My brushless gimbal result

[tstrike]

Beautiful pictures Boltzman and cool idea using the disco. Wonder if you used some of those rctimer 550 long arms, but then I guess it wouldn't be very packable.
How do you support the gimbal, is there a bearing block between the frame plates? I'm asking because I've got a large foldable quad that was designed to have a tilt only gimbal on the nose and this would be perfect.
Also, do you have vibe issues?

 

[R_Lefebvre]

IMO, given that the DJI Flamewheel arms are so incredibly crappy to start with, adding extensions to them is just pointless. The F450 and Disco, etc. are good as really cheap, basic quads. But if you want to move up and fly with a real camera, you're going to have to get a real AP frame.

 

[RTRyder]

IMO, given that the DJI Flamewheel arms are so incredibly crappy to start with, adding extensions to them is just pointless. The F450 and Disco, etc. are good as really cheap, basic quads. But if you want to move up and fly with a real camera, you're going to have to get a real AP frame.

+1, I've got 4 FPV quads based on the F450 arms, two are standard F450 frames and the other two are TBS Discovery frames. Best part about them is they're cheap to fix, those weak arms usually break on impact dissipating most of the inertia that would otherwise break more expensive pieces like the Graupner props and at $4 each I can break all 4 arms and if none of the props break I figure it was worth the loss of the arms. Great for frames that will intentionally get abused and wind up taking some rough landings but less than ideal for carrying anything of value IMO.

I had one of the Discovery setup with Avroto motors but a smooth though moderately hard flat landing broke three of the 4 arms due to the extra weight of the Avroto, swapped the motors back to the standard DJI and it's survived harder landings without incident. Bigger is not better in this case...

Ken

 

[Bartman]

+1, I've got 4 FPV quads based on the F450 arms, two are standard F450 frames and the other two are TBS Discovery frames. Best part about them is they're cheap to fix, those weak arms usually break on impact dissipating most of the inertia that would otherwise break more expensive pieces like the Graupner props and at $4 each I can break all 4 arms and if none of the props break I figure it was worth the loss of the arms. Great for frames that will intentionally get abused and wind up taking some rough landings but less than ideal for carrying anything of value IMO.

I had one of the Discovery setup with Avroto motors but a smooth though moderately hard flat landing broke three of the 4 arms due to the extra weight of the Avroto, swapped the motors back to the standard DJI and it's survived harder landings without incident. Bigger is not better in this case...

Ken

keep in mind though that having truss arms directly under your props is not conducive to efforts to minimize airframe vibration. that truss design couldn't be much worse if minmal vibrations are the goal. especially so close and directly under a pulsing dynamic airflow.

 

[r0beert0]

Hi Guys, loving this gimbal already... this is the unedited (of course no post stab) straight out of the GP3 second and third test flights, first was still getting the power settings down so she had the shakes (thanks Ken for the power settings tip!). I have a camera operator (getting tilt happy though once we get this FPV for him he will not be guessing for framing the shot.

ok so:

1. Some micro vibrations are reaching the gimbal, I will be working to fix that, plus I was pushing the hex harder than I normally would for a shoot.
2. I’m getting the horizon shift, it becomes apparent on my yaw turn at 0:50 and more so at the 1:00 mark of the second flight.

So what do you guys thing is causing the horizon shift? aside from getting these sorted I think that brushless gimbals are the greatest thing since sliced bread!!!

second flight

on this third flight (same day) the horizon shift is not as bad... no settings/setup change from the second flight.

third flight

once all is sorted out I’ll post my settings for anyone using a GP3

PS I’m using a 2s 1000mah pack on this setup.

 

[RC Flying]

Jealous, I'm itching to get my hands on one and give it a try.

 

[R_Lefebvre]

I'm seeing more and more of this horizon tilting issue as more people use it. I have a theory based on my understanding of the AHRS in our system. Basically, when you turn those corners at speed, the accelerometers see the centripetal acceleration as a sideways gravity vector, so it thinks it's tilted, so it tries to compensate.

We solved this by using GPS velocity reporting to tell the AHRS when we're rounding a corner, and subtract the centripetal force, leaving only the gravity vector.

 

[R_Lefebvre]

keep in mind though that having truss arms directly under your props is not conducive to efforts to minimize airframe vibration. that truss design couldn't be much worse if minmal vibrations are the goal. especially so close and directly under a pulsing dynamic airflow.

Everything is wrong with those truss arms. They're terrible. The only reason they exist it to be able to pop them out of a cheap single-slide mold at minimum cost.

 

[jfro]

Got my gimble in today.

Found the manual, where do I find the GUI ?

Anybody know what the range of voltage we can use? Anybody know what's optimal?
I've seen 2s and 3s so far. So 8.4v to 12.6v is in range?

If anyone has one with a gh2 or gh1, let me know. That's what I'm going to try.

