vector sampling rate (in .vec)

Through trial and error, I've found that the formula

     512/AvgInterval

seems to give the sampling rate for my Vectors.  Here, AvgInterval is as read from the 'User Configuration' header (p33 of the system-integrator-manual_jan2011), as a 2-byte little-endian unsigned integer, starting at byte offset 16.

While this formula seems to work in several Vectors, I have *no* idea if it is actually correct.  The problem is that the sys-int-man is a bit light on detail.  All it says is "average interval in seconds" for this item.  So, we have an immediate problem, with the 512 in the numerator.

Please note that I got the formula by mimicking code for aquadoppHR, but in that case, the denominator in the formula was T5, which is at byte offset 12, named "time between burst sequences (counts)".  But that formula gives incorrect sampling rates, in the 4 or so ``.vec`` files that I have tested.

Can anyone advise on whether the formula is OK?   If it's not OK, what should I use, and where is it documented?

Many thanks.  Dan Kelley.

Hi Dan,

Your formula is right.

I understand the confusion. I will update the documentation in the next release of the System integrator manual.

This is just a different way writing the sampling rate. It is done this way due to limited data types and to keep one structure for all our instruments.

Best regards

Jonas Røstad

May I ask a followup?  I am wondering whether the sampling rate will be precise.  To be specific, should I infer times for my data by using time stamps in the "velocity system data" (VSD) headers (Table 5.3 of Sys Int Guide), or by counting time steps based on the sampling rate?  

Background: In working with my own data (in continuous mode), I have been using the "vel sys data" header timestamps, and then interpolating to get times for the "vector velocity data" (VVD).  This works if there is always a VSD before AND after each VVD.  But then I realized, when helping a colleague with data collected in *burst* mode, that the last VVD of a burst sequence does not have a VSD after it, and that means that I could not interpolate.  So, now, I am trying to decide whether I should just calculate times based on the sampling rate, for all cases.

I hope my wording is not too confusing.  My code is all open-source, so you (or anyone) can see it at https://github.com/dankelley/oce in case anyone wants to try it out.

Thanks again.

Hi Dan,

Please have a look at http://www.nortek-as.com/en/knowledge-center/forum/velocimeters/158319581

Please tell me if it doesn't answer the question.

Best regards

Jonas Røstad