Open Source LiDAR - Unruly

Back to project overview

We've been waiting a long time for someone to post the plans of a working LiDAR. Is it really going to happen?


Lincoln Wang wrote 04/16/2019 at 16:10 point


I notice that the bandwidth of the output front end of the APD is 4.4 MHz, (capacitance of APD is 2pF, Rf of MAX3658 is 18kOhm). But the pulse of laser is 15ns, I think a bandwidth of at least 23 MHz is required. Did you sacrifice the pulse response for more gain? 

  Are you sure? yes | no

Laser Developer wrote 04/17/2019 at 05:18 point

Interesting question Lincoln. The gain of a trans-impedance amplifier is modeled using a resistor (18kOhm) but the input does not actually look like a resistor to the APD. Instead the input impedance of the MAX3658 is much lower than 18k, it is 75 Ohm.  This makes the cutoff frequency about 480MHz. 

  Are you sure? yes | no

Lincoln Wang wrote 04/17/2019 at 14:13 point

Thank you. Your reply explains my doubts.  And on the page four on the datasheet of MAX3658, the plot of BANDWIDTH vs. INPUT CAPACITANCE also explains this point. But to be clear, I found the input resistance of MAX3658 is actually 400 Ohm, in which case bandwidth is still sufficiently wide. 

  Are you sure? yes | no

John Stockton wrote 02/11/2019 at 13:40 point

The MAX3658 is an impressive part with very high gain (18.5k Ohms).  It has a DC restoration circuit that looks like it is for dark current compensation that would normally be from a PIN diode.  Does the APD work well with this circuit?  I also assume you can adjust the threshold of the APD by changing the bias through a PWM.  Are you doing single returns or averaging a number of them together?  Great project!

  Are you sure? yes | no

Laser Developer wrote 02/11/2019 at 14:08 point

Thanks for the questions John.

Maxim has a  family of TIAs that originally came from the fiber optic world - MAX3665, MAX3658 and MAX40658. Each has pro's and con's when used in LiDAR with an APD. It is generally better to run the APD just below the breakdown point so that there isn't too much DC signal entering the TIA. It turns out that this is a good operating point for SNR anyway and can be controlled by the bias PWM as you suggest.

Under most operating conditions a single laser shot is sufficient to give a good result but it is very interesting to see what happens when you start analyzing multiple shots. Not only can you improve the timing resolution but you can also apply statistical methods to a much noisier signal to improve the SNR in ways that seem impossible. For example, analyzing 100 shots gives a 10x improvement in SNR, allowing Unruly to measure signals that are well below the noise floor.

The trick here is knowing that firing the laser faster does not affect the eye safety rating nor does it significantly increase the current consumption. Faster firing = more data. More data = better range.

  Are you sure? yes | no

radiusmike wrote 02/08/2019 at 05:25 point

Great project.  I could really use one of these.  Am using LIDAR-lite (gotten too expensive), a homemade solution using PIN diode and caps charged to light level (proportional-inverse to distance), and cheap phase-shift laser rangefinders from China ($30) whcih has good accuracy but slow response.  For $50, this project would take off.  I would buy 500 at that cost.

  Are you sure? yes | no

Laser Developer wrote 02/08/2019 at 05:52 point

Great stuff, welcome to the project!

  Are you sure? yes | no

John Stockton wrote 01/28/2019 at 15:06 point

I've seen some articles where they use a 3d printed item to make a 2-part silicon mold (shore hardness ~40) and make polymer castings of the item.  They have some very impressive resolutions and it seems like the cost could be cheaper than using a SLA printed version for each one.  An injection mold would be a good long term choice, but that's a lot of machining.  On plastic vs. glass, is there isn't a huge difference in cost, I'd at least look at the performance of glass.  Nicely executed project - I appreciate the detail you put into the schematics (PWB layout impedances, layout geometries, ...).  Great project so far!

  Are you sure? yes | no

Laser Developer wrote 01/28/2019 at 17:41 point

Thanks for the ideas John, I really appreciate the feedback.

We have looked at the 3D printed molding process in the past but I'm not sure that it can deliver the precision that we need - it has to be a three part tool. In any event, we'll probably go the injection moulding route as it only takes a few hundred shots to recover the costs and we want to get the price as low as possible to make the Unruly available to students.

We have compared glass and plastic lenses (we had both types custom made and ran side-by-side tests) and were surprised at how good the plastic lenses were. I personally prefer the plastic because the lenses are unbreakable and probably more suited to an experimental environment. Crash!

  Are you sure? yes | no

John Stockton wrote 01/26/2019 at 15:25 point

Even though the optical design is simple, do you have a posting with the prescriptions, spacing and filter bandwidth?  

  Are you sure? yes | no

Laser Developer wrote 01/26/2019 at 15:44 point

Not yet, but it's coming. We've chosen a design that uses plastic (acrylic) lenses with AR coatings and these along with the filter are custom made. Since we have already invested in the tooling for these parts we can offer them to hobbyists or other interested parties at really low cost.

The optical holder hasn't been tooled yet and we are currently 3D printing it. This tool will be quite expensive so unless we get enough interest it will remain an expensive part.

  Are you sure? yes | no

Ytitne the Great wrote 01/28/2019 at 18:52 point

Who is supplying these lenses? 

  Are you sure? yes | no

martin wrote 01/24/2019 at 08:22 point

Just simply STUNNING!!

  Are you sure? yes | no

Laser Developer wrote 01/28/2019 at 17:42 point

Thanks martin! Please feel free to chip in with comments or suggestions :).

  Are you sure? yes | no