2

I'm designing electronics for a time-of-flight sensor that will have an operating distance of 3cm - 1m. I want a distance accuracy of 5cm, like this model:

https://www.pololu.com/file/0J1187/vl53l0x.pdf

To obtain an accuracy of 5cm, we would need a clock speed of $$t = 0.05m /3\times 10^8 ms^{-1} = 1.6\times 10^{-10}s$$ or 6GHz. However, most op amps i've seen have ~50 nanosecond settling times (20MHz). Do they use better chips, or is there something else I'm missing?

acenturyanabit
  • 123
  • 5
  • 1
    Maybe post a link to a TOF with 1mm resolution. It seems to me that 33ps time resolution would be pretty challenging for any low cost type of item. Maybe acoustic TOF would be more reasonable. Then 1mm resolution might be more reasonable. – user57037 Apr 20 '21 at 05:35
  • 1
    You mean [something like this](https://www.st.com/resource/datasheet/vl53l0x.pdf)? – jonk Apr 20 '21 at 05:35
  • Just get a TOF IC, if you roll your own, the system will be very expensive – Voltage Spike Apr 20 '21 at 05:40
  • 1
    Also, Thornkey, your time is wrong. 1 mm divided by the speed of light is about 3.3 ps. So you are off by a factor of 10, already. Your figure is for 1 cm, not 1 mm. And that's just outgoing. The return flight will double the total time in flight. – jonk Apr 20 '21 at 05:47
  • The TOF sensors look like they are pretty complex internally. They may be using chirps rather than continuous bursts of light. Chirps can be demodulated and provide range estimation and resolution based on chirp bandwidth. This is just a guess/speculation. But if that is how they work, they would not need ps timing resolution. – user57037 Apr 20 '21 at 16:39
  • 1
    @mkeith He is asking about accuracy rather than resolution. Resolution (chirped or otherwise) is always limited by optical bandwidth, but accuracy is not. In this case, the TOF sensor he has picked uses SPADs, so the illumination will be a simple pulse. – user1850479 Apr 20 '21 at 16:47
  • 1
    @user1850479 I didn't notice the edits including the link to the vl5310x. OP was originally talking about both accuracy and resolution, though. Where did the OP reveal that "he" is the correct pronoun to use? Modulation changes the optical bandwidth. So a modulated laser, depending on how it is modulated, can have resolution in range. At least in theory. In a sense, timing pulse arrival is just a special case of modulation, and the range will still be limited by the bandwidth of the pulse (which is not zero). – user57037 Apr 20 '21 at 18:10
  • @mkeith 1mm resolution implies an optical bandwidth close to 1 THz, so while modulation does change bandwidth, this has no effect on actual resolution since you cannot modulate optical sources with bandwidths on the order of THz. Pulses do have bandwidth, but they are usually not modulated. Regardless, resolution is not relevant here, so no point in discussing it. – user1850479 Apr 20 '21 at 18:28

2 Answers2

3

You're assuming that timing accuracy is the inverse of bandwidth, but this is not so. How accurately you can measure something is not directly related to how fast it changes. For example, imagine you had a incredibly low noise, exactly reproducible 1 Hz bandwidth sensor. You could measure the shape of that 1 Hz signal extremely accurately and localize when it happened very accurately, probably down to at least microseconds. In that case you would have accuracy thousands of times better than the inverse of bandwidth.

The SPADs in that TOF sensor work on this principle. They are made lithographically to have very, very reproducible single photon response with a very clean rise even though their 3dB bandwidth is very limited (10s of MHz). The hardware then carefully localizes the exact edge of that rise time and calculates exactly when the photon arrived.

user1850479
  • 14,842
  • 1
  • 21
  • 43
  • Thanks! I'm rather interested in how the hardware does edge detection without a 1THz ADC or anything that can switch faster than 20MHz. Could you point me to a book or paper? Thanks! – acenturyanabit Apr 20 '21 at 22:19
  • 1
    @Thornkey 1 THz is the optical bandwidth in that example. It is the bandwidth needed for a NIR pulse to have that axial length and thus resolution. The detector type you linked doesn't resolve anything (it only measures distance) so that number is not important. For high resolution optical ranging, typically interferometry is used to avoid the need for high detector bandwidth. Interferometry can be used to measure the arrival times of photons down to femtosecond accuracy. – user1850479 Apr 21 '21 at 18:20
1

TOF sensors make all the calculations for you.

Your microcontroller has to program the TOF sensor to execute a certain task, for example to measure the distance of the closest object.

After completing the task and after making all the calculations needed, the TOF sensor will fire an interrupt request to the microcontroller.

The microcontroller will read the result data out of the TOF issuing some commands over the I2C or SPI interfaces.

Examples:

STMicroelectronics VL53L1X (up to 4 meters) , VL53L0 (up to 2 meters)

They emit a short pulse of invisible light laser at 940 nm (Class 1)

They detect by SPAD (single photon avalanche diode) the reflected light. They have an array of integrated lens.

Your end-user product will have to declare in the manual that it has a Class 1 laser inside.

Enrico Migliore
  • 4,161
  • 1
  • 9
  • 12