 

[r0beert0]

http://www.simplebgc.com/eng/downloads/

they have 1.3 and 1.2 GUI's I downloaded both, my board reads version 1.2b1 so I'm using the 1.2 interface. As far as voltage... I find that on my GP3 since its a light camera 2s with slightly adjusted power settings is working fine, heavier cameras should go with 3s. though I'm still adjusting settings myself, will post soon after I'm happy enough with what I end up with.

-Roberto

 

[r0beert0]

I'm seeing more and more of this horizon tilting issue as more people use it. I have a theory based on my understanding of the AHRS in our system. Basically, when you turn those corners at speed, the accelerometers see the centripetal acceleration as a sideways gravity vector, so it thinks it's tilted, so it tries to compensate.

We solved this by using GPS velocity reporting to tell the AHRS when we're rounding a corner, and subtract the centripetal force, leaving only the gravity vector.

Interesting... I wonder if there is a way to adjust for this on the settings?

 

[RTRyder]

Hi Guys, loving this gimbal already... this is the unedited (of course no post stab) straight out of the GP3 second and third test flights, first was still getting the power settings down so she had the shakes (thanks Ken for the power settings tip!). I have a camera operator (getting tilt happy though once we get this FPV for him he will not be guessing for framing the shot.

ok so:

1. Some micro vibrations are reaching the gimbal, I will be working to fix that, plus I was pushing the hex harder than I normally would for a shoot.
2. I’m getting the horizon shift, it becomes apparent on my yaw turn at 0:50 and more so at the 1:00 mark of the second flight.

So what do you guys thing is causing the horizon shift? aside from getting these sorted I think that brushless gimbals are the greatest thing since sliced bread!!!

Did you calibrate the accelerometer using a bubble level to make sure the platform was level on the roll axis when doing the simple mode calibration? If you did just the basic calibration you probably need to do it again and do the more thorough extended mode after the simple as its described in the manual though I have yet to fully decipher exactly what the author is trying to say here when stating "to make every of 5 faces “looking up” "...

"Extended mode (preferred): first of all, make calibration in simple mode. Then, rotate sensor board by 90 degree, to make every of 5 faces “looking up” (total 5 positions excluding “base”). Fix sensor in every position, press CALIBRATE_ACC and wait about 3 second. Repeat this step 5 times (the sequence of rotation is no matter). Remember, that calibration in simple mode will cancel results of extended calibration, so make it first.
Hint: Precise ACC calibration is very important to keep horizon stable in case of aggressive flight "

I think it means to rotate the entire multi and calibrate at all points of the compass, N,S,E, and W, or does it mean to do it horizontally then swing the platform 90 degrees so it's in a vertical plane...

I'm seeing more and more of this horizon tilting issue as more people use it. I have a theory based on my understanding of the AHRS in our system. Basically, when you turn those corners at speed, the accelerometers see the centripetal acceleration as a sideways gravity vector, so it thinks it's tilted, so it tries to compensate.

We solved this by using GPS velocity reporting to tell the AHRS when we're rounding a corner, and subtract the centripetal force, leaving only the gravity vector.

A accelerometer calibration makes a big difference in the horizon shift and I'd bet few people have done it. I had a bad case of horizon shift until I recalibrated then 98% of it went away. I think the remainder is exactly what you describe though I don't know what could be done about it without a GPS to get the data from. Given the nature of what we're trying to do here I don't know that it will be possible to totally eliminate horizon shift, just look at the problems all of the standalone IMUs have had trying to get rid of it, I don't know that any of them have come up with a complete and 100% working solution yet. After what I've been through with flight controller gimbal stabilization and independent IMU stabilization I have to say the extremely minor amount of shift I'm seeing now is trivial by comparison and much of it can be compensated for either by adapting the way you fly or in post, or both for that matter.

I'm going to try and get out and record some video with mine over the weekend and see what the results are with the system as-is before I try anything else.

Ken

 

[r0beert0]

Did you calibrate the accelerometer using a bubble level to make sure the platform was level on the roll axis when doing the simple mode calibration? If you did just the basic calibration you probably need to do it again and do the more thorough extended mode after the simple as its described in the manual though I have yet to fully decipher exactly what the author is trying to say here when stating "to make every of 5 faces “looking up” "...

"Extended mode (preferred): first of all, make calibration in simple mode. Then, rotate sensor board by 90 degree, to make every of 5 faces “looking up” (total 5 positions excluding “base”). Fix sensor in every position, press CALIBRATE_ACC and wait about 3 second. Repeat this step 5 times (the sequence of rotation is no matter). Remember, that calibration in simple mode will cancel results of extended calibration, so make it first.
Hint: Precise ACC calibration is very important to keep horizon stable in case of aggressive flight "

I think it means to rotate the entire multi and calibrate at all points of the compass, N,S,E, and W, or does it mean to do it horizontally then swing the platform 90 degrees so it's in a vertical plane...

Ken

Hi Ken, I did do the basic calibration with bubble level on roll as you mentioned.... I am confused also on the (preferred) method. gonna try the basic calibration again before I try this 5 face thing.

