Thread Tools
Sep 13, 2019, 03:03 PM
Registered User
We're now getting pretty good results with the PixRacer mounted in a little 7" racer. We have a stack of them, and fly them in small swarms using an Odroid and WiFi to coordinate. It's been super reliable once we figured out all the nuances and tweaked the flight envelop suitably. They are usually autonomous end-to-end.

However, now I have one mounted in a larger, very lethargic quad, and struggling a bit. I only have a few flights, so some issues can likely be handled with tuning and minor frame mods. The thing weighs about 3kg and has T-motor U3s at 700kv swinging 14" T-Motor CF props on 4S. I performed minor PID tuning indoors in stabilized mode. In the first flight outdoors, switching to alt-hold mode and position hold mode were uneventful, and both performed pretty well. There was a bit of wind, maybe 5mph gusting to 10, and around trees and buildings, so a bit turbulent.

I then set up a very small WP circuit, and everything was still good. I enlarged the circuit a bit - still good. Then I stretched it more, and things got weird. On initial climb out it was supposed to climb to 25m with a max climb rate of 2.5m/s, but halfway through the climb it accelerated, actually maxed out at around 4.4m/s and overshot the altitude. After that, it stabilized OK and flew through the next few WP. Finally, at the last WP before returning to home it had to yaw almost 180 degrees, and part way into this it completely lost it. I switched to stabilized mode and landed it, but it looked disastrous in nav mode. It actually gained about 10m while oscillating wildly. Looking at the logs, I see that two motors were saturated, one for about 3 seconds. I guess yaw authority is super low, but desired yaw performance, by default, is rather high, at least for a quad like this. I've since lowered the yaw rate and yaw acceleration a good bit, but haven't yet flown the changes. Are there any other parameters that, for example, might prevent motor saturation, as autonomous flight is generally unsustainable once that limit has been reached. I may crank the motor axes a bit to provide a yaw-thrust vector, if I can't resolve this another way.

There is a fair amount of clipping during both anomalies, and several "IMU# in-flight yaw alignment complete" messages. There's a GPS glitch recorded maybe 4 or 5 seconds before I took over, which I'm guessing was due to the EKF solution failing during the wild oscillations. The glitch cleared a few seconds after going to stab mode. There are also a few "EKF primary changed: #" messages. What do those mean? Is that an indicator of ailing EKF due to vibes? The PixRacer is actually semi-rigidly mounted to the frame, so I'll soften that up a bit to try to reduce vibes. I'm just trying to understand how to correlate things I experience in flight to messages and errors that show up in the logs.
Sign up now
to remove ads between posts
Sep 13, 2019, 03:33 PM
Registered User
dkemxr's Avatar
I had a quad about that size with 3508-380kv motors and a PixRacer. I would suggest a very compliant mount. I found that a mini ball mount worked well. If it can tame the vibes on a junk F450 frame with extended aluminum arms it's doing a pretty good job!

Quote:
Originally Posted by jones007
We're now getting pretty good results with the PixRacer mounted in a little 7" racer. We have a stack of them, and fly them in small swarms using an Odroid and WiFi to coordinate. It's been super reliable once we figured out all the nuances and tweaked the flight envelop suitably. They are usually autonomous end-to-end.

However, now I have one mounted in a larger, very lethargic quad, and struggling a bit. I only have a few flights, so some issues can likely be handled with tuning and minor frame mods. The thing weighs about 3kg and has T-motor U3s at 700kv swinging 14" T-Motor CF props on 4S. I performed minor PID tuning indoors in stabilized mode. In the first flight outdoors, switching to alt-hold mode and position hold mode were uneventful, and both performed pretty well. There was a bit of wind, maybe 5mph gusting to 10, and around trees and buildings, so a bit turbulent.

I then set up a very small WP circuit, and everything was still good. I enlarged the circuit a bit - still good. Then I stretched it more, and things got weird. On initial climb out it was supposed to climb to 25m with a max climb rate of 2.5m/s, but halfway through the climb it accelerated, actually maxed out at around 4.4m/s and overshot the altitude. After that, it stabilized OK and flew through the next few WP. Finally, at the last WP before returning to home it had to yaw almost 180 degrees, and part way into this it completely lost it. I switched to stabilized mode and landed it, but it looked disastrous in nav mode. It actually gained about 10m while oscillating wildly. Looking at the logs, I see that two motors were saturated, one for about 3 seconds. I guess yaw authority is super low, but desired yaw performance, by default, is rather high, at least for a quad like this. I've since lowered the yaw rate and yaw acceleration a good bit, but haven't yet flown the changes. Are there any other parameters that, for example, might prevent motor saturation, as autonomous flight is generally unsustainable once that limit has been reached. I may crank the motor axes a bit to provide a yaw-thrust vector, if I can't resolve this another way.

