VL53L5CX Operation Theory

[David01]

Hi,

I was wondering if there was additionaly theory avaliable about the operation of the VL53L5CX TOF sensor? I've already read the datasheets and applications notes. I'm particulalry interested in how the SPAD array identifies the transmitted light pulse? Does this work off the prinicple of lambertian relfection? Also how does the sensor account for the the offset between the FoV of the transmitter and receiver? Would you be able to supply sources where I could read into this theory or is this a company secret?

@John E KVAM, @Anne BIGOT, Just mentioned these two names as they seem the most active in response to VL53L5CX quieres.

 

Kind regards,

David 

 

[John E KVAM]

The VL53L5 (and the VL53L7 and VL53L8) use histograms to get the range. Each bin of the histogram represents all the photons that were collect during each clock cycle after the laser turns on. All 3 of these devices have a clock such that each bin is 3.75cm distance. So of you have significant photon counts in the 10th bin, you will have a target 10 x 3.75cm away. (To figure out where in the bit the target lies more accurately one uses the ratio between the first bin and the next bin. 

If you really want a treat, buy the VL53L8CH. With this chip we enable the ability of the sensor to output the bin data (as well as the distance and signal strength) and you can see exactly what the sensor sees. The VL53L8CH is not that much more than the VL53L5 in cost either. It's a good deal.

The 2mm between the transmit and receive is too small to even care about. For any reasonable distance just consider them in the same place. 

Exactly how we do this is a company secret, but several universities give PhD's in this kind of stuff.

I'd suggest buying the P-nucleo-53L8A1 and downloading the VL53L8CH Graphical User Interface. 

You can log the output data and evaluate it against your target scene. Maybe you can come up with a better algo. 

Generally people put this histogram data into AI engines and see if the AI can 'see' anything in the data.

- john

[David01]

Just to clarify the complete 8x8 SPAD array is represented by the histogram or is the histogram just for one SPAD? What is the underlying principle that maps a distance to a SPAD... i.e. How is the distance in zone 0 determined from the histogram if the histogram is for the 8x8 SPAD array?

[John E KVAM]

The 8x8 grid of results is created from a much larger SPAD array. 

But you get 64 DIFFERENT histograms. It's a lot of data. 

We have a new product - the VL53L8CH (H stands for Histograms) where we expose the histograms to the user. Mostly this is for A/I applications, but lots of researchers us it as well.

The historgrams on the L5 work the same way, but we do not have a way of outputting them. 

So read up on the VL53L8CH (from the ST web site) and you will have your theory.

And you might consider the VL53L8CX as a replacement for your L8. It really is better and costs about the same. 

Different hardware though. 

[David01]

Hi @John E KVAM ,

Does the VL53L8CX have better ranging accuracy?

Just in relation to the bin width and distance calculation... can you provide a simple calculation to show me how this process works? For example assuming the object is 100mm away this would indicate a large photon count in bin 3. The range of distances would be 75-112.5mm. How does one select a distance in this range using the other bins? Also how does the clock frequency of the device relate to the bin width?