vl53l4cd variation in readings

[hazeman666]

Hi,

 We have incorporated a vl53l4cd into our product.

 We wish to measure 0 to 100mm with a resolution of 1mm, and an update rate of at most 1 second.

We have set the vl53l4cd 19mm back from front measurement face (so we need to actually measure 19 to 119mm).

We placed statically on a desk I find, the variation of the readings to be from 33 to 40mm.

It does vary from reading to reading but the trend also varies so I cannot simply average it out.

Is this was what I should expect?

I attached a screenshot of what I find. This is against a light teak table. I would say is appears to better when against something black.

Is there any recommendations for settings for getting the best performance between 19 and 59mm in terms of stability (The inaccuracies are less significant beyond that)?

kind regards

Liam

[John_Kvam]

If you read the data sheet there is a chart. It's a little complex but +/- 3% is really what it claims. But that's over most of the range - at the shorter distances, it's a fixed range.

So, at 34mm this is about what you can expect.

So why isn't it better? 

Photons are fast. I think of the speed of light as 3 nano-seconds per mm. (6 if you count out and back.)

(A nano second is one millionth of one millionth of a second, for those who need a refresher.)

Computers are fast, but not that fast - and especially not ones that cost less than 10 dollars.

So, the sensor has to cheat and use statistics.  And statistics will only get you so far.

Another source of error is temperature. Although the chip adjusts ever 2 degrees C, it varies a little with temp. It has to do with a PLL circuit that turns an ordinary clock into a really fast one.  And the VCSEL (laser) does warm the sensor.

So, what can you do?

1) you can try the VL53L4CX - it uses something called histograms. Its main use is to turn the L4CD into a longer distance sensor. It will go out 6M. But I've found it's a tiny bit better in the short distances as well. 

2) you can wait for the next sensor. I can't tell you when it's going to come out, as I've retired, but I know it's in the works. But no guarantee it will be a lot more accurate in the short distances. 

3) try starting and stopping the sensor every little while. That will re-calibrate the PLL. Not sure it will give you better results, but it might eliminate those high and low points. Might introduce an offset though. 

4) use a matte finish target. A glossy target has a tendency to give a worse result. Too many photons requires the sensor to turn off the SPADs (photon detectors) and that plays with the statistics some. 

The sensor was invented to focus the camera in a cell-phone, and it's really good at that. It's also in a lot of laptops to wake the laptop when you sit down in front of it. 

But as a scientific instrument, it has a ways to go.