pwm_input modifications

[arbulgazar]

** Reset feature added for pwm_input driver. **
GPIO_GPIO5_OUTPUT (AUX 6 signal pin) is currently used as reset pin.
Reset is happening if pwm_input don't get any data for some time, to prevent reset-spamming while lidar is not connected ROMFS should toogle pwm_input driver on parameter. (not implemented in this fork)

** Publication to range_finder topic added **
range_finder topic is used by several sensors (pwm_input, i2c lidar interface and range_finder currently) and ROMFS should have cases to enable ony one of them.

[LorenzMeier]

Thanks! Doing a full review now.

[LorenzMeier]

@Zefz This brings the Lidar-Lite to life. I need to think a moment about how to layer this architecturally.

[TSC21]

@LorenzMeier I was planning to get all rangefinders and altimeters data into DISTANCE_SENSOR message. Are there other plans (like keep using range_finder topic) or can I proceed with this? Mavros already is capable of sub and pub into DISTANCE_SENSOR msg.

[arbulgazar]

This shouldn't be there. Included for testing, forgot to remove. :(

[Zefz]

LidarLite still needs working PWM drivers (#1949).

[TSC21]

@arbulgazar it would be awesome if you could do your changes against #2200 👍

[bansiesta]

I can't get this to run yet, I keep getting:

nsh> pwm_input start
[pwm_input] advertising 536948912
nsh> [pwm_input] Lidar is down, reseting
[pwm_input] Lidar is down, reseting
[pwm_input] Lidar is down, reseting
...

Pin 5 and 6 are connected as shown here http://copter.ardupilot.com/wiki/common-optional-hardware/common-rangefinder-landingpage/common-rangefinder-lidarlite/#connecting_to_pixhawk_via_pwm, the rail is powered with 5v, and I'm using a 300 Ohm pull-down.

Any pointers what I'm doing wrong? Thanks.

[LorenzMeier]

Try one kilo ohm instead

[LorenzMeier]

Actually I think it's rather software related. Did you put FMU in PWM mode and / or disabled the driver? I need to check if there are missing dependencies / startup order things.

[bansiesta]

Already re-soldering :). I didn't touch the fmu command...

[bansiesta]

I switched to 1k Ohm and disabled the fmu command, still nothing!
Does the Lidar-lite have any indication that it's running? Any idea how to debug this?

[TSC21]

@bansiesta are you using this #1949?

[bansiesta]

@TSC21: I'm currently trying pwm_input. Is this not needed for ll40ls start pwm?

[TSC21]

As @Zefz said:

LidarLite still needs working PWM drivers (#1949).

[bansiesta]

Sorry, don't get it. Tell me what I need to run in what order and what you suspect to work.

[TSC21]

@bansiesta I can't really tell. From what I know, I2C drivers are working for LidarLite but have problems related to bus hold, which is a problem related to LidarLite only. That's why a PWM driver is needed. But, from what I can see, there's no current implementation for using PWM with LidarLite in the firmware, and that's what @Zefz was implementing, but it didn't finish it. So, I suspect @Zefz code must be done first before getting this to work.

[bansiesta]

@TSC21 Understood thanks. I've already continued where @Zefz left off and I think I have an idea what is going on there. However, for the PWM part to work, this pwm_input needs to work but doesn't for me. The interrupt seems never triggered.

[bansiesta]

Aha, seems to work now, a pwm_input test was necessary.

[TSC21]

Good catch :D BTW, check 43668cc. Big change on structure, as range_finder topic is totally replaced by distance_sensor.

[bansiesta]

@TSC21: thanks for the heads-up. I'll try to merge that in later.

[arbulgazar]

I wasn't modifying start-up scripts, was not sure you want lidar starting with fmu, but probably should have wrote this in comment:
pwm_input start full required in order to both initialize and start timer. pwm_input start will lead only to timer initialization. That's why you had to pwm_input test to get things working - it starts timer. This however is listed in help message you get on pwm_input.

[arbulgazar]

I will do changes with respect to #2200 after week or so. Thesis is demanding time now, can't push it back.

[bansiesta]

@arbulgazar: Thanks for your work, I based my work on it and went from there. It's merged now.

Btw:
I removed the pwm_input start full argument because the read is now triggered from the ll40ls driver. Also the pwm_input only publishes to the pwm_input topic and the ll40ls driver then converts this to a distance, etc..
This should leave the pwm_input driver as a generic interface that can be extended for other use cases.