a birthday present...

[ssgross]

I started to dig. Here is a portion of the raw data for a shot. SNR is signal to noise ratio. Notice anything curious? There is an INCREASE of 10fps in under a yard. The data shows readings for this particular shot at 0yds (which is computed per the docs), then nothing until 13ish yds, where is where it first picks up the round, then at 1ish yd intervals out to 50 something yards. Every now and again there is an increase in velocity instead of a decrease - usually less than a couple FPS, with the exception of the anomaly below.

Time. Vel. Dist SNR
0.033021 2628.51 29.2 17.9
0.034021 2620.81 30.08 17.68
0.035021 2630.47 30.95 13.12
0.036021 2627.19 31.83 10.82

Of course they don't publish whatever statistical method they are using. My guess is, given how primitive the features are (e.g. no way to format your sd card on the unit, or move data off its internal storage, the mobile app issues, etc), they likely didn't put much thought into identifying and eliminating outliers. The big breakthrough for them was likely in getting the sensors affordable enough - and they have skimped on all the other aspects of development. I'm not curious enough to try decompiling their firmware...yet. It's easy enough to play with the data first and see if the results can be made better.

[ssgross]

I think their method takes a clustering of consecutive reading somewhere in the middle of when it first registers the round (11 yds for me) and 25 yds, with the data points used in the cluster identified by high SNR, then computes the linear regression of just the cluster and uses that to predict muzzle velocity. Doing so got me with 0.1 fps of their reported MV. The reported velocities at other distances seem to be from a linear regression of a similar cluster of readings around whatever distance you have in the settings. One expects the noise to increase with distance, but on average I think it would be "locally" consistent, depending on where the round is its trajectory relative to the radar cone. in 50 yds worth of readings at every millisecond from shots taken at 100 yds, the SNR starts high, gradually fades, spikes back up briefly, then crashes. I bet with more careful alignment next time, I'll get more data on more shots. without touching the unit, I seemed to get data out to 100 yds only every 4 or 5 shots.

[usabaker]

the LabRadar is about the only thing I see at the range anymore. I guess me being "vintage" at this point my Caldwell chronograph is the only one I see at the range now. Happy birthday! mine is coming on the 9th so if your wife and kids want to send me a LabRadar too I would not mind at all.

[ssgross]

If I told you I put it in the mail tonight and it's on the way to your porch, you would likely need to wait a few years to call me a liar given the current situation of the USPS. I've had several packages, Christmas and birthdays for the kids, all CONUS, take over 2 months. Santa had to get creative.
I'm having a lot of fun with this. I think I can even save ammo by bringing my laptop to the range and playing with data in between shots instead of just blazing away group after group.

[ssgross]

In the little spare time I have had to non-work-intellectual things....I've discovered how I think the Labradar does it's calculation from the raw data. I think it's a weighted least squares regression, with the weights supplied by the signal to noise ratio. I took a stab at doing a weighted least squares in excel...ugly ugly. It's probably easiest to just parse the csv in python - the object being a readable table of calculated velocities at more than just the 5 set points in the preferences, and then write an open-source ballistics calculator to go with it. I'll get to it some day.

I surprised myself last night measuring strings with under 5fps std. dev - not in a milsurp though!...A few years back I built a poor-mans f-class rifle...a brand new factory Remington 700 left hand with heavy varmint barrel something, ditch the cheap plastic stock it came with for a Black Friday special XLR chassis. I then splurged the next year for a jewel trigger and Nikon glass. I have it set for 0.5lb trigger right now as I am only shooting off the bench at my indoor range. It shoots 1/4" groups at 100 yds with favorite handholds, and the Labradar is confirming very tight velocity groups too.