How to reduce the sampling duration of VL53L3CX? As it is obvious longer than VL53L1x.

[DLI.1]

Currently i have a project based on VL53L3CX, and we are planning to use it for short ranging mode, long ranging mode and proximity detection mode. But some confusions happened and trapped me.

1. In first stage, I am trying to use standard driver (sampling twice) to realize both the short ranging (<1.5m) and long ranging (1.5m-3m) mode, and i found:

a) In short distance ranging, everything seems ok.

b) In long distance ranging, output value are not valid when distance more than 2m.

2.Based on b) i am suspecting that sampling twice cannot ouput the valid value in long distance, so i did some tests including adjusting time budget and sampling times, and roughly get the conclusion: for example when sampling 4 times and Time budget=56ms, it can only measure up to 4m。

But it has a problem, that is the current consumption is too high and this project is using coin cell battery. So i am searching for other solution.

3.I also used ULD driver of L1x for L3 (seems no official ULD driver for L3？), still can only measure for short distance rather than long distance(>2m)。

So my questions are:

1. How can i get the valid measurements both in short distance and long distance based on sampling only twice? As it is related to product battery lifespan.
2. How to reduce the whole sampling durtion of VL53L3CX？(For example, TB=8ms, when sampling twice the whole sampling duration is about 68ms), in VL53L1x it seems much shorter than this with same time budget.
3. ​What is the maximum detection distance of VL53L3CX i n proximity detection mode with using ULP driver? (specs didn't clarify but API docs says it is 0.4m)

Here i have attached one picture to specify what i mean for the "sampling twice" and "Sampling durtion".

[John E KVAM]

The basic issue is the number of photons returned from the target is the limiting factor.

When the target is close, you have no problem as there are plenty of photons.

But as the target gets farther away, it takes a longer integration time to collect enough photons to get a measurement.

Sometimes you can lower the signal rate. It will return an answer - even when the signal is so low as to be somewhat inaccurate, but if you only want presence and not accuracy.

Why sample twice?

There is a condition called aliasing or wrap-around. When the photons from pulse N come back after pulse N+1 goes out, the sensor incorrectly assumes the returning photons are from the most recent pulse.

So if we send out pulses assuming enough time for the light to go out 4 meters, and the object is at 4.2 meters, we can get could fooled and return 0.2 as a distance.

So we use two samples at two different pulse intervals. If the numbers match we know we have it right. And if the number are off by a known amount of time, we can work out distance to 6Meters.

But longer distances requires a VERY reflective target - like a projector screen.

To increase the speed of return data, we can return a result after every cycle, but the minimum is 2 - because we want to do that check.

The ULP saves power by using a very short integration time. But with that short time you give up accuracy. If you only want to know if someone is there, then it's a good way to go. However with a short integration time, you also give up distance - unless the target is reflective.

Skin is 50% relective - and most synthetic fabrics are as well, but soft cotton fleece is only about 17% reflective. And so your max distance depends on what your subject is wearing.

In very battery constrained projects, some people use passive IR - which is very power efficient, then use the VL53 sensor to gain accuracy when needed. But besides the obvious cost issues is that the PIR coverglass must be made out of something not compatiable with the ToF.

• john