Jump to content
 

Twitching Servos


bigal10

Recommended Posts

I have a number of points that are operated by servo, mostly HK white cheapies, both analog and digital, and the same thing occurs.

At standstill they twitch a bit, not enough to move the blade, but enough to annoy me!

I have 3 different types of accessory decoder, ESU SwitchPilot Servo, MegaPoints, and GFB designs' IMP.

Only the GFB ones don't twitch!

 

In spite of setting the CV's in SWPServo to discourage the twitch, they still do at rest.

And it only occurs on points, which are all Peco 100. (ie: not on my home-made uncoupling ramps)

Could it be that the servos are battling against the spring? Should I remove the springs, I know they aren't necessary when using servos?

Or, have I set the arc of the servo throw too short for the internal electronics to settle at the end-stop?

 

Any thoughts?

 

Regards,

Alan.

Link to post
Share on other sites

It is very difficult to adjust a servos endpoint to be absolutely perfect.

 

Once the points have moved the spring will be reinforcing the position so the servo will not be fighting against it unless the end point is set too short. In this situation I would use a linkage that allows the spring to do the final positioning, and provide the holding force, and only use the servo to get past the centre.

 

The other arrangement is a flexible linkage or an omega loop to take up the slack and set the end point just slightly past perfect. The servo will, however, be pushing against it and this can result in excessive power dissipation in the 5V supply to the servos.

 

The best solution is to turn off the control pulses after each movement. This turns off the servo but the friction in the gearing holds the position. Maybe this is what the GFB controller does.

 

Servo "twitch" is usually used to describe unwanted movement when the system is turned on, and has it's own set of solutions.

Link to post
Share on other sites

Modern servos tend to have digital processing in them, and some of the cheaper ones appear to only have a few steps, so if you are sending them a pulse length which is between two steps they will judder between the two steps, adjusting the position slightly will often shut them up. If you can find an old analogue servo and it does not do this you have found the problem.

 

Excessive load on the servo can be a problem too. If there is load the servo arm will be pushed back until the feedback circuitry within the servo detects it is out of position when the motor will activate to reposition the servo. You really need to remove any springs and create a springy linkage and make sure that the servo endpoints are adjusted to match the exact throw required.

 

Some controllers do judder a bit.

 

On controllers that switch of the pulses to get round this problem you have to watch out for the servo wandering out of position because the gearing on modern servos can be very free and they do not have enough friction to hold position against a strong spring - this method is really a bit of a bodge and you can come unstuck.

 

It could be a lot of things!

Link to post
Share on other sites

  • RMweb Premium

On controllers that switch of the pulses to get round this problem you have to watch out for the servo wandering out of position because the gearing on modern servos can be very free and they do not have enough friction to hold position against a strong spring - this method is really a bit of a bodge and you can come unstuck.

But with Peco points the spring holds the points in the desired position, the servo is not fighting the spring, and the spring will not be pushing the servo back. So don't remove the Peco springs, they help. With analogus servos having the controller turn the pulses off works very well. With some digital servos the servo remembers its set position so turning the pulses off does not help, best to avoid these for point drives.

Regards

Link to post
Share on other sites

Yes, there are two types of twitch that I have observed:-

 

  1. The first is twitch from external interference such as current spikes in adjacent wiring. These are usually wild twitches that occur infrequently and can usually be related to some activity that causes the current spike (operating a point, running a train in a certain area, when a short happens somewhere).

    Modern servos with digital signal processing should be immune to this and will not respond to a single pulse that is wildly different to the applied pulse stream, but it is still sometimes seen. Old analogue servos are very prone to this.
     
  2. The second type of twitch is a very small movement of typically one 'step' and it usually appears that the servos are hissing or randomly buzzing depending on how fast the motors are. The movement is typically too small to be seen (unless you can see the gears turning) and does not result in the operated device moving at all because the movement is normally taken up in the linkage. The sound is very irritating!

I think that the OP is referring to the second type. It mostly occurs within the servos and then only in modern digital ones, but controllers like the ESU Switchpilot Servo do it at times too, I cannot comment on the ones listed because I have not played with them. Fortunately on the Switchpilot it ceases as soon as you send it a command!

Link to post
Share on other sites

  • RMweb Premium

In my experience servo twitching/chatter can be cause by any number of outside influences, and digital servos are no better - actually worse in my case - than analogue ones. After trying analogue ones that twitched repeatedly on a random basis I tried digital versions which I found would twitch/chatter constantly as soon as my DCC system command station was switched on in the same room, this without it being connected to anything, and with the servo(s) powered off a battery pack in a test bed setup, so no electrical connections whatsoever.

 

These problems are I have since discovered - I should have done a more thorough web search before inversting time/money/effort in servo control - quite common in radio model control cars/aeroplanes/robotics etc, where generally only small numbers of servos are used unlike with model railway use, and with various solutions devised to try and overcome them.

 

EMI interference is I believe the main causal problem, and so the nature of the electrical signals that the servos use/need can be distrupted simply by the wiring being alongside other wiring let alone other electrical items being used. The idea of using R/C servos is great in theory, but thus falls down in the practical application for some like me, while other would appear to have no problems.

 

Not being interested in electrics/electronics past what I need it to do for me, I threw in the towel and went back to using more conventional electrical control which doesn't suffer from such issues and is thus more consistent and reliable.

 

Izzy

Link to post
Share on other sites

  • RMweb Gold

When I built the level crossing on Ravens I used cheap servos and found them to twitch as described. Upgrading to more expensive servos cured the problem.

 

I've got one of the original cheap ones on the workbench now connected to an Arduino project, it will twitch at certain end positions with nothing attached to the horns.

 

Andi

Link to post
Share on other sites

  • 2 weeks later...
  • RMweb Gold

WIth cheap servos, I've seen it suggested by arduino users that if the servos are only used "intermittently" (ie for points, signals, uncouplers, level crossing gates) then to stop movement jitter and the noise they make, a workaround is to switch off the power to the servos (relay/power transistor switched through an output port) and only activate them when you want to move them.  Its probably cheaper than buying good servos!

 

Of course, if you're not using arduinos, then its not an option and you either have to get earplugs or bite the bullet and replace the cheap servos.

Link to post
Share on other sites

  • 3 weeks later...

Hi bigal10,......... all of my points(x17) are fitted with gfb designs imps and new power xl-09 servos, the points have all the springs removed, 2 twitched.. one faulty which was replaced and one cured by re-alignment... highly recommend the gfb imps.... :yes: 

Link to post
Share on other sites

Hi Dave,

I'm already a GFB fan, (as Fraser is a near neighbour of mine) and have started to remove springs on my points, that does seem to help overcome some of the twitching, and also the 'thump' as they throw, so you get a nice slow movement.

By trial and error, I've found that the cheapo digital version (rather than analog) of HobbyKing servos work best and without twitch on both IMP and SwitchPilot Servo2.

Unfortunately I cannot say the same for MegaPoints, which is a shame, maybe Dave will look at fixing this?

 

Regards,

 

Alan.

Link to post
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...