Improving the video performance of a PS1 mkTR Pro camera gimbal/mount...

[GGoodrum]

There's a couple things I like about the mkTR Pro mount, from PhotoShip One. The first is that overall, it is very light. The second is that it has a very effective vibration isolation scheme, which uses two frame-mounted platforms, that also serve as battery trays, that have these very squishy rubber doughnuts, which act as "cushions" for two carbon fiber tubes that go through rubber grommets in the landing gear/mount. I think this setup provides close to the same amount of vibration isolation as does the patented system on my CineStar 6.

There are a couple of problems, however, with using this mount for video, in a "stock" configuration. First of all, the tilt axis uses a direct-drive setup so the movements tend not to be very smooth. My first attempt to improve the smoothness was to replace the standard Hitec digital servo with a high speed Align DS610 servo, which uses a 12-bit A/D converter in the electronics. This did, in fact, improve the smoothness, but still not enough. There's still a very slight, but detectable "stair-stepness" in the video.Clearly, some reduction is needed in order to smooth out the movements. PS1 does offer a od to the tilt axis that implements a geared 2:1 reduction, but this really needs to include a hires, 12-bit servo to have adequate performance. Also, gears can add play, or backlash into the mix. The version I had, which granted, was one of the original prototypes, had a bit of play and there's no way to adjust it out. What I did instead was to first use a 12-bit, very fast HS-7966HB digital and then converted the tilt axis to a belt drive. I chose 2:1 as the ratio, figuring that this will be a good tradeoff of speed vs smooth movements. I can tell you, it works quite well, and is buttery smooth now. :02.47-tranquillity:

The roll axis normally uses a single "mini"-sized servo, with a mechanical reduction of around 2:1, using a special servo horn and a special link that attaches to the arch. Again, in its stock configuration, this is not all that smooth, even with the reduction. Part of the problem is that there is a bit of backlash/play in the linkage. PS1 does offer a "fix" for this backlash problem, and that is to use two servos, with two sets of linkages that have slightly different lengths set into the servo horns. The idea here is that the servos fight each other slightly, which removes the backlash/play, but I was not very happy with the smoothness of this operation. My first attempt at improving this was to use two Savox SH-1357 digital minis, in place of the Hitec minis. This took a bit of shoehorning, to get them to fit, because they are slightly larger than the Hitecs, and they have a different output spline. In any case, this still didn't have enough smoothness. Part of the reason is that the Savox minis are 10-bit, not a full 12-bit. More than that, though, I couldn't get the servo horns adjusted just right, to not cause a little jitter. My latest change was to also convert the roll axis to a belt drive, and use the same ultra-fast, 12-bit Hitec HS-7966HB digital. This is a ton smoother, now just as "buttery" as the tilt axis.

Finally, what I came up with, mainly as a workaround for my issues with the wandering/drifting WK-M gimbal outputs, was the idea to use an XA FC and AHRS setup, in addition to the WK-M, but only to drive the gimbal servos. This works quite well, actually. Using the v1.31 firmware, it allows you to change the servo update rate. For some reason, they made this a fixed setting of 10ms/100Hz on the v1.33/v1.34 update, but v1.31 lets you change this, all the way down to 2ms/500Hz. There is a very noticeable difference between the two, for sure, especially for small movements. Actually, I'm guessing that the WK-M's output update rate is probably close to 10ms/100Hz as well, because it seems the same as the XA with v1.33/v1.34.

One other change I'm trying out is to move the XA's AHRS to the top of the mount, instead of being part of the platform. The idea here is that with the AHRS on the non-isolated side, it will be generating "corrections" to the gimbal servos, due to vibrations. With the gimbal on the other side of the isolation, these corrections will end up affecting the video, to some degree. By having the AHRS on the same side, it should theoretically work better. We'll see.

Anyway, this is now mounted back on my X8. It is too late today, but I will try and get some test flights in tomorrow.

-- Gary

 

[GGoodrum]

I've done some testing, with mixed results. Actually, the modified mount worked quite well. This X8 platform, however, has some major vibration issues. Even with the isolation on the mount, pretty severe vibrations are still getting through to the camera. I could see the green and pink pool noodles, on the landing gear, vibrating at one point, and they are on the isolated side of the mount. The vibrations were so bad that it was causing a bit of a wobble, that I had a hard time dialing out. Actually, the video came out better than I expected. You can definitely see it in the video, but I think a combination of the mounts isolation and having the XA AHRS on the stabilized side of the mount, helped quite a bit. It should have been much worse. In any case, the mount itself worked quite well. You can see the platform movements, which are pretty severe, and the camera stays put. I need to dial in the roll gimbal gains just a bit, but it is pretty close. The tilt axis is good.

I just finished swapping out all the APC 12x3.8 and 11x4.7 props for 11x5 Graupners. While doing that, I think I found one thing that was a big contributor to the vibrations. It turns out I must not have used Loctite on a couple of the arm cross support screws, because I found two supports that were missing both screws. I've replaced those, used Loctite this time, and checked all the rest. I also added one more support on each arm. This should stiffen things up a bit, and hopefully that, combined with switching to the Graupners, will cut the vibrations down to something manageable.

