That’s true - filtering if it’s simple will introduce significant delay. There are methods to move this lag behind, especially predictive methods. They are especially good if they take into account physics of signal source and additional data that has an impact on the result. But I can tell you for sure - a velocity that Granity scope shows is not calculated well at all. You can see that you can do way better calcs almost without any lag introduced.
Could you comment on the feeling for ten counts TSP1-2? Was it at the edge of what you can feel if you touch the wheel? Large Mige is much more inert, so I would assume that it can’t do such small oscillations as Small one does. From your charts, it looks like positional feedback was much cleaner.
Somehow you have 2 bit noise from the same encoder. Could you set lower sample rate like 1250hz?
EDIT: Bullshit below, ignore.
For 2048 lines encoder you setup your FBR wrong - you’ve selected FBD 256x, but set FBR 0.5M. FBR should be 2048x4x256=2097152. I don’t think it’s working right at all now…
10 was a very, very, light vibration, I tried 5 but I couldn’t feel it , 25 and 50 I could feel pretty well, at 200 of course I could feel it with no issue and the wheel moved slightly.
Mash corrected himself (and I missed the x4 on his post) on that the FBR should be 2048x256 as that is the number of points in a quadrant then the system multiplies the number by 4… I think Granity messed up a bit on this but unless they have a corrective calculation then cannot fix it as it would cause massive issues across their customers with it so they probably should re label it… I am not sure what the difference is with this error between the X64 and X256 SinCos settings since in reality the separation isn’t needed with the FBR error unless it changes some parameters elsewhere in the system.
The thing is of course that naturally a 5000 line unit would have over twice the optical accuracy of a 2048 unit so it would make sense there… A 5000 line unit is essentially like a 40K incremental with an additional SinCOS out for the interpolation.
Unfortunately I don’t have a 1024 hooked up to it… Maybe I will get up to switching out the servo again with the Small Mige I have here with the 1024 line Heidenhain on it.
No idea what any of this means… should I be concerned about anything or just get on with life?
I installed the encoder and other than it supposedly being much smoother/room to grow for titles with support for higher resolution, I am clueless.
The 2048 will be more precise generally, there is the possibility of SinCOS erring a little here or there because it is an analog signal and interpolated but because of that it offers much more points of reference so things like th filtering will be better applied to the servo.
I believe that at these levels, the most important thing is the feeling it gives on the steering wheel, which I find personally fantastic. Even if there were some small positional errors with 2M values, I do not think our feelings would ever come to mind.
I fitted the sincos 2048 today.
The crasstalk noise over 680 is maybe less but not at level I read somewhere (that was nearly inaudible with the new encoder…). Can it vary from one unit to another?
I use only 2% damper (the less I can only to reduce olliciation) and 0 on other filters. (reconstruction at 4).
So maybe this is the worst scenario to get benefits.
At the moment I do not feel differences. Will test more today and tomorrow and will decide which one to use.
I think the sound depends on my mige motor so I can’t do aything to remove it.
Anyway I switched back to the 40k encoder. The simucube new firmware is perfect with this encoder and I can’t feel differences to justify the cost.
This is really strange to me because I read lot of opinions here and there and all were super good so I do not understand.
I made 4 races with 40k, then 3 races with the sincos, the 2 races with the 40k again and if I was blind of the change I would never been able to say which one I was using.
iRacing, 45 min races with Ford GTE.
I do not agree at all. I do not own iracing, but with Rfactor2 Assetto Corsa and PCars2 there is a big difference in terms of precision, control of understeer and oversteer and speed of reaction
Change is massive with Simtronix (Heidenhain) Encoder on Iracing. Much more refined.
Don’t know about Mige Sincos encoder though, but I guess it should feel different too.
