I have the servo driver (mx-04) connected to a servo. I know this servo controller is of limited capacity, but I was wondering if servos all have a kind-of "jerky movement at very slow speed or is this an issue caused by the limits of the controller? Such as; the pulse width or frequency? OR is there a different servo I should get? Maybe a slower servo?
Can we get a really smooth movement from a servo at slow… really slow… speed?
You probably need to better define the magnitude of the jerky behavior you are observing. I’ve tested a cheap servo and the smallest movements it can make are approx. .5 deg increments.
Servo jerkiness can be caused by a number of factors. A lot of time, it comes from overshoot of the commanded position, which can be limited in a number of ways:
Running the servo against some small amount of load can tend to reduce the overshoot, as it has to actually do a bit of work to get to the commanded position. Note that depending on how much and the method by which this load is imposed can have an effect - possibly detrimental - on your positioning accuracy and the ability to hold a position steadily. Also, too much load can make the servo work hard all the time, leading to premature failure.
Running the servo at a lower voltage means that it will be less “peppy” when trying to move it a commanded position, and will therefore be less likely to overshoot its target. You’ll lose a bit of torque, but if you’re only moving something light, this might not be a huge factor. Speed is also reduced, but you’ve already said that you’re trying to go slow.
Adding a reduction gear to the servo’s output will slow it down, possibly adding a little bit of “slop” into the system, so that twitchiness would also tend to be reduced. This approach would reduce the range of travel, unless you were willing to do major servo surgery, and move the feedback pot to the driven gear, instead of the servo’s output shaft. Adding a form of anti-backlash to the gear train (a light load, spring, or anti-backlash gears) would help to remove any induced slop.
There are other ways to achieve the same result, but these are the quickest and easiest options that come to mind offhand.
You might also find that a digital servo has a smoother response than an analog one. I recently used an HS-5645MG servo for the first time, and was amazed at how smooth the motion was, as compared the the bargain-basement analog servos I’ve always worked with.
Any incremental movement is not wanted. I am trying to obtain totally smooth movement. I am trying to build a pan/tilt head for a video camera and need fluid like movement. I don’t want to go to a gear drive if I can avoid it. I want to know if the Bot Broad and programing can give me what I want. (I do not have the Bot Board yet) I have the HS-645MG servos, High Torque, $40 each and the MX-04 servo controler.
I was looking at Futaba’s web site and they have 3 or 5 pole motors in some of there servos…this 5 pole motor may be what I need. BUT they are digital 
What does this mean? Can I still control them with a Bot Board and for now the mx-o4 servo controller?
I see someone else posted…
Can I control a digital servo?
From a servo controller’s standpoint, the number of poles on the motor probably won’t have an appreciable effect on the jerkiness, since any “cogging” of the motor’s poles is quite isolated from the output shaft through at least four stages of intermediate gearing. then again, for all I know, it might - I’ve never heard a decent explanation as to the benefit of a 5-pole motor over a 3-pole unit in a servo application, where the fractional rotation at the motor’s shaft yields a change at the output that is much smaller than the backlash in the geartrain driving it.
You’ve mentioned that you’re using the servos for a pan/tilt on a video camera. Some more information on the camera would be good, here. Is it one of those tiny “sugar-cube” cameras that’s less than an inch on a side? Is it an enclosed security cam that weighs several pounds? A production-quality camera? Are you using a wide-angle lens where small twitches won’t translate into noticeable changes on the screen, or a long lens where looking at it funny causes the image to shake visibly?
In practice, digital servos function pretty much the same as analog ones. They respond to the same pulse trains, and behave in the same way. They plug into the same ports, and are controlled by the same commands, and are, for all practical purposes, controlled in exactly the same way. (they’ll also hold a position after the pulse train is removed, so long as power is maintained, which is something that analog servos won’t do.) The main difference is that they can be programmed with the proper equipment, and - most likely the more important bit for your project - they have a higher update rate than analog servos. Because they check and correct their position more frequently, I find that they have a somewhat smoother response, and seem to be more solid than analog servos when it comes to holding a position.
If you’re looking for an absolutely smooth motion, I would tend to think that a high-reduction gearhead motor would be the way to go. You’ve said that you’re trying to avoid that approach, but that’s essentially what a servo is - a gearhead motor with a closed-loop control system built into it. It sounds like you’re really looking for a slow, precise pan/tilt platform with feedback, in order to provide absolute positioning. My thought would be to use small high-ratio gearhead motors controlled either by the guts of large-scale servos, or by a board specifically designed to provide servo-type functionality from the user’s own motor and feedback hardware. I believe Jim’s been working on a gearhead-based servo which would be just what you’re looking for.
“Any incremental movement is not wanted. I am trying to obtain totally smooth movement.” is somewhat fantasy thinking. There will be no totally smooth movement. You probably want movement increments that are small enought to be greater than the pixel surface in your cam chip. You probably need to go with stepper motors and high gear reduction to get the smoothness you want. You might be able to hack the guts of a printer that can get 2,400 descrete positions in the span of an inch. Also, have you tried controlling the servo with a servo controller instead of the servo tester? The tester itself is probably not high resolution and may be causing the jerkey motion you see.
Yes, I am still working with a servo TESTER.
