Optical chrony's will error out if too close to the muzzle - no reading. Rifle rounds require more distance than pistol rounds. Whether this is due to the muzzle blast or shockwave shadows or whatever I don't know. It is possible that this effect is taking place more subtly at marginal distances, causing the crony to trigger prematurely/erratically depending on individual trajectories over the sensors. You may want to move it another 10 or 15 feet out to see if that helps. This might also explain somewhat the sunny/cloudy differences, as a shockwave shadow would have more effect on a sunny day. The screens do seem to help with this while they survive
I can't test this theory at the moment but it would be interesting if anybody else could have a look. Like my old prof used to say - 'One good test is worth a thousand expert opinions'.
Yes, for reasons too lengthy to explain, I have a very keen interest in all things chronograph. I have tested this problem at length under lab conditions. This weakness of optical chronograph is yet another reason that I am so impressed with the LabRadar.
My assessment revealed multiple sources of error when the screens are too close to the muzzle.
The first is movement of the first screen itself and when they are close coupled on a fixed space bar (like the CED) there can be movement in both screens.
The second is hysteresis in that movement. They don't always settle back the same way after each shot.
The third does appear to be shock wave errors - not from the bullet itself, but rather from the muzzle blast of compressed supersonic gas that leaves the muzzle before the bullet does.
The fourth is very easy to manage, but it always amazes me that so many shooters don't know about it or care. That is returning the rifle to the exact same position between shots.
All of these errors can add up and/or cancel out.
In all cases, the main problem goes away with adequate distance from the muzzle. Before I got my Labradar, I used to put my chronograph screens 25 ft ahead of the muzzle. It is easy to back calculate the muzzle velocity at this distance.
Of course, the Labradar is not prone to these errors. What a terrific advancement!