Twitchy servos

User avatar
David B
Posts: 1509
Joined: Sun Jul 20, 2008 5:30 pm

Twitchy servos

Postby David B » Tue Sep 25, 2012 7:58 pm

I am attracted to using servos for various things on a layout, but this line of thought received something of a set back a short while ago.

I was watching a layout that used servos for changing a turnout as well as operating signals. There were a couple of locos which, when they went near the turnout and signal, caused said items to a frenzy akin to an attack of the DTs. Other locos on the layout had no effect on the servos.

What is it that actually causes this twitching? I remember from many years ago, loco motors having some sort of suppressor attached to stop them interfering with the television. Is this a similar phenomenon and if so, how does one overcome it? Is the loco/motor affecting the servo or the driver? If it is the driver, should it be placed some distance from the servo and perhaps in a Faraday cage?

A friend had a problem with twitching servos a couple of years ago which he cured by buying more robust (and expensive) servos.

Lots of questions - can anyone explain the cause and perhaps prevention of, this phenomenon? Please don't get technical. Simple language will be appreciated.

David

hughesp87
Posts: 137
Joined: Tue Aug 05, 2008 8:25 pm

Re: Twitchy servos

Postby hughesp87 » Tue Sep 25, 2012 9:16 pm

David,

Greetings from another electronic ignoramus who is having exactly the same problem. For me the problem was all the more acute because it initially presented itself in the middle of the first exhibition after fitting, and nothing seemed to indicate what the cause might be.

However, talking to a friend who knows much more about this than me (and often confounds me with techy responses to simple questions), the problem might arise because some of the electronics boards that drive servos maintain a current through the device at all times, and it may be that there is interference between this and a passing loco, or in my case the operation of an AJ magnet or Tortoise motor.

The answer appears to be to select your drive electronics carefully, because there are systems that operate the servo and then shut off all power to the circuit. The suggestion is that replacing the board under the layout might solve the problem - we'll see! The downside of this is that it apparently makes the nice bounce that I had achieved with the old system a little harder to set up.

Perhaps my friend (DB) would care to elaborate through this thread for those with a higher level of knowledge in such matters?

Hope you manage to sort the problem out.

Regards,

Geraint Hughes
Geraint Hughes
Cromford & High Peak in P4
Danish Railways in P87

User avatar
David B
Posts: 1509
Joined: Sun Jul 20, 2008 5:30 pm

Re: Twitchy servos

Postby David B » Wed Sep 26, 2012 11:48 am

I should have mentioned that the driver board is the one that comes from MERG, capable of dealing with up to 4 servos.

David

nigelcliffe
Posts: 747
Joined: Mon Jul 21, 2008 8:31 am

Re: Twitchy servos

Postby nigelcliffe » Wed Sep 26, 2012 2:11 pm

Dave,
if you are using the MERG Servo-4, then the processor (PIC) on the board can be upgraded to have the latest bouncing "Sema4" software, or the non-bouncing "Servo4" software. Both of these have the automatic power-shut-off after a few seconds. No need to build a new board to replace the old, or even buy new parts, just carefully lift the PIC out of the board for reprogramming. Then, re-insert the PIC and reset the distances/speeds for your servos. You will need a PIC programmer (eg. PicKit-2) to do this, or ask a MERG member with a PIC programmer to do the job for you; many will do it for just the cost of return postage for the PICs.
The revised Servo-4 board can be selected or de-selected the power-off capability by bridging a couple of pins with a little plug, plus a few other additional features.


The general case problem is probably interference, perhaps with cable runs too long, or too close to each other. Possibly also dips in power supply to boards. It gets horribly complicated tracking down electrical interference.

- Nigel

Chris Mitton
Posts: 258
Joined: Sun Aug 31, 2008 1:18 pm

Re: Twitchy servos

Postby Chris Mitton » Wed Sep 26, 2012 2:34 pm

hughesp87 wrote: For me the problem was all the more acute because it initially presented itself in the middle of the first exhibition after fitting, and nothing seemed to indicate what the cause might be.



I well remember a time in the early "noughties" when I was heavily involved in IT in the educational world and a number of manufacturers were introducing a new whizz-bang techhnology called Wi-Fi. They all spent fortunes developing new kit, they all worked perfectly in their research laboratories, so they all spent even bigger sums going into volume production. They all then blew huge PR budgets trumpeting big launches at a major educational exhibition.....and they all took them to Olympia and they all interfered with each other so nothing worked properly and they all spent even more PR budget wiping the egg off their faces!

Moral - much electronic technology depends on what else is in the vicinity, which might not be under your control.

Regards
Chris

User avatar
David B
Posts: 1509
Joined: Sun Jul 20, 2008 5:30 pm

Re: Twitchy servos

Postby David B » Wed Sep 26, 2012 2:55 pm

Thanks, Nigel. I have not done any installing yet, but would like to know what it is that causes the phenomenon so that ways to eliminate or minimize it can be incorporated.