 

[RTRyder]

I found this and it was pretty much what I eventually came up with as the meaning for the "5 faces"... http://www.gcdataconcepts.com/calibration.html

Think of it as a cube, the sensor needs to be positioned as though it were sitting flat on each face of the cube with the bottom horizontal being the "base". After doing the initial base calibration with a bubble level, then do the remaining "5 faces" of the cube by moving the camera platform so it is vertical to the four points of the compass as well as fully upside down, doing a calibration at each point. Three of the five are easy, the two positions to the sides are more difficult because the platform can't rotate a full 90 degrees in that direction so you'll have to be a little creative to do those positions. I'm going to run through all that on mine and see how it does out in the wild tomorrow.

Ken

 

[GotHeliRC]

I found this and it was pretty much what I eventually came up with as the meaning for the "5 faces"... http://www.gcdataconcepts.com/calibration.html

Think of it as a cube, the sensor needs to be positioned as though it were sitting flat on each face of the cube with the bottom horizontal being the "base". After doing the initial base calibration with a bubble level, then do the remaining "5 faces" of the cube by moving the camera platform so it is vertical to the four points of the compass as well as fully upside down, doing a calibration at each point. Three of the five are easy, the two positions to the sides are more difficult because the platform can't rotate a full 90 degrees in that direction so you'll have to be a little creative to do those positions. I'm going to run through all that on mine and see how it does out in the wild tomorrow.

Ken

Yes it can rotate fully 360 degree each direction Ken, for the roll... undo the 6 screws on the aluminum tower... drop it one screw down... now you can rotate and do all your calibration then raise it back up.

 

[RTRyder]

Yes it can rotate fully 360 degree each direction Ken, for the roll... undo the 6 screws on the aluminum tower... drop it one screw down... now you can rotate and do all your calibration then raise it back up.

I just propped up the frame on one side while holding a level on the top frame of the camera platform, quicker and easier. Batteries charging now, I'm going to try for an early morning trip to anyplace that looks interesting and record some video. After I did the calibrations I powered it up with the camera on and swung it all around the room, still a little tiny bit of drift doing that but it's not like anything it's likely to see in flight (I hope!).

Ken

 

[Boltzman]

Thanks! No, I have not used the RC Timer long arms, though they caught my mind.
No there is no bearing block, just a piece of black MDF board betweeen the frame plates where the roll motor is directly hard mounted. No vibration dampers, nothing. I was pretty amazed that it worked so good. Though, naturally you need to have good motors, tried with some Turnigy 3536 910 kv motors first and got terrible results.

For me its to build the most portable drone that I can fit preassembled in a backpack. I travel to countries with extreme heat, humidity and there is no time to assemble each time. I also have a 80 cm foldable quad with AV130 mount - it took me around 20 min to assemble each time and the weight was 4,5 kg vs now around 2,5-3 kg. I got some considerations from others before building this front mount design, but it really works and is so amazingly ready to fly.

Balancing the drone is now much more tricky as you slide in the battery at the very far end.

I will try to go to 2814 motors and remove the extension arms, but people do not recommend it.

Beautiful pictures Boltzman and cool idea using the disco. Wonder if you used some of those rctimer 550 long arms, but then I guess it wouldn't be very packable.
How do you support the gimbal, is there a bearing block between the frame plates? I'm asking because I've got a large foldable quad that was designed to have a tilt only gimbal on the nose and this would be perfect.
Also, do you have vibe issues?

 

[Boltzman]

Hi Ken!

Agree with your statement. The arms are really weak but take the blast when you crash. I had a lot of considerations of using the DJI arms for my TBS Discovery brushless gimbal quad but apparently it worked out pretty good.

When you are at the field you need to have something easy to repair, you cannot start to drill your alumnium arms.

Have anybody seen alumnium based DJI arms?

Also, have anybody seen a similar design to the TBS discovery that is foldable and can handle a motor size of 2814 or 2820?

+1, I've got 4 FPV quads based on the F450 arms, two are standard F450 frames and the other two are TBS Discovery frames. Best part about them is they're cheap to fix, those weak arms usually break on impact dissipating most of the inertia that would otherwise break more expensive pieces like the Graupner props and at $4 each I can break all 4 arms and if none of the props break I figure it was worth the loss of the arms. Great for frames that will intentionally get abused and wind up taking some rough landings but less than ideal for carrying anything of value IMO.

I had one of the Discovery setup with Avroto motors but a smooth though moderately hard flat landing broke three of the 4 arms due to the extra weight of the Avroto, swapped the motors back to the standard DJI and it's survived harder landings without incident. Bigger is not better in this case...

Ken

 

[rocjock]

thanks for the info on precise calibration. it seems like you would need to have a level on both the roll and pitch axis at each position to be precise.

so after the base calibration for example, if you bring the camera frame forward 90 degrees, are you holding a level on the roll axis and calibrating, or are you leveling the pitch axis to ensure you went 90 degrees?

 

[vip]

maybe someone can make a short video.. this is getting confusing.