Measurement Delay

There is a delay in the video stream. When machine stops, it will take some time before the camera gets an image, codes it for USB, transfers it to PC, the PC buffers it and the video section decodes it. Depending on the cameras, used resolution and the performance of the PC, this delay can be significant.

Therefore, the video measurement system can wait for a set number of frames before a measurement. In other words, when the system takes the machine in position for a vision based measurement, the video stream might lag the machine movement and stabilize only after some frames.

The arrow buttons increase/decrease the number of frames to wait.

The “measure” button measures the required delay. It takes the machine a bit off the homing mark, then to the homing mark and measures the mark position. This is repeated 15 times with different delays. Finally, the results are reported on the log window. (Please note, that the software prioritizes measurements over visual stream; what you see on screen during the process does not reflect what is going on in the measurement system.)

Here are the results form my computer at 2084 x 1536  resolution. The camera frame rate is 25 fps, up looking camera is also active.

Results:
Del| X | Y
0 | 0.751 | 0.843
1 | 0.005 | -0.015
2 | 0.000 | -0.005
3 | 0.005 | 0.005
4 | 0.000 | -0.010
5 | 0.000 | -0.005
6 | 0.000 | 0.000
7 | 0.000 | -0.010
8 | 0.000 | -0.010
9 | 0.000 | -0.005
10 | 0.000 | 0.000
11 | 0.000 | -0.010
12 | 0.000 | -0.005
13 | 0.000 | 0.000
14 | 0.000 | -0.010

The last result, which is taken with biggest delay, is our reference. We’ll see that the first result is clearly wrong. On following measurements, we see small errors, one or half pixel worth. These are caused either video noise or machine repeatability. Just to be safe, I settled with two frame delay.

In practice, the machine doesn’t spend very much time in measuring when it is placing components; one or two frames of extra delay doesn’t much increase job run time, but gives some margin. Therefore, set the delay one or two frames more than you really need.

However, the same system at resolution 3840 x 2160 gives very different results. The computer is not powerful enough to keep up with two high resolution video streams. The frame rate drops to 12fps and the lag is noticeable.

Results:
Del| X | Y
0 | --- | ---
1 | --- | ---
2 | --- | ---
3 | --- | ---
4 | --- | ---
5 | --- | ---
6 | 0.666 | 0.800
7 | 0.772 | 0.871
8 | 0.287 | 0.325
9 | -0.003 | -0.007
10 | -0.003 | -0.003
11 | 0.000 | -0.007
12 | 0.000 | 0.000
13 | -0.003 | -0.003
14 | -0.003 | 0.003

First six results indicate a failed measurement, three following ones are clearly wrong. With this resolution and my (outdated) computer, I would think a delay of at least ten frames is needed.


Back to Setup Video Processing page

Software reference top page