Replacement for the GS-1?

edited March 2014 in General

It's an interesting project. It doesn't have the slew rate of the brushless DC gimbals, but it's got the torque of a large heavy-duty metal geared servo. So if your suspension doesn't need the speed of a brushless gimbal or you don't want to have to deal with temperature dependencies and servo tuning, this may be a good alternative.

Some things that differentiate it from the GS-1:

It's continuous rotation, and can either be used in a way that drives to a particular location (traditional servo mode) or it can be used in pan mode (similar to a continuous-rotation pan servo for a KAP rig.) Apparently this is set up in software, so no hardware mods necessary.

It comes out of the box being able to rotate 360. So using one for the tilt axis of a KAP rig should let you tilt from horizon to horizon with zenith and nadir included. I know not everyone wants that, but it's kinda nice for vertical panos.

I've got a question in to the designer about maximum axial load and whether one can be used in the pan axis. (I want to see if I can safely move away from gearsets on the pan axis on my rig.)

You can power it directly from a 2S LiPo battery. I know this isn't of interest to everyone, but for people who are running LiPo batteries on their rigs, there's some attraction since it means there's no need for a voltage regulator. (But keep in mind that the cables and peanuts from James Gentles NEED to run below 6V or they burn! So careful running higher voltages on your rigs!)

Anyway, figured I'd share. The things look expensive, but for DSLR-sized rigs they may be cheaper than a comparable brushless gimbal so long as your suspension can take out all the high frequency / high angular velocity jitter.



  • It's galling to see a replica of Scott’s brilliant GS1 presented as a 'new solution'. Apart from the smoothing I get the same control over 2.4gHz RC using GS1 with out having to pay into a Russian start up! I have been looking at a much cheaper 'off the peg' ($20) replacement method:

    Tom, you fly an RC plane have you tried one of these?


  • I haven't, but from some of the comments I read on it about a year ago, more than one person said they've used one to stabilize a camera gimbal on their multirotor. I'd be willing to try it out and post a write-up.

    One thing I'm curious about on the OrangeRX flight stabilizer:

    I've got a geared pan axis. When I was using my Coolpix and later my A650, running with a bare servo on the pan axis was ok. The weight wasn't extreme, so the chance of the pan servo's output gear shaft failing or the case cracking was minimal. But I'm not comfortable hanging 1kg+ off of a servo shaft, even with a metal geared servo. Call me paranoid. One thing I like about the Brooxes Better Gear Guide is that the pan axle is captive. The top nut may fall off, but nothing else will cause the rig to fall out of the sky. So pre-flight safety checks are very simple: check the top nut, make sure everything is tight.

    But it makes gyro stabilization of the pan axis tough. My guess is the gear ratio I'm using means it simply won't have enough available velocity in that axis to keep up with rig motion. But I'm curious if the OrangeRX flight stabilizer has enough range in the gain for the pan axis to let me use it anyway, for whatever it's worth.

  • What happened to the GS1?
  • I think Scott stopped making it.

    Which is a serious bummer for all the reasons Bill mentioned. It was awesome when Scott first came out with it. Still just as awesome now.

  • edited March 2014

    I'm not sure gyro servo stabilisation of the pan axis is effective, I remember Scott looking at this and although he used a gyro servo on the pan he also devised a 'reaction fan' to use airflow to keep the pan heading controlled. Not a simple matter at all.

    The big win is in the tilt and HoVer (horizon levelling- a really nice feature of a' roll rig') axes. For both of these a transfer gear will take the load without grief.

    I only have the tilt axis managed by GS1 and its absolutely brilliant with a pendulum in the down wind plane. Most of the time its all I need. The worry is the tilt servo is the one that gets mashed if the worst happens.

    I think I’ll punt an Orange Rx 3S 3 too. I have one of their 'standard' Rx units as backup and, although its a bit scratchy at pairing, it is fine for what I need.

    An alternative would be one of the growing number of heli gyros such as looks nice at 7g.

    Let's see what the cheapie Orange can do.

  • The Indiegogo product uses two technologies that I have tried before: a magnetic encoder feedback and a gyro. Both are good ideas, but $179 is a little steep for something with about $40-$50 in parts (just an estimate on my part based on experience). I have been burned by Kickstarter projects that never materialized and Indiegogo has an even worse reputation for that. My approach now is to bypass the crowd-funding phase and wait for the product to mass-produced and sold by someplace with stock on hand. Usually, this means the price is lower, too.