I'm hoping these fixes/changes will make this a viable video platform for my GoPro 2. If I can't get the vibes under control, I will move the motors, ESCs and the gimbal/mount to some other platform, like a mid-size hexa of some sort. In any case, I'm going to be finishing up my CS 6 build this week, so I'll likely be moving the WK-M and XA gear over to it. If the X8 is "fixed", I'll probably leave this setup alone, and put something else on the CS 6.

-- Gary

 

[GGoodrum]

Not sure anyone is paying attention to this, but in the odd chance that somebody is, here's another update.

I found a major source of the bad vibes I was getting. It turns out I had a bent prop adapter, from my very first flight with the WK-M, back when I was using the PM cable DJI provided that had "Cheans" plugs. These didn't work well with my regular Deans and I had a complete loss of power that caused the X8 to drop out of the air, from about 8 feet. As it fell, it sort of rolled over and one prop/motor hit the concrete.. It didn't look like there was any damage, and but I replaced the prop anyway. What I didn't realize is that aluminum prop adapter apparently took the brunt of the force from the drop, and it bent slightly. Anyway, I noticed it yesterday, after I switched to the Graupner 11x5s. I replaced it today, and did another test flight. Oh, and I also adjusted the roll gain so the mount is really dialed in now.

I still need to get the WK-M gains sorted out, as I also changed to v4.22. I had to dial them down, from the 130% I had before with v4.2, but I need to fine-tune these still, because I'm getting quite a bit of wobble, especially when descending. The vibrations are drastically reduced, but I'm still getting a lot of servo "chatter", which you can hear in the video, and which causes a lot of jelly roll. Again, the mount is working really well. I was purposely yanking this all over the place, which you can see in the video, and the camera stays level. Here's the latest video:

I'm thinking that maybe putting the XA AHRS unit on the mount, instead of on the frame, with the rest of the electronics, may be actually making things worse. My next test will be to move it back to the top, and see if that makes any difference.

-- Gary

 

[jffry7]

Gary, I know you are more experience than me but have you tried raising the gain. In my x650 V8 I thought dialing down the gains would lessen the wobble but it made it even worst specially on decent so I raised it from 135% to 210% (using X2 gain adjust it was actually 215 but reduced it). I find it the combo 210% basic and 70% atti was good enough for the quad then I started to sort out the gimbal gain. Tho I know I read somewhere using the XA elect jsut for gimbal stab is a viable option.

 

[GGoodrum]

Gary, I know you are more experience than me but have you tried raising the gain. In my x650 V8 I thought dialing down the gains would lessen the wobble but it made it even worst specially on decent so I raised it from 135% to 210% (using X2 gain adjust it was actually 215 but reduced it). I find it the combo 210% basic and 70% atti was good enough for the quad then I started to sort out the gimbal gain. Tho I know I read somewhere using the XA elect jsut for gimbal stab is a viable option.

Yes, I will give that a try, thanks.

-- Gary

 

[mbsteed]

This is great information and an excellent mod. Could you give exact details on the parts you used and the sources you obtained those from. A video tutorial of the mod would be useful.

I have used moon gel and it takes out some vibration perhaps not all. Put it between the camera and the mount. Have you balanced the props?

 

[GGoodrum]

I used two full-size Hitec HS-7966HB 12-bit digital servos, with 32t MXL drive pulleys from Lynxmotion.com. The 64t large pulleys came from Andrey Kim, but I'm not sure what his source is. I also got some 60t versions from McMaster.com, along with the MXL belts.

Although these mods made the mount work well, I'm afraid there just isn't any hope for eliminating the low frequency vibrations from the VC-600-X8 frameset, so yesterday I "retired" it, and moved everything to an XA hexa frame. I'm also going to try one of CNCHelicopter's MC6500PRO 2-axis gimbal, as I've seen several of these that seem to produce good video. I just want to have a good, stable platform to use with my GoPro2, and be done with it, so I can get back to working on my CineStar 6 build. Anyway, I'm about to start a new thread on the XA hexa.

-- Gary

 

[mbsteed]

Thanks for the info - there is a thread on the RCgroups where a fellow uses wood frames and has no vibration. It just blew me away, he sat his GoPro on top of the wood frame with no dampening material, no gimbal, and no software stabilization and got perfectly smooth video. I wonder if the pine booms provided a natural dampening that took out all the vibrations. The booms were 3/4"X3/4: pine hot glued together in a T configuration. It was interesting. Perhaps in all of our efforts to go with hitech solutions we may be missing a simple but more effective solution.

 

[GGoodrum]

Thanks for the info - there is a thread on the RCgroups where a fellow uses wood frames and has no vibration. It just blew me away, he sat his GoPro on top of the wood frame with no dampening material, no gimbal, and no software stabilization and got perfectly smooth video. I wonder if the pine booms provided a natural dampening that took out all the vibrations. The booms were 3/4"X3/4: pine hot glued together in a T configuration. It was interesting. Perhaps in all of our efforts to go with hitech solutions we may be missing a simple but more effective solution.

I think I read that thread. The one thing that stuck was the example of why aluminum baseball bats have rubber handles, but wooden ones do not. Makes sense.

-- Gary