Running TauLabs software on hardware that isn't supported? Work with others to get a new platform supported.

Naze32

User avatar
User

jhitesma

Posts

132

Joined

Tue Sep 30, 2014 8:14 am

Location

Yuma, AZ - USA

Re: Naze32

by jhitesma » Mon Apr 13, 2015 7:08 am

Well that explains a lot. Found a burned trace on my Naze board. The ground trace running under the resistor next to the boot pads right on the edge. Guessing that happened at the same time my orange openLRS RX suddenly went dark on me.

Ran a jumper on it and now things are acting a LOT more sane.

Looks like I'm very close to a test flight. Just for some reason cant' get it to arm. Inputs seem to be setup correctly, and if I set it to always armed it arms ok. But when I set up yaw right for arming I see the status blink to armed then right back to disarmed in GCS even though my inputs all look good.

But it's late and I'm tired so I'll fight with that tomorrow. Even if it means my chances of a test flight at lunch tomorrow are now slim to none :(
User avatar
User

jhitesma

Posts

132

Joined

Tue Sep 30, 2014 8:14 am

Location

Yuma, AZ - USA

Re: Naze32

by jhitesma » Thu Apr 16, 2015 5:08 am

Success!

Well, if you can call a little hovering in my tiny back room success :D

Arming problem was apparently because I had my throttle min and neutral at the same value and didn't notice. Once I fixed that it armed. But then it was acting like a had a motor set to the wrong channel...but after repeatedly checking them they all were responding correctly. Finally I found that apparently the last time I reset the board I forgot to adjust the yaw on the attitude tab to compensate for my board being mounted 90 degrees off center. Doh. With that fixed it started responding correctly!

Way too dark outside for a proper test flight, so I did what I hate to do and gave it a quick hover test in my way too small back room. Definitely not tuned well yet...but it flew.

Will try to give it a bit better of a tune tomorrow and get it outside (if we aren't having the crazy 30-40mph winds we had today again) with a video camera for a proper test.

Now to decide if I want to reflash my minim or just go without OSD....though without a working beeper I'm not sure I want to go without so I don't overfly my packs...
User avatar
User

jhitesma

Posts

132

Joined

Tue Sep 30, 2014 8:14 am

Location

Yuma, AZ - USA

Re: Naze32

by jhitesma » Fri Apr 17, 2015 4:02 am

Well did a bit of testing over lunch.

First test flight was a bit rocky, but impressed my mailman (not the one I don't like, got a new guy today. Wish I had had time to give him more details!)


This video isn't public yet since I haven't had a chance to review or edit it - quick preview looked like the mailman is anonymous enough that I feel safe posting it here ;)

Then did a test with autotune. Didn't go so hot. Tau numbers seemed lower than I expected, and the suggested PID's had an I on pitch that was so crazy high the GCS wouldn't even let me try to use it. But I expected as much because it was windy and my error values were through the roof. I tried doing the test under my carport to minimize the wind but I was still getting blown all over (mostly forward/back which is why my pitch numbers were so far off I suspect.)

Here's the autotune test:



Hopefully the wind will let up and I can do a better test this evening. Then I can tweak the rates and should have it flying more like I like it :D

Lot of potential there already...can't wait to get this tuned in a bit better.


Gave it another try after work...shared my autotune results here

It was still kind of windy but I gave it 2 more tries. Numbers just kept getting worse. Like Grouch/accelerotor I had suspiciously low Tau values and crazy high noise values. My board is mounted with o-rings already though so that isn't a solution.

What did seem odd was that in all three tests I got way more noise in pitch than roll and wound up with a crazy high I term for pitch that GCS wouldn't even accept it was ridiculously high. Also all three times the P term on roll came out lower than default while the P term for pitch came out considerably higher than default.

I did do a bit of quick manual tuning and upped my rates to 400 (pitch/roll) 500 (yaw) and it's feeling pretty nice - better than it did with baseflight IMHO. In Horizon it was super silky smooth, in acro it felt nice and responsive and like it would get better if I could spend some time to improve the tune. Did some flips but it's not really dialed in well enough for me to be happy with the way it did them yet.

Hope I get some calm air tomorrow to try autotune without wind to see if it does any better. But given how similar my results are to grouchos I'm suspecting there's something about autotune in this build that isn't happy. Having read the autotune development thread though and how much of it was so far over my head I can't begin to guess what the problem may be :D
User avatar
User

jhitesma

Posts

132

Joined

Tue Sep 30, 2014 8:14 am

Location

Yuma, AZ - USA

Re: Naze32

by jhitesma » Sat Apr 18, 2015 2:21 am

I have an idea what may be the problem with autotune on Naze32.

I noticed that when I enable the autotune module I get an error about the board missing the SystemIdent object.

A quick glace at the source seems to show that for most targets the objects are setup in a UAVObjects.inc file - but there's no such file in the naze32 tree. I also don't see any code like what in the UAVObjects.inc file in any of the other files in the naze32 tree...so I'm not sure just what's going on here.

