149 replies to this topic

[Brian]

The problem is getting deeper...  I thought I had saved settings on both board, but I may have been running on defaults on one.  I just saved settings on the "good" one, and now it's showing the same symptoms.  I removed reading of the settings, so it's just getting the defaults, and it's the same, so it's not a settings problem.

It appears now that flashing the settings has effected USB somehow.  Does that make sense to anyone?

The address that I'm writing to is:  0x08000000 + 0x20000 - 1024, which I believe should be the last page of flash.  Is it possible that there is something having to do with USB at that location.

In anycase, I wouldn't try saving settings with this firmware until this is worked out.

[Kenn Sebesta]

Brian, on 13 May 2012 - 11:37 PM, said:

The status/configuration interface looks like:

CropperCapture1.png

Mine looks more like this:

 Screen Shot 2012-05-13 at 10.43.24 PM.png   407.18K   22 downloads

Aside from that, it's working fine so far. I'm going to give it a flight test tomorrow, I hope, in the batplane.

[Brian]

Kenn Sebesta, on 14 May 2012 - 02:45 AM, said:

Mine looks more like this:

Screen Shot 2012-05-13 at 10.43.24 PM.png

Aside from that, it's working fine so far. I'm going to give it a flight test tomorrow, I hope, in the batplane.

I don't profess to be a QT wizard! .It seems like there's enough space below for it to expand into, but I don't fully understand how it works.  I wonder if it's trying to keep the two columns the same size.  I thought I had an expansion thinggy on the bottom of the right side, but maybe it works differently on the Mac.

Gook luck with the flight test.  I hope it works!

[Brian]

It turns out that there was a race condition on reserving TX/RX packets that was causing the USB stream to backup.  I just pushed an updated version with this bug fixed.

I uploaded a new firmware to the first post.  The default configuration works with a UAVTalk connection over either USB HID or on the telemetry port.

This firmware also has support for transparent serial over either of the serial ports or the USB VCP, but this hasn't been tested.  the VCP currently doesn't work on Windows, so I either have to setup a serial to serial test setup, or wait for the VCP update to stabilize.

Please try this out and give me any feedback.

[Kenn Sebesta]

Aha, that might explain why I had loads of telemetry dropouts on Monday. It was as if the board were being plugged and unplugged, as GCS physically lost contact with it.

It also might have been because I was communicating with a Revo prototype board, so maybe there was something going on on that end. Any possibility it's sensitive to the UAVTalk recipient?

[Kenn Sebesta]

Brian, on 14 May 2012 - 02:53 AM, said:

I don't profess to be a QT wizard! .It seems like there's enough space below for it to expand into, but I don't fully understand how it works.  I wonder if it's trying to keep the two columns the same size.  I thought I had an expansion thinggy on the bottom of the right side, but maybe it works differently on the Mac.

Gook luck with the flight test.  I hope it works!

Peabody124 just pushed a new branch, called james/ui_unbreak. He might be able to advise on how to fix the problem where the fields crunch together, instead of going to a scrolling page. I think he did this for one of the config gadgets by changing the style sheets.

[Brian]

Kenn Sebesta, on 16 May 2012 - 10:01 PM, said:

Aha, that might explain why I had loads of telemetry dropouts on Monday. It was as if the board were being plugged and unplugged, as GCS physically lost contact with it.

It also might have been because I was communicating with a Revo prototype board, so maybe there was something going on on that end. Any possibility it's sensitive to the UAVTalk recipient?

Let me know how the latest version of the firmware works.  I started it streaming last night before I went to bed, and the link was still up when I got up.  Unfortunately, the GCS had locked up at some point, so I'm not sure how long it was actually streaming data!

I hope to have a Revo to test with at some point. (Hint, hint, wink, wink, nudge, nudge)

[Brian]

Kenn Sebesta, on 16 May 2012 - 10:03 PM, said:

Peabody124 just pushed a new branch, called james/ui_unbreak. He might be able to advise on how to fix the problem where the fields crunch together, instead of going to a scrolling page. I think he did this for one of the config gadgets by changing the style sheets.

Since I don't have access to a Mac, I'm not sure how much I can test the GCS issues on the Mac.  I suspect it's something fairly simple, probably with that filler at the bottom.

[peabody124]

I'll try and test it tonight.  Also soldering your revo is high on my list for this week

[Brian]

peabody124, on 16 May 2012 - 11:01 PM, said:

I'll try and test it tonight.  Also soldering your revo is high on my list for this week

Whohoo!

I just pushed an update to the PipX V2 branch that fixes a bug in the (as previously untested) transparent com configuration.  You can now configure a pair of PipXs to transmit raw data on each of the three com ports (telemetry, flexi, and VCP).  I haven't actually tested the flexi port or VCP port.  I need to update the VCP code to work on Windows when that gets through review.

Top priority now is trying to figure out why characters are being dropped on the COM port.  Looking at the statistics, it looks like it drops 1-2 per second under heavy load, which seems pretty bad.