DAQ Toolbox Counter Input integer overflow problem

Hello everyone,

I am trying to sample a counter output from a position encoder attached to a servo motor. I am sampling the counter through an National Instruments USB-6259 acquisition unit, using the daq.Session object in MATLAB 2011b/DAQ Toolbox 2011b. I have set up a counter input on a channel in the Session object.

I have observed that when using the counter in MeasurementType= 'Position' mode, the counter correctly tracks the position clockwise, and the Z-Reset works correctly (i.e. the position returns to zero once the maximum counts per revolution is exceeded). However, when the motor is rotated counterclockwise, the counts go to ~4.3billion. I discovered that this number is the intmax for an unsigned 32-bit integer, so I suspect integer overflow when the motor is rotated in one direction past the "zero" position. Is there a way to avoid this overflow, and instead allow negative positions?

The properties of the channel are as follows:

    EncoderType: X4
   ZResetEnable: 1
    ZResetValue: 0
ZResetCondition: BothHigh
      TerminalA: 'PFI8'
      TerminalB: 'PFI10'
      TerminalZ: 'PFI9'
           Name: empty
             ID: 'ctr0'
         Device: [1x1 daq.ni.DeviceInfo]
MeasurementType: 'Position'

Thank you for your assistance.

Regards, Matt

[Walter Roberson]

MathWork's documentation on this topic is certainly sparse! I had to go to NI's site to find out what Z-Indexing was. Regrettably I was not able to figure out what combination of phases corresponds to the condition you are interested in. BothLow maybe. I do not see any mechanism for resetting on two different phase conditions.

Answer 1

[Bart]

I had the same problem, with a NI USB-6216. This is not an overflow problem in Matlab, but a 'feature' of the NI hardware. The values are actually correct, when you subtract 2^32 from the (negative) measured values.

this should do the trick:

>>data(data>2^31) = data(data>2^31)-2^32

This divides the range into 2 parts, the part from 0 to 2^31 (unchanged), and the part from 2^31 to 2^32 (the 'negative' values). By subtracting 2^32 from the second set, these values become negative and can be used as normal.

[Walter Roberson]

Perhaps

typecast(data, 'int32')

[Bart]

Thanks for the hint, but I don't think that would work. The data is already in double format. I think the NI device internally turns these values in uint32. Some values I encountered: pos2 =

   1.0e+09 *

   0.000000024000000
   0.000000013000000
   0.000000014000000
   0.000000008000000
   4.294967283000000
   4.294967254000000
   4.294967226000000
   4.294967212000000

when you plot this, or look at it in short format, it looks like total crap. but when you subtract 2^32 from the last 4 entries, the values make sense: pos3 =

    24
    13
    14
     8
   -13
   -42
   -70
   -84

[David]

Since the data comes in double format, you can just convert it to uint32 format first. I just solved it using the following code

typecast(uint32(data), 'int32')