    I would agree that gyro stabilization of the pan axis is problematic. The biggest problem is that moving the camera in the pan direction will twist the suspension due to the "softness" of the suspension and the gyro will have to sort out which panning motions are side-effects of intentional panning or gyro corrections and which are undesirable motion to be canceled out. This requires careful tuning of the gyro's response for each rig. In addition, the stiffness of the suspension changes as a function of line angle, making it even harder to tune. But roll and tilt both work well with gyro stabilization, but make sure the camera is well balanced around the pivot axes.

    The GS-1 was dropped because the volumes never got high enough to justify high volume production. I sold 1300 of them over about 2 1/2 years and I got tired of hand-building each one. BTW, most were sold as roll stabilizers for cameras mounted on motorcycles.
  • Scott, thanks for the info on pan stabilization. I'm curious, though, is that for pendulum suspensions as well as Picavets? I would've thought they'd be stiffer in pan, and less dependent on line angle. (Just wondering out loud.)

    That's wild that most of your GS-1 units went toward roll stabilizers for cameras on motorcycles! Funny thing is, I remember showing some of your videos to a mountain biker friend of mine for just that reason. But by the time he decided to do something about it you'd stopped making them.

    Yeah, I have no plans to be an early adopter of this new stabilized servo. When I find it in stock at a reputable supplier, count me in. 'Till then... er... no.

  • edited March 2014
    OK got Orange 3s. Turns out I ordered a stabiliser only unit, despite being badged 'Rx' it's not the combined Rx and flight stabiliser, no problem ( in fact its better because I can bypass it for the channels that don't need the gyro like the shutter and pan). I paired it up with an Orange DSM2 RX and the oddness began.

    The reaction of the servo in all cases is nothing like as fast as the GS1 after a lot of tweaking I got close, the goal is to have the nadir position maintained as the rig swings, but its no substitute. With practice I could find the frequency it liked to work at and a smooth motion was achieved. In practice such a frequency is not likely which leaves the movement 'notchy' at best.

    Fortunately I have a pal who will find the stabiliser useful for his glider but for a KAP rig it is no good.

    Cheap and cheerful it maybe but it's not even close to the GS1.

  • edited March 2014
    Thanks for the report, Bill! That's good to know. I might still pick one up for one of my planes, but it looks like it's not really suited to KAP.

    Actually, I did some poking around and found this:

    It's a new firmware flash for the OrangeRX stabilizer. It re-purposes a bunch of stuff and lets you tune on the fly using the AUX channel (no longer just on-off - makes it proportional). But the most interesting bit is that they optimized the interrupts to remove a lot of the servo jitter. From what you're saying about the speed that won't be enough to be useful. But it might get rid of some of the notchiness.


    P.S. I kept reading on that web page. Apparently the OrangeRX is an AVR ATMega running an Arduino bootloader, along with a 3-axis gyro, 3 potentiometers, 3 switches, and headers for servos. LOOOOTS of things that could be used for in the world of KAP, if you don't mind doing a little programming. First on my list: A radio pass-through that, with the flip of a switch on the transmitter, goes into autoKAP mode. Even better if it could be used with a receiver that can "safety" the AUX channel to a value that puts the rig into autoKAP mode. Bring a radio, you get RCKAP and autoKAP. Leave the transmitter at home and you get autoKAP. HMMM! (Looks like I might be getting one of these for KAP after all!)
  • Aaaight. I sprung for one. If I don't use it as a stabilizer for a rig or a plane, I'll write something to make it an RCKAP/autoKAP switch and post the code.

  • Wow! wonders will never cease: for a minimal outlay you could be onto a very handy thing indeed. Code is beyond me I'm afraid I can barely organise the required characters for English. I think you will find the correction it applies is a kind of spike or pulse like motion rather than an opposite balancing one. Do let us know how you get on these things are too cheap!
  • Oooooh! That may explain something, though! On an airplane, a sharp impulse on a control surface is sometimes handy to give it a bit of a "kick" to make it move. I'll take a look at that open source alternative firmware for the RX3S to see if that impulse is there, and what it looks like in the code. If it's present in the open source version, too, and was intentionally done, I should be able to intentionally remove it.

  • Video here:

    I think the GS1 cycles at 400Hz this thing is much slower.
  • AAAAAH! That makes sense. And it's also probably going to be tough to fix. Servo control signals typically run at 50Hz (though some will run at 100Hz). But without switching to some other control scheme, servos won't run at 400Hz. To do that you'd have to control the motor directly, as with the GS1.


    So one way to overcome this is to use a really long pendulum. Slow the period down to the point that the RX3S can more or less keep up, albeit with significant lag. But my guess is such a rig would become ungainly in short order.

    Anyway, mine's on order and should be here some time in the next year, depending on the speed of the boat they stick it on.

