I have been performing a simple drop test on a load cell using a golf ball, recording the impact using a Phantom camera as well as collecting the data from the load cell.

The problem arises when you compare the 2 time measurements for the impact of the golf ball on the load cell plate. The phantom camera gives an impact time of approximately 0.65ms whereas the load cell reading from start to end gives a much greater reading (approximately 0.85ms) as seen in the picture - also the 2 gradients from start to peak and from peak to end differ significantly, which we know is not true during a golf ball impact. enter image description here

  • How do you explain the time readings being different?
  • How I could go about getting a more accurate value from the load cell that matches the phantom camera?

I have also been calculating the impulse of the impact and comparing it to the area under the graph on the load cell readings to check that the force readings are correct, however I cannot determine if they are right - most likely due to the time reading being wrong.

  • \$\begingroup\$ It's far easier to help you if you don't repeat that you need help... less stuff to read that distracts from the problem we're trying to understand. \$\endgroup\$ Feb 28, 2019 at 16:41
  • 4
    \$\begingroup\$ also, as a guy who will be grading 270 exams of EE students in the next couple of days: label your effing plots :) What is what in your plot? \$\endgroup\$ Feb 28, 2019 at 16:51
  • \$\begingroup\$ How are you calculating your gradient? Why are there two of them? \$\endgroup\$
    – TimWescott
    Feb 28, 2019 at 17:39
  • 2
    \$\begingroup\$ It looks like the load cell, it's signal conditioning electronics (if separate), your data acquisition equipment, or all three, are low-pass filtering the load cell signal. This is a profoundly sensible thing to do if the load cell is designed to weigh stuff, because the use case for weighing something is to put it on the load cell and leave it alone while getting a reading. Please post a data sheet of the load cell -- perhaps look to see if it has a response time or a bandwidth specification before doing so. \$\endgroup\$
    – TimWescott
    Feb 28, 2019 at 17:41

1 Answer 1


Any real-world system has limited speed. So, that something that is very fast (optics) measures a short event "shorter" than a slower, mechanical system isn't surprising at all.

If your sensor is linear, then you can model the slowness and "memory" effect by what we literally call an impulse response, being convoluted with the input signal (which, in your case, is the force of the golf ball). There's rich theory of how to model these; in Electrical Engineering programs, we are taught about this in lectures called "signals and systems", or the like.

Now, these properties of real-world systems simply mean that the output of your load cell contains less information than the output of your phantom camera. You can't fully reconstruct what you see on that, mathematically – one system is more band-limited than the other.

What you can do, however, is try to correlate any impact with the best estimate for a true impulse response of the system you can get. That's easy to do, but it requires you to estimate the impulse response – but maybe a bit of correlation applied to the load cell signal and a signal modeled from your camera data suffices as an estimator for the impulse response.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.