There is a fair amount of clipping during both anomalies, and several "IMU# in-flight yaw alignment complete" messages. There's a GPS glitch recorded maybe 4 or 5 seconds before I took over, which I'm guessing was due to the EKF solution failing during the wild oscillations. The glitch cleared a few seconds after going to stab mode. There are also a few "EKF primary changed: #" messages. What do those mean? Is that an indicator of ailing EKF due to vibes? The PixRacer is actually semi-rigidly mounted to the frame, so I'll soften that up a bit to try to reduce vibes. I'm just trying to understand how to correlate things I experience in flight to messages and errors that show up in the logs.
Sep 13, 2019, 05:11 PM
Registered User
Quote:
Originally Posted by dkemxr
I had a quad about that size with 3508-380kv motors and a PixRacer. I would suggest a very compliant mount. I found that a mini ball mount worked well. If it can tame the vibes on a junk F450 frame with extended aluminum arms it's doing a pretty good job!
Thanks for the quick feedback. I'll end up with something like that. On all my old fixed-wing applications of Pixhawk, we used the stock foam sticky pads along with a printed mount that let us snap the Pixhawk in and out pretty easily without damaging the foam. I've never tried the foam on copters though. In both the swarm copter and in a modified Finwing Penguin we use the PixRacer with elastomer standoffs and O-rings - fairly rigid, and they have been good.

Do you know if excess vibes can cause it to climb faster than expected? I think that's the main unknown behavior I experienced. Also, any thoughts on keeping yaw commands from saturating motors?
Sep 13, 2019, 05:54 PM
Registered User
dkemxr's Avatar
Yes, altitude instability can be caused by high vibes. I don’t think I have ever had a problem with a mount that is too compliant despite the warnings about that.

The yaw issue I have seen on that quad [I posted about] and another “dead cat” geometry that was even worse. I gave up on that craft. The Yaw authority was poor and I don’t have an answer for it. I tried various parameter changes with little improvement. I chalked it up to low kV Motors and spinning mass and no help for it.

Quote:
Originally Posted by jones007
Thanks for the quick feedback. I'll end up with something like that. On all my old fixed-wing applications of Pixhawk, we used the stock foam sticky pads along with a printed mount that let us snap the Pixhawk in and out pretty easily without damaging the foam. I've never tried the foam on copters though. In both the swarm copter and in a modified Finwing Penguin we use the PixRacer with elastomer standoffs and O-rings - fairly rigid, and they have been good.

Do you know if excess vibes can cause it to climb faster than expected? I think that's the main unknown behavior I experienced. Also, any thoughts on keeping yaw commands from saturating motors?
Sep 15, 2019, 05:29 PM
Registered User
Quote:
Originally Posted by dkemxr
Yes, altitude instability can be caused by high vibes. I don’t think I have ever had a problem with a mount that is too compliant despite the warnings about that.

The yaw issue I have seen on that quad [I posted about] and another “dead cat” geometry that was even worse. I gave up on that craft. The Yaw authority was poor and I don’t have an answer for it. I tried various parameter changes with little improvement. I chalked it up to low kV Motors and spinning mass and no help for it.
Thanks for the info on the yaw issues. I may try to crank the motor axes into the yaw direction a bit then. This is really common on hex, oct and higher, and always present on giant scale (air-taxi) prototypes. I just haven't seen it on a lot of quads. I haven't thought about it much, but should probably think a bit about pitch and roll and make sure it's not going to mess those up too much. I guess it should still be symmetric about those axes, so should be alright. I guess a small loss in overall efficiency, but should be pretty small. With say 5 degrees of cant, I'd lose about 0.4% thrust, but add about 8.7% of the differential thrust from commanded yaw into a direct yaw thrust component. Seems reasonable.

