I've been working on an easier-to-use version of the gyro stabilization system I posted about a few months ago. It turns out that there is just enough space in a standard servo to put all the gyro electronics and microcontroller inside a standard servo. Doing this creates a gyro-stabilized servo that just drops in and installs like any other servo.
The GS-1 starts with a commercial servo (a Futaba S3004 analog servo with ball bearings), but replaces the circuit board with a board of my own design. In addition to the gyro functions, the new circuit makes the servo into a digital servo.
This servo works best for stabilizing tilt and roll axes. I hope to have a solution for pan as well, but that is proving to be a tougher problem.
I expect that the main advantage of the GS-1 will be to keep the camera pointed in the same direction while the rig swings back and forth. This will help for both still images and for movies. But, like many new products, I expect to learn from users about what works the best.
Currently, I am looking for about 10 beta testers for this product. The beta testing process would involve buying the servo(s) and trying it out. Feedback from people (good or bad) will help me refine the hardware and software. There may be some back-and-forth shipping to upgrade the software during the testing phase, so I am only considering testers in the U.S. at this time.
You can visit my GS-1 web page to get more information about the GS-1 gyro servo.
That's pretty cool. I wasn't exactly sure how it was going to work, so thanks for posting the video, as that makes it crystal clear. How much did your customized rig weigh?
Wonderful development. I'm just about to finish Rig 3.0, but this is definitely on the shopping list for Rig 4.0. As I said before: dayum, I need this!
One question which I didn't see answered in you FAQ: you say it's a digital servo - will it work with a Dunecam system?
Wow! That makes it simpler. I think you said before that you thought the primary use would be for video. Do the small servo responses add any noise to still shots, or do you think it's just not as necessary for still shots?
I'm excited, but my Pan/Roll/Tilt rig is third or fourth in line in my current project list. How fast do you expect the beta testing phase to proceed?
Although the servo is digital, it interfaces to a receiver like an analog servo, so it works fine with DuneCam (as seen in the video). The only way you would know it's digital is that it "sings" when running because I update the motor drive 4000 times per second instead of 50 times per second as in an analog servo.
I originally intended the gyro stabilization for video stabilization and it does really help make KAP movies less nauseating to watch. But it also helps for still images since it keeps the camera pointed at the same subject as the rig swings. It may also help with low light/long exposure shots, but I don't have any proof of that yet.
So the roll servo doesn't really need a controller unless I plan to switch between horizontal and vertical orientations. Assuming I don't need that, what's the simplest way to create a circuit for this servo on the roll axis?
Scott, the roll servo does not need to be controlled remotely, but like any other servo it needs a signal to tell it how much to turn. In DuneCam, if you are not using the shutter servo channel, you can use that for roll by going into the Servo Test screen and setting up the Shutter channel to make the rig roll the proper amount. Otherwise, you can use a Y servo cable off of another servo output and connect a servo tester to that to generate the roll signal. A cheap servo tester is available from york-electronics.com. If you go that way, be sure to cut the PWM signal wire from the Y cable going to the servo tester so that the servo tester only sees the power and ground signals.
Linnar, There is one gyro needed for correction in the one axis that the servo controls. Each additional axis would have its own gyro servo and each would operate independently from the others.
MEMS gyros have finally gotten cheap enough to do this for a reasonable price. I expect piezo gyros to become extinct very quickly.
All servos (including the GS-1) idle under no load at a pretty low current level (a few mA). Servos draw more current when doing work. Under the heaviest load, the battery is connected to the motor almost 100% of the time. Since I use the same motor as in a S3003/4, that maximum current should be about the same.
BTW, no matter what kind of servo you use, the most important thing that you can do to reduce power consumption is to make your rig balanced. A servo motor constantly struggling to keep the camera's lens up can make your rig draw several times as much power as a balanced rig.
Edit: In use, gyro servos will take more power that a servo without a gyro just due to the fact that they constantly correct the position, as opposed to just slewing to the desired position and stopping. How much more power they consume due to this constant correction will depend on how much motion is in your rig to start with.
that's amazing and love that rig too! Can any bright spark figure out a way to incorporate this into a basic setup like mine (using Brooxes Autokap rig and one continuous rotation servo)at all ??
Mathew, I think this thread and the one Wicherd started after it might help. I think Peter posted a mockup of a Pan/Roll/Tilt rig with Brooxes parts in the latter.
thanks for the photo Peter it looks remarkably simple to do it - I think even I could do this! I don't seem to have the L-shaped bracket but I'm sure I can make one easily enough. Sorry for the stupid question but could this be powered from the 2 AAA battery pack that also power the rotation servo, and if so where do I plug this servo in? (I'm guessing a y-type cable is added to power both servos?) would love to see this working on the setup peter has shown if anyone has done this yet?
Depending on what you are trying to accomplish, you may not necessarily need roll stabilization to the same degree as you may need tilt stabilization. When using a Picavet suspension and the camera is pointing at right angles to the line, there is a limited amount of rotation in roll since the rig does not swing back and forth parallel to the line very much. When the camera is pointed parallel to the line, swinging in the roll axis makes the horizon not level, but this is creates less motion blur than uncompensated tilt.
