Incorporate pulse/sample parameter

[jgvictores]

We are currently counting on a fixed pulse/sample value (0.001 (?) for velocities and thus 0.001^2 for accelerations):

• yarp-devices/libraries/YarpPlugins/TechnosoftIpos/IVelocityControl2RawImpl.cpp

  Line 25 in 4b2c591

  double val = (encoderPulses / 360.0) * 0.001; //-- if encoderPulses is 4096 (4 * 1024), val = 0,011377778

• yarp-devices/libraries/YarpPlugins/TechnosoftIpos/ICanBusSharerImpl.cpp

  Line 743 in 4b2c591

  double val = (encoderPulses / 360.0) * 0.000001; //-- if encoderPulses is 4096 (4 * 1024), val = 0.00001138

• yarp-devices/libraries/YarpPlugins/TechnosoftIpos/ICanBusSharerImpl.cpp

  Line 761 in 4b2c591

  double val = (encoderPulses / 360.0) * 0.001; //-- if encoderPulses is 4096 (4 * 1024), val = 0.01138

This looks dangerous and undocumented!

Incorporate a pulse/sample parameter, such as pulsePerSample or better name?

[PeterBowman]

Incorporate a pulse/sample parameter, such as pulsePerSample or better name?

samplesPerPulse and you get rid of fractions 😏

[jgvictores]

Related: yarp::dev::IMotorEncoders::setMotorEncoderCountsPerRevolution

[PeterBowman]

Sample loop is 1 ms default, see e.g. docs for object 6081h (profile velocity). Regarding object 207Ah (interpolated position 1st order time), it is stated that time "is given in IU which is by default 0.8ms for steppers and 1ms for the other configurations". We don't care about steppers, I'm just curious whether the sample loop refers to that same time dimension in both objects.

[PeterBowman]

samplesPerPulse and you get rid of fractions smirk

Makes more sense the other way around: pulsesPerSample.

Regarding object 207Ah (interpolated position 1st order time), it is stated that time "is given in IU which is by default 0.8ms for steppers and 1ms for the other configurations".

That is, we want to sample a given amount of pulses per one second. Assuming we already translated a degrees/seconds input into pulses/seconds (by applying the reduction factor and taking into consideration the number of encoder lines and pulses per line):

pulses/second * (samples * second)/pulse

I might be wrong in my interpretation regarding time units; however, the maths are clear and 1 millisecond as stated in the manual translates into 1000 pulses per one (sample*second) in my previous formula. For steppers, that would be 1250. And I like this more than fractions.

[PeterBowman]

Ready at 36b6360.

[PeterBowman]

I might be wrong in my interpretation regarding time units

Yup, even though the result was correct (not accounting for #253), this is not how real dimensions should be translated to internal units. Instead:

• Position: degrees -> pulses
• Velocity: degrees/second -> pulses/sample
• Acceleration: degrees/second^2 -> pulses/sample^2