer made the torque shaft, worked out the twist deflections vs torque, then we custom made the interim unit.
We made one using 2 hollow shaft non contact, ie magnetic, decoders.
I had a DUE reading both units and doing the differencing calculations, it provided a 4-20mA output which the receiving PLC used its software filters to extract the smoothed reading.
Why 4-20mA, that was what the output spec was for the proper torque sensor.
If the ESP32 had been around then, it would have been even more precise, but the makeshift allowed them to advance their project while awaiting the needed product.
The encoders were like this but in plastic housing;
https://www.turck.com.au/en/product/0000003600024d6b0002003a
They had SSI comms and I was impressed by the stability.
It was the first time I had worked with these 25bitm Hi-Res encoders, fascinating bits of gear.
Tom..
And that is the key. It have extremely limited twisting, not like a solid shaft. Who knows what the OP is suggesting.
Hi, @Matheus_Markies
What is the shaft material?
I hope it is not the plastic flexible tube your prototype is using, your readings will be distorted by the non linearities incurred as you flex and warm up the plastic.
Can you please tell us what the actual application is?
There may be other methods, like measuring the forces produced in the motor mounts.
Thanks.. Tom..
I'm using PVC, the same as in the video.
The motors that do the torsion don't have a high torque, so I opted for something very flexible.
Do you think it is necessary to change the material?
What material could I use instead?
It is a mechanical engineering research project (Developing a non-intrusive torque sensor.).
What video? The picture you included seems to be ABS tubing. PVC tubing, like PVC pipe from the Home Depot will shatter is twisted very much, where ABS likely will not.
I didn't say the authors were not self-aggrandizing. 
1 Like
The article notes that the smallest possible pitch of the zebra tape is limited by the expected torsional displacement.