Bringing up the GluonPilot board

[elkaim]

Hello Tom (and others),

I'm trying to bring up my new GluonPilot board and having some interesting difficulties. I have a serial port level shifter to connect to the board, and 5V power coming in from a bench power supply.

When I open the com port in RealTerm, I can see the data coming by. When I open it in the Config utility, I get nothing at all, not a single message.

I have the serial port set at 115200 8-N-1, and the Gluon Config v0.3

Any suggestions?

Thanks, Gabriel.

[Tom]

Hi Gabriel,

So the connection is no longer the problem.

Did you uncheck the "Hide parsed lines" checkbox?

If yo leave the connection open and restart the module, you should also see the boot-log of the module: http://www.gluonpilot.com/wiki/Bench_te ... the_module

Tom

[Mitch]

The gluonpilot board will dump status messages to uart1 as soon as it is powered on. If you are seeing readable text in realterm then the baud rate is correct. If you are seeing just gibberish characters then adjust the baud rate to find the correct one where the text is readable. Gluonconfig uses a fairly simple com port interface (the source code is available in the svn repository). Double check the port you're using (windows device manager) and the baud rate.

I've used ttl rs232 converters on pic uarts for many many projects and it should work fine - just sometimes the baud rate isn't so easy to set up. I've also had a few TTL interfaces that would not function reliably at higher baud rates. Usually 57.6K and slower works OK but I have had to back down to 38.4K.

I think most of us are using ftdi interfaces simply because the usb is so much easier to do in hardware once everything is set up and seems to be more reliable at higher baud rates. Here is a link to a very simple FTDI that can be configured for 5 or 3.3 volts - also inexpensive and easily available in the states http://www.sparkfun.com/commerce/product_info.php?products_id=9115 unfortunately out of stock right now. This little adapter does interfere with the gps header connector when it is plugged directly onto the gluonpilot board but it is really necessary to eventually eliminate the +5 connection when the board is powered with a battery. Thus, I made up a 3 wire jumper for the sparkfun ftdi that only connects tx, rx, and gnd - solves both problems.

I also assume you are using a single power supply that is powering both your ttl-rs232 converter and the board.

BTW I assume this is Prof Elkaim. Welcome to the forums. It is very cool to see you are interested in this project! I know you are an expert in some of the algorithmic things we are doing here. Tom has done some great work on the hardware and firmware, and several of us are working on additional firmware for better three space IMU/sensor fusion/Kalman filters for vtols, helis, etc. I've run your estimator algorithm on the board with the addition of an hmc5843 magnetometer triad and it has great potential if we can get the code streamlined a bit. I posted a little demo video of the estimator here http://gluonpilot.com/forum/viewtopic.php?f=3&t=8&start=10

[elkaim]

Hi Mitch and Tom,

Thank you for both jumping in here. I have the board set up in the lab, and I'm still getting no response from the Gluon Config program. The baud rate, parity, and connections all seem to be good. I have a separate bench supply powering the Gluon board, and the RS232 has a common ground.

In realterm, I see the text of the Gluon boot and start dumping data out just fine. Then I close the port, open it in Gluon Config and connect, and nothing (yes, I have the hide parsed messages unchecked). Nothing, no response at all. Close the port on GluonConfig and open it in RealTerm and there is the data.

I'm running XP on a Dell Laptop that has its own serial port, and it seems to work fine. I've checked it several times and can read and write to it on other serial adapters (same thing happens when I use a USB adapter).

Any ideas?

Thanks, G.

PS: Yes, I am that Prof. Elkaim. We've developed a larger, more capable autopilot board (see http://slugsuav.soe.ucsc.edu/ ) that uses the same processor and is programmed directly from MATLAB. I'm very interested in seeing if we can move most of that code onto the Gluon as a "smaller and simpler" version of what we do.

PPS: I'm very happy to help you with the algorithms using the two-vector estimation. Really, though, you want to keep track of the whole DCM or quaternion. I've got some really good algorithms for calibrating the mags and accels, and if you have a rate table and a lot of patience, the gyros.

[Mitch]

It sounds like you have the hardware issues covered. Perhaps there is a gluonconfig dependency missing on your laptop. Maybe a dll version. Perhaps it would be worth trying to run gluon config from the visual studio environment rather than from the precompiled version. I uploaded a zip with a snapshot of gluonconfig source at r38 here http://baxtercam.no-ip.org/images/gluonconfig%20r38.zip (1.8 Mb too big to attach here).

Tom has much more expertise here than I do. If you don't have visual studio 2008, you can install the free express version for C# from here http://www.microsoft.com/express/Downloads/ You can access the entire svn repository here http://gluonpilot.googlecode.com/svn/trunk/

Cool stuff at ucsc. I couldn't get a connection to the git repository at git://overtux.cse.ucsc.edu/malife/apcode.git
I did however get a few snapshots to peruse.

[elkaim]

Hi Mitch,

I am in the process of downloading and installing VS2008Pro. I had 2005, but that did not do it for the Gluon code. Once I have that, I can see what changes when I compile it.

I believe that we moved the SLUGS repository to github, search for Malife and you should find it. They system is really quite slick.

--G

[Mitch]

Well you solved my problem! I found it here http://github.com/malife/SLUGS git://github.com/malife/SLUGS.git

When you run the gluonconfig source - there may be a few nuisance bugs. I believe they are VS2008 bugs. If you get a window that asks to run the previous working version then do that.

[Tom]

This is a really strange problem. We all use FTDI cables and this seems to work fine... My laptop doesn't have an RS232 serial port so I can't test it.

Do you see any output with this tool: http://msmvps.com/blogs/coad/archive/20 ... -.NET.aspx

This uses the same SerialPort lib as I do for the gluon config.

Tom

[elkaim]

So, the plot thickens...

I have a FTDI USB board for my XBEE radio, and I used that to talk to the GluonPilot, and it worked. Basically, that library that talks on serial only works with FTDI converters.

Both a built in RS-232 and a Prolific USB->Serial don't work (either on Toms code or on the other one he linked to). In any case, now I can talk to the board and will be seeing if I can get everything else that I need to run on it, and possibly make it do some interesting things.

By the way, I am assuming that I can go through the XBEE radio bridge using the modules on either side if I so desire. Is that correct?

--G

Next stop: R/C receiver and a servo output.

[Mitch]

Some issue with the net 2.0 serial! I use xbees for the serial connection to make it easier to move things around - usually with no problem but I wouldn't try to upgrade flash. Sometimes I think the throughput is too slow and things will lock. I configure them at 115200bps but I only have very inexpensive xbees with limited buffer and rf data rate.

I just posted a rambling post in the General UAV section for Lucasz about ppm and pwm receiver connections.