VL53L4CD / VL53L0X wavelength detection

[TH3]

Dear all,

I'm working on a product where we use 850 nm LEDs with a typical spectrum like shown below:

Since we emit quite some light (10's of Watts in short, ms bursts) with our device in a narrow cone (about the same FOV as the ST sensors), I'm wondering if this could potentially cause interference when reading the distance values. I remember seeing some picture that showed the acceptance window of the incoming sensor but I can't find it any more. 

Thanks for your help in this,

 

Best, Tom

[John E KVAM]

Tom - 

In theory there should be no issue. There is a bandpass filter on the VL53 sensors that only allows 940 +/- a few nm through. And 850 is well away from that. 

But that's only the theory. I had a customer with a cutting laser, and it emitted so much power that the 'blew through' the filter. As it was a continuous beam, it looked like a huge ambient signal. But the point is that you can overwhelm the filter if your light source is bright enough. 

If you have a powerful laser, and its reflecting light back into the VL53, I might suggest you make the protective coverglass out a filter glass - one that filters out anything below 900nm say. That would go a long way toward solving any potential problem. 

- john

[TH3]

Dear John, 

 

I've talked with our LED supplier and after calculating that approx 0.1% of the light emitted from the source still falls within the acceptance window of the sensor I can imagine this has an effect.

One way to solve it would indeed come from the cover glass. Another method, since we used a pulse mode device, might be to run the sensor only when the LEDs are off. We have a 9 ms window without light emission from the LEDs. In the VL53L0X I read that the minimum range measurement period is 8 ms. In theory this sounds like a "perfect fit" but does this sound viable as well from your point of view? Any experience with similar projects perhaps?

Thank you once again for your help John, much appreciated!

 

[John E KVAM]

that's cutting it pretty close. 

The secret is that with more time (and more power of course) you get more accuracy. 

In general, 4x time gives you 2X more accuracy. 

8ms will give you a distance, but the standard deviation will be kind of high. 

If all you need is 'target is there', then fine, but if you need more accuracy, you are going to need 30ms. 

I'd look into a real band-pass filter - or a high-pass filter with a 900nm cutoff. 

You will be happier in the end. 

- john