This is sort of the bane of high efficiency systems. The more efficient your propulsion system is, the less drag torque the props have, and therefore the yaw authority is reduced. Probably there is a strong correlation between FOM (g thrust per Watt power) and yaw authority. All my 5" and smaller racers, which live in the 1.5 to maybe 3 range for FOM have fantastic yaw authority, but of course pretty pitiful endurance. All my high endurance aircraft have FOMs between about 8 and 10 g/W, but are crazy weak in yaw.

I guess one other option is to eliminate yaw maneuvers entirely. For most copters, in particular with a camera payload, this seems like a non-starter, but for this particular quad there is no sensor that needs to be pointed and the airframe is completely symmetric, so there is really no requirement to point in any particular direction. Any idea how ArduCopter likes leaving yaw fixed during WP navigation?
Sep 15, 2019, 09:42 PM
Registered User
dkemxr's Avatar
I don't see why this wouldn't be a good solution. Set WP_YAW_BEHAVIOUR to 0 (never change yaw) and check it out. It's funny that I fly plenty of missions where the yaw direction makes no difference but have never set this parameter for other than 2.

Quote:
Originally Posted by jones007
I guess one other option is to eliminate yaw maneuvers entirely. For most copters, in particular with a camera payload, this seems like a non-starter, but for this particular quad there is no sensor that needs to be pointed and the airframe is completely symmetric, so there is really no requirement to point in any particular direction. Any idea how ArduCopter likes leaving yaw fixed during WP navigation?
Sep 17, 2019, 06:48 AM
Registered User

Pixracer on a plane, with an analog airspeed sensor


Hi All,

I'd like to install my pixracer R14 in a plane, and want to add an airspeed sensor, I have an analogue airspeed sensor to hand, but am wondering how to wire it, it has 3x wires, I assume signal, vcc and gnd, is it possible and were should I connect it ?

Thanks

Andy
Last edited by adibbins; Sep 17, 2019 at 12:08 PM.
Sep 19, 2019, 04:57 AM
Berlin/Germany
I remember, there was a comment to connect an analog sensor to the RSSI port of the Pixracer. Unfortunately I never tried it.

Andreas
Sep 22, 2019, 02:14 PM
Registered User
Quote:
Originally Posted by adibbins
Hi All,

I'd like to install my pixracer R14 in a plane, and want to add an airspeed sensor, I have an analogue airspeed sensor to hand, but am wondering how to wire it, it has 3x wires, I assume signal, vcc and gnd, is it possible and were should I connect it ?

Thanks

Andy
Sorry, haven't tried it, but it works great with the new temp-compensated digital sensor. ;-)
Oct 07, 2019, 04:38 AM
Registered User

Pixracer with Arduplane and telemtry


Hi All,

So I've installed my pixracer in a plane, it's working well, except getting the frsky telemetry output.

I've followed the guidance, and linked the FS-In and FS-Out, pushed that thru to my RX smart port, set serial4 to protocol 10 and baud rate to 9600, tried to discover new sensors on Taranis and nothing.

Can someone help, have I missed a step ?

I've also tried to create a telemetry cable using one of those tiny RS232 to ttl converters, plugged into TL2 or TL1 , no joy. FYI: A Slik radio in TL1 or TL2 works, so I'm confident these serials work.

FYI: I'm using the latest stable arduplane pixracer release 3.9.11

Thanks in advance.

Andy
Oct 07, 2019, 07:26 AM
Registered User
dkemxr's Avatar
Protocol 10 is for pass through. If you just want to discover sensors use protocol 4. You don't need a converter with Pixracer. A much better solution is to use The Yaapu Frsky Telemetry Script and then use Protocol 10.
https://github.com/yaapu/FrskyTelemetryScript


Quote:
Originally Posted by adibbins
Hi All,

So I've installed my pixracer in a plane, it's working well, except getting the frsky telemetry output.

I've followed the guidance, and linked the FS-In and FS-Out, pushed that thru to my RX smart port, set serial4 to protocol 10 and baud rate to 9600, tried to discover new sensors on Taranis and nothing.

Can someone help, have I missed a step ?

I've also tried to create a telemetry cable using one of those tiny RS232 to ttl converters, plugged into TL2 or TL1 , no joy. FYI: A Slik radio in TL1 or TL2 works, so I'm confident these serials work.

FYI: I'm using the latest stable arduplane pixracer release 3.9.11

Thanks in advance.

Andy


Quick Reply
Message:

Thread Tools