  • edited April 2014
  • edited March 2014
    This is the set up so far. I'm beginning to think gyro is operating in the reverse direction to that required for the required balancing motion- this makes sense as it's designed as a counter force for an aircraft control surface...
    ... it's surprisingly difficult to figure out what the thing is doing as I made the mistake of jerking the trimmers about before I realised it's trying to gobackwards!

    If you can stand a re-visit to the blog I have posted another vid of the movement:

    what larks eh?

  • Hahahaha! You're right! Oh man. It's a pity there's no way to reverse the correction. Maybe the open-source firmware has something for that (or at least a way to switch it in the code).

    But your video also shows how much lag there is in the corrective move. Probably fine for a control surface, but not so good for camera stabilization. It'll always be just a little too late.

    I'd be willing to bet some of that is because it's trying to average readings off the gyros to get rid of noise. Again, I'd like to see the open-source software for this to see how they do it. It might be possible to do some other filter (Kalman?) that won't result in so much lag.

  • I always wondered whether you could eventually get things even smoother by combining realtime-sensor data with an analysis of the last few pendulum swings to provide a predictive element to the control. The period of oscillation is system dependent and the velocity at the bottom of the swing gives you a prediction of the height and angle from vertical when it reverses. I guess this would be a sort of "model predictive control" and would require a lot more computing power and code. But maybe in 2020.

    I'll be interested to see how you guys re-purpose the Orange.

  • Tom, you're right - the pan stability of a rig will change with line angle, but that is only for a Picavet.

    However, a pendulum suspension will change its stability depending on the amount of pull on the line. When a camera is turned in pan with a pendulum, the bar attached to the line will put a bend in the line in reaction to that movement and the tighter the line, the more stiff it will be.
  • edited March 2014
    Good point, Scott.

    Man, it looks like all the simple problems were taken first. Now we're just left with the messy ones!


    P.S. Meh. They're all messy...
  • edited March 2014
    Here's a link to a thread on the "old discussion board" that discusses a Pan Hold function on a pendulum that I have implemented on my rig. (See Pan Hold comments towards the end of the thread)

    This does a pretty good job of compensating for unwanted pan motion around the kite line as the pendulum swings from side to side. It's a bit tricky because my pan gyro is part of a 3 axis gyro mounted on the side of the pan servo, so the point of reference moves when the pan servo commands a new position for the camera.

    I think an ideal configuration would be to mount the pan sensing gyro directly to the pendulum above the pan servo so that it does not move when the camera is panned to new positions. In this case, the only pan motion it senses is "unwanted" pan motion from swinging around the kite line. This configuration would be much easier to code but it requires a separate gyro just for pan.

  • edited March 2014
    FCB: I use a fairly short pendulum rig (18 inches) and have had good results synchronizing the shutter with the peak of the pendulum swing. This allows the pictures to be taken when the camera swing motion is at a minimum. I have had good success using SDM on an S100 and more recently with a Ricoh GR using a mechanical servo shutter control.

    (See previous thread post for more info on the rig).

    Ricoh GR Servo Control (See thread on 3/31/14)

    Pendulums can be fun to work with :)
  • FCB
    edited March 2014
    Very cool Mike. Impressive response time with the servo shutter. Actually your work is always impressive and nicely laid out.

    I used an accelerometer trigger on an old point and shoot SDM rig (subtracting current 3-axis acceleration from last acceleration to look for a time of zero tilting or motion) and it worked pretty well down to 1/60th of a second with normal (~ 1m) picavet motion. Since I switched to DSLR, I haven't tried to go back, but it might be useful for twilight shots (or rough winds like yours).
  • Bill, now that I've cleared my project plate a little I'm starting to take a look at the OrangeRX stabilizer.

    Some discussions on the Open Flight Stabilizer thread on RCGRoups indicate that what you're seeing on your rig is the expected behavior of the firmware that comes with that version of the device.

    Good news is that the Open Flight Stabilizer firmware fixes a lot of it. The jittery servo issue apparently was tied to the funky interrupt timing that was mentioned earlier. Also, the OFS firmware sets the center position of the pots as zero gain. By dialing the pot one way or the other you can tell it to correct in either direction.

    My programming adapter showed up yesterday, so today I installed the latest version of AVR Studio (not necessary, but it's what I'm used to), plugged the cable into my STK500, and verified that I could talk to the chip on the stabilizer. I'm a little bummed that I can't dump the firmware off of the chip to a .HEX file. I swear a previous version of AVR Studio let me do this. It could be that the STK500 only wants to go one way. No biggie. I have other programmers to try first. Anyway, I want to grab a copy of the original firmware binary before flashing the Open Flight Stabilizer firmware. As soon as I do I'll test it out on my rig and see what I get. I know it won't be as good as a GS-1 since the timing loop is too slow, but it might work to keep horizons relatively horizontal even in gusty wind. We'll see.

