Best Sensor for Close Readings

[will3]

We currently use the VL53L1x and have struggled with precise measurements to close objects. 

Our application needs precise measurements between 1mm and 400mm.  

Reviewing the available sensors ST has it appears to us that the VL53L4CD is the best possible sensor given our needs.

Would you agree?

Thanks,

Will

[John E KVAM]

this is going to sound odd, but I'm going to suggest you go with the VL53L4CX. I know it has a distance spec of 6M and so it's a lot of overkill, but by experiments, it's actually a touch more accurate at the very close distances.

Hopefully your target moves in and out like a piston of some sort. If it's a different target every time, you are going to have issues. Really reflective targets and exceptionally dull ones give the most variability. 

And about that 1mm - I can't do it. Below about 5 or 6mm the variability in accuracy means you know it's close, but not how close. 

I suggest finding a way to back the sensor into your housing a bit so when the sensor reads 5mm you are to the limit of your range. 

I'd try this. For $56 you can buy the P-nucleo-53L4A2. This demo board will allow you to test and see if the part meets your expectations. Just download the GUI from st.com and start testing.

But you can also use it to test the VL53L4CD. (The L4CX is a superset of the L4CD). Give that a try. If it's accurate enough for you the L4CD is a few cents cheaper. 

But one of the L4's is the way to go. The VCSEL (laser) is a single-stage 18 degree field of view and it's a bit more accurate than is the 27-degree dual-stage VCSEL in the other parts.

- john

[will3]

Hmmmm...

Ok.  Now you have us wondering.  

Since the CX, not the CD, has multi-object detection the question I would have is seems the CD would be good enough for us especially if we can recess the sensor back a few mm's. 

However,  the CX could be valuable for us if we were able to use it to detect multi-objects given our container.  You are spot on that what we have operates like a piston.  It is the same target and it moves front to back.  However,  let's say our container smallest width is 6cm and largest width is 18cm (same length 0mm to 400mm) would the CX even work for us for multiple objects if let's say this same container contained two independent 'pistons'?

Thanks,

Will

[John E KVAM]

The VL53L4CX CAN resolve multiple targets, but there is a trick. The two targets must be separated in depth by about 80cm. It has to do with our 2ns pulse width. 

So for your application, it's not going to do much. 

The L4 has an 18degree FoV. This means that the diameter of the target is 1/3 the distance to that target. 

So if your piston is 6cm wide, you can go 18cm before you have to take into account the side walls of the cylinder. 

At 18CM width you can go 54cm before you start worrying about the side walls.

If your cylinder is very deep (40cm) and very narrow (6cm) you are going to have a calibration step. 

The signal you get back is going to be combination of the piston and side walls.

Very black or very polished side walls will help. Having a reflective piston top will help as well. But you are going to have to do some calibration.

- john

[will3]

John,

All makes sense and good advice. 

One last question (I swear),  in your example would you calibrate against the full 40cm or half/part way?

Thanks as usual,

Will

[John E KVAM]

One calibrates the VL53L4CX differently than for the CB. As the sensor can simply count the photons coming from the coverglass. If you point the sensor where there is no target within 80cm, you can do the crosstalk calibration. 

If you can't get away without a target, flat black as you can make it, and try for at least 60cm.)

Of course, to do the offset calibration I'd use a white paper at 20cm. (anywhere between 10cm and 30cm is fine as well.