Kit Encoders...Why x2 and x4 have extra noise!?

We are sticking with x1!

Team 599 measured the 2009 Kit E4P encoder rates and found that using x4 mode the error in rate was about 6% at a robot speed of 10 fps. In the x1 mode, this error dropped to around 2%. No measurements were made with x2. In discussing this with Joe Ross (Team 330), he suggested that the phase errors could be the problem. In reviewing the encoder specifications, I think he is correct. We recommend that teams only use the x1 mode when using the Kit or Andy Mark SuperShifter encoders unless they are just counting pulses.

As I interpret the USDigital specifications and how the WPILIB encoder.cpp code works it seems clear that in the x2, and x4 modes the symmetry and quadrature errors would come into play, Eg, for just the A channel there can be 16 to 30% symmetry differences between 1 and 0 state widths which would show up in the x2 mode. In the x4 mode, the variations due to quadrature can cause 10 to 60% alone. These of course are the worst case errors over the environment.

Since the dist/pulse is assumed constant for all pulses then the distances and rates have errors proportional to the phase variations. Our encoders seem to have smaller errors but they are significant relative to the control signals which they are used for. Operating in the x1 with added digital filtering makes them usable.

For those that want to look further see http://usdigital.com/assets/general/168_aedr_datasheet_0.pdf

USDigital.com Data Sheet excerpt for the AEDR module which is referenced for the E4P encoder( not sure which 8300 we have)

Symmetry Error (Ch A) 15 deg typical, 55 deg max
Symmetry Error (Ch B) 25 deg typical, 75 deg max
Quadrature Error between Ch A and Ch B 12 deg typical, 60 deg max.
Definitions:
Symmetry:

A measure of the relationship between (1 state) and (0 state) in electrical degrees, nominally 180 electrical degrees.

Quadrature:

The phase lag or lead between channels A and B in electrical degrees, nominally 90 electrical degrees.

2010 Thread update:

Team 1771 took some near max speed encoder data in support of this thread http://www.chiefdelphi.com/forums/showthread.php?p=953400&posted=1#post953400. I analyzed the data and posted it here:

http://www.scribd.com/doc/29880227/1771-Encoder-Tests. Downloading will give a better format on the pictures and data.

It compares about 1.5 seconds of data for 1x,2x,4x encoder configs.
Summary:
Left Encoder
1x sigma = .30 fps (4.1% of average)
2x sigma = .48 fps (6.5%)
4x sigma = 1.68 fps (18.2%)

Right encoder
4x sigma = .49 (5.2%)