    Oh! One other thing: I took a look at the source for the Open Flight Stabilizer firmware. It's quite nicely written and commented. It would be a decent place to start on an autoKAP/RCKAP switch. I think there's even a couple of spare DIP switches that are unused with the OFS firmware. I might see about setting one as a toggle between flight stabilizer and KAP controller.

    Eeeeh... I'm getting ahead of myself. It's still World Wide KAP Week. I'm not going to mess with this any more until afterward. Right now I gotta fly.

  • @Benedict,
    That's encouraging news. I dug out the ORX yesterday and had a play (waiting for the Wx's doing its WWKW rain, rain, cloud rain thing all week again) I reckon the cycle speed will be ok for a pendulum, not so sure about Picavet. I hooked it up to James's cPp controller and it worked for a bit and then stalled. Glad to hear you have a path to its code now.

  • I ordered two more so I could leave one on the stock firmware and have two more to play with. I did manage to dump the original firmware into a .HEX file so I can reload it later. But this way I can swap them out in the field.

    I agree with you on the pendulum vs. Picavet. I expect it would have a hard time keeping up with all the directions a Picavet likes to move. I'm eager to see what it does with a pendulum. Even if it only really works for directly-down aerial mapping, that's still a handy thing to know and have available.


    P.S. Weather here has been spitting rain for the past couple of days. I'm bound and determined to get out at least once more for WWKW. If there's a lull in the weather this evening I've got a spot picked out in town. I just hope the wind cooperates!
  • edited June 2014
    @Tom Can you let me know which method is best to flash/ reflash the chip?

    When there is code to test I want to be ready!

    I have discovered the Orange 3s works really well if it has no input! If you power it on the Aux channel and just use the rudder output on the tilt axis it will react correctly to the pendulum swing! If you plug in an input from an Rx it reverses the motion. I don't think there can be much wrong with it's code: the issue seems to be in mixing the Rx signal with the default action.

    WWKW was a fail for me here, but no shame in that. Just didn't get the Wx and time to coincide...I got proof of concept for a KAP mapping project by Friday and then I got to the point where I resented the pressure... a sure sign the safer thing is to keep out of the sky!

  • I wound up hodgepodging a flashing setup using stuff I got when I was actively doing Atmel AVR development. If I was starting over I wouldn't build it this way, but this is what I used:

    Since I want to play around with writing KAP code for this thing I splurged and got this from Hobby King:

    ATMega Socket Firmware Flashing Tool

    With some practice you can press the tool down over an ATMega 168 / 328 / or whatever ATMega that shares the same form factor, and flash it. This thing is basically a cable adapter, so you still need a programmer. The ATMega 168P on the Orange 3S runs at 3.3v, so be sure to pick a programmer that'll operate at 3.3 and not just 5.0v.

    One caution with this adapter cable is that the other end is a 10-pin IDC connector. Most of the programmers these days come with 6-pin headers. Only six wires are actually connected on that tool from Hobby King, but they went with a 10-pin connector. ARGH! I've got an AVRISP, an AVR ICE, a Dragon, and an ancient STK-500. With the exception of the STK-500 none of them have 10-pin headers. You can build adapter cables, but come ON! I wound up using the STK-500 to program the Orange 3S.

    I think a good alternative to picking up an STK-500 would be this programmer from Sparkfun:

    Pocket AVR Programmer

    It even comes with the right header! (YAY!) And at $15US it won't break the bank.

    The last bit is the software. If you're just after flashing the chip something like AVRDude would be a good route to go. I think you can get it for Windows, Mac, and Linux. I've used it in Windows and Linux. Since I'm planning to write code for the thing I installed the latest version of AVR Studio. Overkill for flashing, but it's a nice IDE for writing software and debugging.

    As far as the actual procedure goes, this is a good guide:

    The one difference between what I did and what he did is how I got the board out of the case. He wound up cutting the case apart to access the AVR chip (which is on the bottom of the board). I found if you use smooth jawed needle nose pliers you can pull the board out of the bottom half of the case by gripping one of the pins on the servo header. It's a snug fit, but it's not glued in place or anything.

    Also, be careful getting the two halves of the case apart. I managed to snap one of the tabs off on mine while getting the board out. Irritating, but certainly not a show stopper. I'll be more careful with the other two.

Sign In or Register to comment.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

In this Discussion