Hoping someone more familiar with the code can give me some pointers as to whether I'm on the right track with this or not.
User avatar
User

peabody124

Posts

1339

Joined

Tue Feb 18, 2014 12:33 pm

Re: Naze32

by peabody124 » Sat Apr 18, 2015 3:08 am

No, it is included: https://github.com/peabody124/TauLabs/b ... efile#L209 after a reboot you'll see that warning go away.

I'm pretty sure Naze just lacks resources to service the autotune EKF calculation frequently enough.
User avatar
User

jhitesma

Posts

132

Joined

Tue Sep 30, 2014 8:14 am

Location

Yuma, AZ - USA

Re: Naze32

by jhitesma » Sat Apr 18, 2015 4:00 am

peabody124 wrote:No, it is included: https://github.com/peabody124/TauLabs/b ... efile#L209 after a reboot you'll see that warning go away.

I'm pretty sure Naze just lacks resources to service the autotune EKF calculation frequently enough.


Ahh, there it is. I was looking in the right place just got thrown off by the case when I was searching ;)

What made me suspect that was I also got that as an error instead of a warning when I tried to disable it after having rebooted and run autotune. When I unchecked the enable autotune module box I got a red error saying systemident wasn't present...though it makes sense that it must have been since I did get SOMETHING from the autotune.

I don't doubt that the Naze lacks resources...but given that it did work on CC3D for awhile which is so similar I'm still tempted to keep poking at it :) Though it seems that just about everything that could be disabled to free up more resources already is.

A little manual tuning tonight and it was flying pretty well though:


Little more tuning and some fresh props and it should be a LOT of fun. Got a new VTX on order for this frame tonight too so I can give it a go FPV style again soon too ;)
no avatar
User

Yamin

Posts

4

Joined

Wed Oct 28, 2015 2:10 pm

Naze32 Full Rev 5

by Yamin » Tue Nov 03, 2015 9:34 pm

Hi,

I have a naze32 full rev 5 that I am trying to build by enabling hmc5883, mpu6050 form
flight/targets/naze32/fw/pios_config.h

//#define PIOS_INCLUDE_MS5611
//#define PIOS_INCLUDE_HMC5883

However, I am getting a build error

pios_board.c:83:29: error: 'pios_exti_hmc5883_cfg_v5' defined but not used [-Werror=unused-variable]
static struct pios_exti_cfg pios_exti_hmc5883_cfg_v5 __exti_config = {
^
pios_board.c:53:29: error: pios_exti_hmc5883_cfg causes a section type conflict with pios_exti_mpu6050_cfg_v5
static struct pios_exti_cfg pios_exti_hmc5883_cfg __exti_config = {
^
pios_board.c:172:35: note: 'pios_exti_mpu6050_cfg_v5' was declared here
static const struct pios_exti_cfg pios_exti_mpu6050_cfg_v5 __exti_config = {
^
cc1: all warnings being treated as errors
make[1]: *** [/home/yami/dev/TauLabs/build/fw_naze32/pios_board.o] Error 1
make: *** [fw_naze32_tlfw] Error 2


then I added this line #588

//TODO: if(board_v5) { /* use PC14 instead of PB12 for MAG_DRDY (exti) */ }
if(board_v5) {
// rev5 hardware use PC13 instead of PB13 for MPU_INT
pios_hmc5883_cfg.exti_cfg = &pios_exti_hmc5883_cfg_v5;
}


and then I got this error
pios_board.c:83:29: error: pios_exti_hmc5883_cfg_v5 causes a section type conflict with pios_exti_mpu6050_cfg_v5
static struct pios_exti_cfg pios_exti_hmc5883_cfg_v5 __exti_config = {

well I have not worked on C for 10 years and I just start learning about this product, however it look like these sensors are not implemented for Naze32 full.


Can someone please help me to build the naze32 full version. I understand that there's not enough RAM to enable these things without trimming some other stuff down, but that is very fine for me as I want to experiment with the "follow me feature" mainly by doing the heavy processing form the ground station while sending navigation commands to the drone in real time.
no avatar
User

Yamin

Posts

4

Joined

Wed Oct 28, 2015 2:10 pm

Re: Naze32 Full Rev 5

by Yamin » Wed Nov 04, 2015 2:39 am

okay I found a way to fix this

add
SRC += $(PIOSCOMMON)/pios_hmc5883.c
to MakeFile for naze32

in pios_board.c
also added "const" keyword to pios_exti_hmc5883_cfg and pios_exti_hmc5883_cfg_v5


static const struct pios_exti_cfg pios_exti_hmc5883_cfg __exti_config = {

but still the Mag is not working !!
Previous

Who is online

Users browsing this forum: No registered users and 1 guest

Powered by phpBB ® | phpBB3 Style by KomiDesign
cron