If you do construct a PRT rig such as what Peter shows, I would suggest adding an additional metal axis for roll with gearing to the servo. Otherwise, the entire camera/tilt mechanism is supported from the plastic servo output shaft. This is a lot of weight and the center of mass is some distance from the servo, producing a lot of bending force on the servo shaft and servo mounts. Or you could make the roll axis from a metal shaft supported at each end by ball bearings, with the servo shaft coupled to this support shaft without it having to support the weight of the tilt cradle.
For the 3-axis rig shown in the video on my GS-1 page, I originally supported the tilt cradle directly from the shaft of a metal gear roll servo, but changed it to a .25" aluminum shaft supported by ball bearings when I noticed that the servo flexed too much when attempting to support my 125 gram camera.
What a clever idea ! I see some applications in robotics, too. I would love to have some more detail on the auto-calibration, gain tuning, and integration method (this is not a rate-gyro, but a "heading-lock" gyro, am I right ?)
No feedback yet. Maybe everyone is busy with their other projects. There are some new gyros coming out soon, so I hope to be able to evaluate those before committing to a production design.
I used this servo while shooting the Statue of Liberty and I'm very impressed.
I was only using it on the tilt servo, and most of my shots were perpendicular to the line. When using this servo on the tilt axis, it really shines when the camera is pointing to the left or right of the kite line. The picavet handles some of the leveling when the camera is pointing forward, but the servo helps there as well. It's amazing how easy it was to add this to my rig. Just replace the tilt servo on your rig with the GS-1 and you're ready to go.
For me the difference was that I was able to frame a shot faster and keep the subject in frame. With the standard servo the constant movement of the rig would make it difficult to know if the angle I was viewing in the DuneCam preview screen was an angle resulting from rig swing or from the tilt setting. With some trial an error I could usually figure it out, but the GS-1 made this much easier. It's difficult to say if the GS-1 reduced blurs or not. I think it can in some cases, but it is difficult to prove.
At some point I'll try it with the autoKAP controls as well. It may reduce blur and will probably keep the tilt angles more consistent.
The servo only misbehaved once. I grabbed the picavet lines half way between the clips and the cross. With this shorter swing, the movement caused by the correction would cause the camera to swing slightly in the other direction. This forced an opposing correction that would cause the camera to swing back to the other side. This would repeat over and over again until I let out more picavet line, turned the camera, or reached down to hold the camera steady. It only happened when the camera was pointing perpendicular to the kite line and when the picavet lines were choked short as I held them.
I too have finally had a chance to experiment a bit with the GS-1, but I used it on my AuRiCo /Canon570 rig, and my testing was all indoors.
The servo was really designed to be used on an r/c rig, not for autoKAP, but I thought I would try it anyway. I theorized that it should make shooting of panoramas much easier.
However... the AuRiCo sends no signal to the tilt servo in any of the modes where there are no tilt actions. Those modes, of course, would be where I would want to use it to take a horizontally-aimed panorama. Without that signal, there was nothing to drive the gyro correction of the servo, so there was no benefit. If Peter Engels can set up the AuRiCo to send a zero-tilt signal to the tilt servo during mode 0 or the two Turbo modes, I'm sure the GS-1 would be effective.
In all the other modes, the GS-1 did an admirable job of keeping the camera pointed as it was aimed by the AuRiCo tilt program, except when the standard position was at the end of the servo's travel (e.g. 90°). In fact, it did it so well it was, as I described it to Scott, unspectacular. It's smooth and quiet, and it doesn't call attention to itself. It wouldn't take long for the user to take it for granted, and miss it terribly when it's gone.
So as long as there's a signal going to the servo, the GS-1 is dandy. I plan to test it in the air with AuRiCo, and also with CAMremote, once I get better acquainted with that device, but I can't guess if it'll be a better match.
This thing will be a smash hit, IMO, and I'm honored to be included among the beta testers. I look forward to adding it to my Parts page.
I've done some indoor testing in an AutoKAP setup too. And like Broox I'm very happy so far.
I'm using two Gentles products to get the rig going. The Ricoh camera and the pan servo are driven by a Gent360-Ricoh. To get a steady servo signal to both the tilt and the roll servo I use a gent360-Pan (the anti creeper) in setup mode. On the output I have a Y-cable to connect the two GS-1s to the 1.5 msec signal produced by the gent360-Pan.
As soon as the weather gets better I''ll do some outdoor testing.
As for Brooks' comment about the GS-1 not doing anything when the servo control signal stops, my intent was to make it emulate a standard analog servo and that is what they do. However, they don't have a gyro, so I'm considering changing the software so that after seeing the first valid servo control pulse, it always actively drives the servo using the most recent control pulse (even if that pulse has since stopped). This is how most digital servos work anyway, so it's not something that is all that unusual.
