Replies to This Discussion

3D RoboticsPermalink Reply by Chris Anderson on August 24, 2009 at 2:42pm

1, 2) Yes, we're noobs with SVN and never got it working correctly. And Jordi doesn't like the whole checkin/checkout process ;-) For now, let's just email patches or post them to the ArduPilot thread. If there are a lot, we'll get SVN working.

3) Yes, the Windows version of 17 is stable. If you're using Linux and having trouble, stick with 16. ArduPilot works fine with either.

4) That sounds right. It's been a while since I looked at the NMEA parser (we don't use it anymore), but that's my recollection

5) Long term, we want one code base and one hardware configuration. Given how many people have the EM406, we'll continue to support that but as soon as we release the uBlox5 combo, we'll make that the recommended GPS. It's too hard to support random combinations of hardware, which spiral into an unlimited number of one-off configuration problems. If people want support, they'll have to use the recommended configurations. (Of course, they're free to use other hardware and software just as long as they don't expect us to debug their setups).

6) GPS error (other than heading) is not a big problem with uBlox 5

• ▶ Reply

Permalink Reply by I Heart Robotics on August 25, 2009 at 3:06pm

Ok today I am looking at the servo PWM subsystem...
This afternoon I got the connectors soldered to the ArduPilot and hooked up a scope and it looks like the way the PWM signals are generated is interesting. I went through the source and it looks like...

Channel 1 and Channel 2 are generated by a hardware PWM generator that gives approximately an effective 9 bit resolution due to the 2us resolution. This uses roughly 0 cpu since everything is in hardware.

Channel 3 (the throttle) is generated by hitting an interrupt every 8us and running a counter to the right time to flip the i/o line. This give us a 7 bit resolution on the throttle and is drive by an interrupt.

On the scope it looked like channel 1 and 2 turn on at the same time and from the code it looks like the period for channel 3 is probably different than the period for 1&2.
It also looks like the capture for flyby wire is driven by the hardware interrupt pins.
So the ArduPilot does not output PPM compatible PWM signals.

Does this sound correct?

If you have the software internals documented somewhere let me know and I can read that instead of asking all these questions.

• ▶ Reply

3D RoboticsPermalink Reply by Chris Anderson on August 25, 2009 at 10:06pm

That sounds correct, but it's a bit over my head. PM Jordi, who actually knows this stuff, if you want confirmation.

• ▶ Reply