I have seen other examples and heard of others, so it would seem this 'twitch' is something that could benefit from the application of brain power. As it stands, electronic ignoramuses like me & Geraint can merrily incorporate servos, get this problem and be stuffed.

Chris Mitton wrote:Moral - much electronic technology depends on what else is in the vicinity, which might not be under your control.


As the twitch happens, apparently haphazardly, does anyone have any answers or suggestions that will make servo use more reliable? In view of what Chris suggests, why are only servos affected and not locos, relays or other electronic bits? I have not heard of locos being driven by poltergeists yet, even on a DCC layout!

David

User avatar
Russ Elliott
Posts: 930
Joined: Thu Jun 02, 2011 6:38 pm

Re: Twitchy servos

Postby Russ Elliott » Wed Sep 26, 2012 3:09 pm

NLG had an excellent presentation on servos from David Barham on Monday night, and to be fair to David, he did not shirk the issue of servo twitching. As a bit of homework, I've been reading through some disparate RMweb threads on servos, and it would seem the cause(s) of twitching are or could be:

- Use of cheap servos, the more expensive ones being reported as being more immune to twitching.

- Extraneous pulses being picked up from or induced by other circuits. Since most servos get their control pulses every approx 20ms, this is about 50Hz. There could be plenty of 50Hz inductions or crosstalks in layouts from DCC, lights or transformers or any motors (locos, turnouts) doing in the region of 3krpm. A suggested solution is to turn off the servo power when not being required to operate. This however invites the next potential problem.

- On startup, servos receiving their power before their control signal, thereby making them unable to decide where they are, where they should be, which direction they should be going in, and how fast they should go there. I understand the software in later MERG units addresses this issue. An alternative simple solution being reported is to use a pullup resistor on the control wire, which holds it in a known state while the servo decides what to do.

User avatar
grovenor-2685
Forum Team
Posts: 3918
Joined: Sun Jun 29, 2008 8:02 pm

Re: Twitchy servos

Postby grovenor-2685 » Wed Sep 26, 2012 4:14 pm

In MERG we have sold a lot of Servo4 boards and a lot of cheap servos. Most customers keep coming back for more. A while ago we have reports of twitching at power on. This was looked at and a number of teweaks made to the boards and software to greatly reduce the twitching.
As this occured only on power up it was probably unnoticed or ignored by many users. Recently we have had a few reports of interference during use similar to that at the start of this topic. This is being actively discussed on the MERG forum but is not proving easy to replicate and hence solve.
Solutions involving use of expensive servos or modifying the actual servos are not helpful as the whole object is to use cheap off the shelf servos instead of expensive point motors etc.
The problems tend to be reported in the form of "my friend has this problem" so the amount of data available to replicate is minimal.
We would selcome hard information on the circumstances when the interference is seen. I would request any non-MERG members who have obtained MERG kits via a friend to report any problems back to MERG via the friend with as much data as possible.
It will all help in finding a solution.
Regards
Keith
Regards
Keith
Grovenor Sidings

User avatar
Ian@Exton
Posts: 88
Joined: Sun Feb 20, 2011 10:12 pm

Re: Twitchy servos

Postby Ian@Exton » Wed Sep 26, 2012 6:31 pm

In response to the request for hard information, one of the locos which caused the "servo twitching" on the layout DavidB refers to at the start of this thread is a pannier fitted with a Mashima 1220 motor, Branchlines 53:1 two stage gearbox, and a 6mmx12mm flywheel. It is not DCC fitted. All in all a pretty conventional loco.

User avatar
grovenor-2685
Forum Team
Posts: 3918
Joined: Sun Jun 29, 2008 8:02 pm

Re: Twitchy servos

Postby grovenor-2685 » Wed Sep 26, 2012 6:49 pm

That's a bit different from other reports which have indicated that older locos were causing the symptoms. Is there any apparent difference between this loco and others which do not have a problem, eg with or without suppressor capacitors.
How does the wiring between servo4 and servo relate to the track feed wiring?
Is there any common power supply?
If you watch them in the dark can you see any sparking from the wheel/rail interface or pickups, and does this differ between good and bad locos?
Does the twitching occur with the points in either position or just one?
Regards
Keith
Regards
Keith
Grovenor Sidings

User avatar
Ian@Exton
Posts: 88
Joined: Sun Feb 20, 2011 10:12 pm

Re: Twitchy servos

Postby Ian@Exton » Wed Sep 26, 2012 9:07 pm

That is a lot of questions, which will need a bit of discussion to answer.
The locos only seemed to make the servo twitch when it was in one position, and didn't affect it in the other.
Ian

User avatar
David B
Posts: 1509
Joined: Sun Jul 20, 2008 5:30 pm

Re: Twitchy servos

Postby David B » Wed Sep 26, 2012 9:40 pm

Thank you, Keith. It is good to know that those who understand about the electrical side of things are looking in to this phenomenon. Are there any early suggestions of cause or solution?