As for the roll axis, many systems do not have an extra channel available to drive the roll axis with a constant pulse width. Maybe someone should make a little widget that has a T-connection from another servo (for power) and uses a small pot to control the pulse width of a roll servo output. Are you listening, James? :)
I had the same question as Brooxs....I am flying Auto KAP with the AuRiCo device that briefly sends a signal to move the servos (pan and tilt) in a set pattern / frequency but does not send a continuous signal to the servos....
Scott, I like your thinking with changing the software to always drive the servo using the most recent control pulse.
I am trying to picture how the servos would hold position as the AuRiCo device moves through the selected tilt pattern (e.g., 0, 22.5, 45, 90). .... maybe a bit of collaboration between Peter E. would help.....
The other question I have is battery life given the servos are "running all the time" to activey compenstate for the constant motion...what is the practical burn time for a typical KAP flight? Currently I am getting over 10 hours on 3 AAA batteries that drive both the servos and the AuRiCo device....suspect shorter flights / more battery power will be required....any field reports yet in this area?
You are right about gyro servos taking more power from the batteries. They don't draw more current, but they move more often, so the average power consumption will be higher. This was noticeable on my 3-axis stabilized rig (I needed a bigger battery), but I haven't noticed it yet on my rig with only tilt stabilization. It will depend a lot on how much the rig is swaying, so it's going to vary from one situation to another.
It may also depend on the tilt axis balance. If the servo has to struggle to move the camera up or down, the servo current can increase by several times when moving in that direction. And a gyro servo would make that extra current draw more noticeable.
Scott, I'm afraid that James is not listening -- he's just gone on holiday in the boonies for two weeks. But that would surely be a good idea, as is yours to tweak the software.
James is listening, at 56Kbs... As Peter has already identified that gent360-PAN emits a stream of 1.5mS pulses to keep a servo centered if it is powered up without an input signal. If I understand correctly you can use a Y-lead and the gent360-PAN to power the extra servo - just cut the signal wire on the Y lead between the Y and the gent360-PAN
James, to be effective as a driver for a roll channel, it would need a trim pot to adjust the pulse width over about a 1000~2000 us range. Is it able to do that as well?
No, James. The Y-cable is behind the gent360-PAN, at least in my setup. And in this setup the signal wire is needed to give pulses to both the tilt and the roll servo. The AuRiCo doesn't produce a servo signal constantly. So that wouldn't work.
Hy Scott I like a lot your design but I only work with micro servos on my <300g rigs. The one I use the most is the hitec hs-65hb, an analog servo with a lot of power and carbonite gears. I use it because it is easy to modify.
Do you think your board could fit in this servo. It could be outside the servo thought.
My original design was a separate board that was in-line with an unmodified servo. I changed to this design for simplicity. A standard-sized servo is needed to have enough room inside for all the parts. But you are right - the HS-65HB is a very nice servo. As is the HS-56HB, for right-angle mount applications.
Peter, The Y lead is just a method of providing power to the gent360-PAN? Scott, gent360-PAN only provides 1.5mS. Why would "roll" need a variable pulse length? Can't you just set the servo mechancially to set the servo in teh center.
Hi Scott, I have the Beta servos and like them on my helicopter. I set them up with one of Brooks mounts. One on tilt and one on roll. The roll I controlled with a gent360-pan since I ran out of channels and it does work, but I have to do some more testing. I think it slows down the reaction time of the servo so it's playing "catch up".
Here's a practice video I shot over a construction project. http://www.youtube.com/watch?v=_ZHio0wsr88
gent360-PAN when in calibrator mode sends a 1.5mS pulses every 20mS, which should be typical of most RC systems, may even be a little faster than some of the 8+ channel systems so I'm not sure why appears to react slowly?
I wondered if any of you have purchased this servo and integrated it into your rigs could post photos of it - (I'm keen to see how you've used it, if only to copy your setup and try to intgegrate into my basic autokap rig!)
I just read a very interesting idea by Scott Dunn wrt pan servo stabilisation. The idea, which has been expressed in that thread, is to use IC electronic compass to keep the directional orientation. As compass doesnt not drift it could maybe make panning stabilisation easier.
Besides 2 more ideas (mine)... It might be a good thing to use endless rotation (modified) servo for stabilised panning - that would solve the problem of movement limits. Well, DuneCam could then use that compass info for absolute positioning of endless rotation servos, getting the best of two words.
One of the problems with using magnetometers for heading sensing is that in some areas on earth (such as where I live) the magnetic inclination is close to vertical, meaning that a slight tilt of the rig towards north will result in a magnetic field that is parallel to the pan axis, making it hard to determine direction. Even when not tilted, there can be so little horizontal component to the earth's field that accuracy can be poor. This effect would probably be less of a problem in Europe than at similar latitudes in North America, due the north magnetic pole being located in Canada.
Just finished my second try at a stabilized Aurico rig using the GS-1. This time it's what Scott calls "PRT" -- the roll axis carries the tilt frame instead of the reverse. See the discussion here.
Haven't tried it in the air yet, but it works well in the shop.
and I'm very impressed.
