Date: Thu, 27 May 2004 17:00:39 -0600
From: Milan Pipersky

To: rilling
Subject: Re: beam timing question

Bob,

I have carefully reviewed the mmw drx software w.r.t the question of 
timestamping beams. The timestamp is computed for THE END OF THE BEAM. A 
brief outline follows:

Beamnumber is computed from the pulsenumber:

beamnumber = pulsenumber / hits per beam

pulsenumber = epoch second * (hits / second)

epoch second is the GPS time at which data acquisition commences -- the 
current second plus one; necessary delays to begin at precisely the 
first pulse of the first whole beam in the next second are applied in 
hardware and software and have been thoroughly tested.

The mmw drx main application starts acquisition as described above and 
literally waits for a beam of data to collect. After each pulse, an 
interrupt occurs and the pulsenumber is incremented. When a full beam is 
collected, the system beamnumber is incremented, the dwell is computed 
(pulsepair calculations), displayed, and a piraqx data packet is 
assembled. THE INCREMENTED BEAMNUMBER is used to compute an epoch 
second/nanosecond pair that is written into the packet.

To "discipline" this method, on each GPS second transition (PPS, "pulse 
per second") the "hitcount", that is, the #hits accumulated toward a 
full dwell, is compared with that expected by the epoch second and beam 
parameters (hits/dwell, prf), and a correction is applied if necessary. 
For instance, with prf = 1000 and hits = 30, the hitcount at PPS should 
neatly rotate through 20, 10, 0, 20, 10, 0, ... This has also been 
tested and it works flawlessly. In the future we could store the 
corrections in the header, for diagnostic purposes. For now, the beams 
should sync within 1 hit.

Good Luck!
Milan


Bob Rilling wrote:

> Milan,
>
> The Ka radar provides a timestamp on its beams.  Does the timestamp 
> mark the start, middle, or end of the beam?
>