We saw the VCSEL getting damaged in random cases in our board. We have followed the same design as in ADI-96 Development Kit (but a different VCSEL). To find the reason for this, we did multiple experiments to check if ADI Processor can increase the ON time of VCSEL.Based on our probings, we found the ON time at which the VCSEL is driven was more than our expected condition. That is, from the actual ON time of 68 ns, the VCSEL ON time was found to be around 650ns.This observation was seen while removing the TAL feedback resistor (Resistor R30 in Laser board schematics). We observed this damage only during the initial boot of ADI processor..What could be the possible reasons for the damage? Also, does ADI Processor have control to dynamically change the ON time of LD signals?
Thank you for the inquiry. Your clarification on few items below would help us to understand the issue better.
- Have you observed similar behavior with the 96 boards and standard SDK using one of…
- Have you observed similar behavior with the 96 boards and standard SDK using one of the standard three modes (Near/Mid/Far) ?
- I suppose you use a custom firmware because none of standard modes has the nominal pulse width of 68 ns. Could you give us an idea about your firmware?
- When TAL feedback resistor is removed, how long do you observe 650 ns of ON time ?
- What would be the peak current and duty cycle of VCSEL in your system?
Hi,Please find answers for the queries below,1. Have you observed similar behavior with the 96 boards and standard SDK using one of the standard three modes (Near/Mid/Far)? No. this observation was seen in a board designed by e-con2. I suppose you use a custom firmware because none of standard modes has the nominal pulse width of 68 ns. Could you give us an idea about your firmware? Yes, we use a custom firmware that supports a range of 1m-6.5m.3.When TAL feedback resistor is removed, how long do you observe 650 ns of ON time? Initially. the 650ns is seen. Later after a few microseconds, the ON time crossed values around 20us. 4.What would be the peak current and duty cycle of VCSEL in your systemWe use 5V as the supply voltage to VCSEL with ON time of 50-55ns and Peak current of 12A was observed.
Thank you for the answer.
I have a few suggestions here.
1. The timing of VCSEL is fully controlled by the firmware running on AFE and your custom firmware might behave differently from the standard firmware provided with the 96 board. If you followed the standard design, you should be able to load up one of the standard modes you can download from the github (https://github.com/analogdevicesinc/aditof_sdk/) onto your hardware. Please let us know how that works and whether you observe the same behavior. It is difficult for us to debug issues with your custom firmware.
2. It is still not clear how long the 20 use pulse continues. 650 ns pulse width for few microseconds may not have enough energy to damage the device. And I suppose the pulse width becomes something close to 68 ns eventually? At this moment, I am not sure whether specific firmware can generate such a long pulse during initialization, however, longer pulse width for short period of time may not be the root cause of your VCSEL damage for the reason mentioned below as long as the average power is kept low.
3. It appears that your pulse width and peak current have been pushed to the limit. The most probable failure mechanism of VCSEL is usually thermal breakdown. It means that the average power applied to VCSEL matter more than instantaneous pulse width. VCSELs and driver circuit of 96board is designed to handle the duty cycle less than 5% with the peak current of 7A. It looks that the duty cycle and peak current of you system seems higher. If your firmware is designed to handle 1~6.5 meters with 68 ns pulse width, the average power applied to VCSEL could be higher than our standard configuration by few times as I believe your system has wide range mode, wider pulse width, and higher peak current. Please check the duty cycle and check with the VCSEL vendor whether it can handle that much of average power.
Hope this is helpful.