Measurement error in reading Encoder Pulses

[Harish]

I am using PCI-6601 card to read a Linear Encoder which gives 2000 pulses per mm of distance.The accuracy of the Linear encoder is 2Microns.In my application I am measuring a distance of 7mm.But the card is reading with an error of 30 to 50 Microns error.We had checked up all the factors related to mechanical and Encoder.In fact the same application was earlier working with Advantech Encoder card which has provisons to connect the A,B,A Bar,B Bar inputs from Encoder.There was an accuracy of 0 to 5Microns.But the application was replaced with NI Card because of the continuous problem of the Digital I/O Card used from Advantech.
Mail me the possible causes and also the troubleshooting procedures for correcting the above problem
.

[rgharrison]

Hi,

What I think is happening is there might be some noise on your signal lines. The PCI-6601 counters are capable of detecting a pulse as small as 20ns. If the Advantech card can't detect pulses at that small a resolution, it might be ignoring "noisy" pulses that shouldn't be there. Are you reading more pulses than you should be or less?

There are a couple things I might be able to suggest to clean up your signal. The counters support several different encoder modes. Try using X4 Encoders. An X4 encoder counter increments or decrements on each edge of channels A and B. Whether the counter increments or decrements depends on which channel leads the other. Each cycle results in four increments or decrements. I believe this X4 mode is more noise resi
stant than the standard mode.

Another thing you might want to consider is applying a filter to the input signals (debounce or deglitch filters). Each PFI line coming from the I/O connector can be passed through a simple digital filter. The filter operates off a filter clock and a fast internal sampling clock. It samples the signal on the PFI line on each rising edge of the sampling clock. A change in the signal is propagated only if it maintains its new state for at least the duration between two consecutive rising edges of the filter clock (one filter clock period). The frequency of the filter clock determines whether a transition in the signal may propagate or not.

6601/6602 User Manual
http://digital.ni.com/manuals.nsf/webAdvsearch/DD0DBB6404BC139886256721007CF16D?OpenDocument&vid=niwc&node=132100_US

Hopefully one of these suggestions will do the trick. Have a good day.

Ron