So, I should invest in the Bot Broad/Atom 28 and get someone to help me with the program. I have an industrial joy stick. I should do this before I go out and buy digital servos???
I hear thats what your saying?
A little background on this is: I do a little video production and have invested in a 12’ boom for a video camera. The video camera is a small semi pro camera and light weight. You have to counterbalance the boom with weights, so what ever I build for the pan/tilt mechanism must be light. Whats available on the market is ether VERY expensive ($1500-$4000) see: varizoom.com/products/motion/vzmc100.html
The less expensive option is slow and limited movement see: 21best.com/21_best/electroni … html#mp101
The market has hole in it: nobody makes a motorized pan/tilt head like I am putting together. So far I am pretty pleased with what I have, see: i32.photobucket.com/albums/d36/r … G_3770.jpg
I suggest you get the ssc-32 servo controller and do some testing with a computer to see if servos will actually do what you want. I’ve made simple programs for using both an analog joystick and a gamepad with my servo base pan/tilt web cam. It was an interesting exercise. As your cam is light weight, it is much more succeptable to jerky movement than the heavy weight commercial types. If operating in the servo mode does not work, then the servos could be converted to simple gear head motors and other speed/direction controls possibly used.
You are saying to get the ssc-32 board to do some testing.
Will this broad later work in my application where there is no computer available? Or am I just getting the broad for testing.
The SSC-32 requires some sort of computer or microcontroller and software to send it commands, so it would not do anything useful by itself.
8-Dale
The ssc-32 would just be for testing using a computer. It might avoid getting into programming a bot board only to find out in the end that servos won’t do what you need. You are going to have to sort out issues such as what to do with a spring return joystick (if that is what you have) when you need to let go of the stick (remove the springs). I’m interested in seeing the bot board joystick programming myself to get around the $$$ for controllers like below. For the simple side, if you just use the servos as gearhead motors, a rheostat and two toggle switches from radio shack might get the job done.
From what I have been reading I don’t think these servos (HS-645MG) are capable of smooth movement. Someone said they where impressed with how smooth “digital” servos where. (mine are not)
I think the direction I need to be going is this:
-2 digital servos
-Bot broad
-Adam 28
-Analog joystick, spring loaded, 5V
-A program that overcomes the issue of the servos staying where I left them when I pushed the joystick, (no return to center)
-Someone to help me write that program. (i HAVENT A CLUE WHERE TO STATRT ON THAT ONE, except here on this forum,
What I have now has solved the issue of being able to point the camera in a desired position just not doing so while tape is rolling.
I should point out that my mention of the smoothness of the HS-5645MG servos’ response is based only on my limited use of them over the last couple of weeks - I have not done any head-to-head testing of them versus HS-645 in the same installation. They do have an extremely silky action over a broad range of motion, and I suspect that all other things being equal, they would provide a smoother response, but I have no actual proof of this.
As to the joystick control system, my initial thought would be to wire the pots of the stick as voltage dividers, as noted in someone’s previous post, and use the Atom’s A/D conversion to turn the analog values into numeric values that can be manipulated within the program.
One way to approach the positioning would be to have a variable “dead zone” at the center of the axes. Within this dead zone, the position sent to the servos does not change, but when an axis is moved beyond the dead zone limit, the commanded position is changed by an amount proportional to the distance from the center. That way, a small movement results in a slow motion of the servo, while cranking the stick all the way over to the side results in a fast transit of the servo’s output horn.
Alternately, and quite a bit more natural, from a video production standpoint, you could ditch the joystick idea, replacing it with a pan-tilt platform, with the motors replaced by potentiometers that feed the A/D inputs of the Atom. Then the servos on the camera head can simply be slaved to reproduce the motion of the pan-tilt controller. Mount a couple of handles to the pan-tilt controller, and place an LCD monitor on top of it, and it would be just like driving the camera normally, except you’d be doing it via remote control. This is the same sort of remote-control setup that they use for operating cameras in dangerous locations, like the ones right along the walls of racetracks.
The HS-5645MG appears to be not to expensive (ebay) and plenty of torque.
The camera controls you mention sound cool, a little overkill for what I need. Have you seen the pic at:
i32.photobucket.com/albums/d36/r … G_3770.jpg
It is a small camera and not much need for rear lens controls, the zoom and lenses are crap. I do have an infrared remote that will zoom and pause/start tape. I only have my thumb available for the pan/tilt controls, the rest of me is busy booming up or down. In the picture you can see an LCD screen in the BG.
So what about this programing? Can you tell me where to go to get the software? Can you email me the program when I am ready for the proram for the Bot Broad?
Just taking a close look at your pan/tilt construction, it appears that the pan portion is two metal surfaces, one sliding on top each other with out any bearings and such. This type of construction itself probably will result in jerky movement.
No, it has a Teflon like thrust bearing between the two metal surface’s.
It slides very freely. What really made the difference was finding the center of gravity/pivot points, thus creating little exertion on the servo motors. The original design was the video camera pan/tilt head that lynxmotion sells. That turned out to like sitting the camera on-top of springing diving board. Those metal brackets, lynxmotion sells, as strong as they appear, are very flexible. And the center of gravity was way to high, servos couldn’t even hold up the weight.