I encourage people who have experienced 'twitch' caused by locos to supply as much detailed information as they can, either here or to MERG. I am endeavouring to get the relevant information for the incident I witnessed.

David

User avatar
grovenor-2685
Forum Team
Posts: 3918
Joined: Sun Jun 29, 2008 8:02 pm

Re: Twitchy servos

Postby grovenor-2685 » Wed Sep 26, 2012 10:30 pm

There are several postulations, but not yet enough data to reach any meaningful conclusion, I look forward to gathering more.
Regards
Keith
Regards
Keith
Grovenor Sidings

User avatar
Will L
Posts: 2516
Joined: Sun Jul 20, 2008 3:54 pm

Re: Twitchy servos

Postby Will L » Thu Sep 27, 2012 1:32 pm

IANATEXTON wrote:That is a lot of questions, which will need a bit of discussion to answer.


That is the problem, without a proper understanding of the circumstance, and probably a reliably repeatable example, there is little chance of finding the cause. What "they" will need is somebody with a repeatable problem who can demonstrate it in person to somebody like Keith so that they can start to track down causes. Hopeful when that has been done and an underlying cause found, then a simple solution that the rest of us can understand and implement will become available.

I know we have seen an associated event on Knutsford, not on a signal, but on a servo driven working gate on our the recently introduced MOD siding. I will check with our "expert" and see if we have found a solution yet.

Will

User avatar
grovenor-2685
Forum Team
Posts: 3918
Joined: Sun Jun 29, 2008 8:02 pm

Re: Twitchy servos

Postby grovenor-2685 » Wed Oct 10, 2012 3:30 pm

I don't know if this will help but at least its a suggestion from Gordon Hopkins of MERG.
If anyone coming to the Missenden event this weekend has a loco which is particularly prone to causing servo interference, perhaps they could bring it along so that some experiments may be undertaken and potential solutions tried out whilst we have all the appropriate kit and people there.

Anyone suffering from this twitching problem and has the opportunity to take it to Missendon, please do so.
Regards
Keith
Regards
Keith
Grovenor Sidings

User avatar
Jol Wilkinson
Posts: 1113
Joined: Mon Jul 21, 2008 7:39 pm

Re: Twitchy servos

Postby Jol Wilkinson » Wed Dec 12, 2012 4:00 pm

Keith,

I have just read this thread with great interest as I experienced this problem "big time" when I set up London Road recently in the local village hall.

I have just set up the new board which has the most servo driven signals (seven) on it ( the station approach) in the workshop. These are drive by two MERG servo driver pcbs, one earleir version with a SRV4E chip, one more recent with SRV4G chip. . The other board has three servos with one MERG driver pcb SRV4G chip. The servos are a mix of JPS7.5 and Tower Pro SG90. I will work my way through possible resolutions at the weekend. I can't start on it earlier as I have to be away from home for a couple of days (and the workshop is a little chilly at the moment).

In the meantime, this the following describes the set up and the symptoms.

The MERG drivers are provided with 12V DC from a dedicated 12V regulator in the control panel box (the one at bottom left). SPST panel switches are used for the input controls to J6 on the PCB.
CP voltage regulators.JPG


The PCBs are mounted under the baseboards on cross braces. This is the later version. Only one of the demountable signals has been connected.
Earlier type MERG servo driver.JPG


This is the earlier type, used to control the ground signal servos, showing the position of other electrical components, etc.
Earlier type MERG servo driver.JPG


The servos function correctly when no locos are running and aren't affected by operation of the turnout motors (Cobalts and one Tortoise). However, when locos with Mashima motors (mainly 1224s during the weekend) are run then intermittent but severe "twitching" happens. One loco fitted with a Portescap didn't cause a problem and trying both Pentroller and Modelex controllers made no difference. The servos don't all twitch at the same time or to the same amount. The "twitch" seems to be at a fast speed, irrespective of the MERG driver speed setting by the MERG control box.

I'll start (at the weekend, I'm committed over the next couple of days and the workshop is rather chilly at the moment) by disconnecting the servos and reconnecting them one at time, as this may show if the servo leads are picking up interference. I've bought some shielded cable to connect the switchgear and 12v supply, in case those cables are causing the problem. I'll report on progress as I proceed.

I have an invite to take these two boards to expoEM in May to demonstrate ILC (indifferent layout construction) so, if the problem isn't resolved, then the members of MERG will be welcome to descend on it with test meters in hand to find a solution.

In the meantime, any sensible suggestions are welcome,

Jol
You do not have the required permissions to view the files attached to this post.

User avatar
grovenor-2685
Forum Team
Posts: 3918
Joined: Sun Jun 29, 2008 8:02 pm

Re: Twitchy servos

Postby grovenor-2685 » Wed Dec 12, 2012 4:16 pm

Investigations have been progressing on this and we do now have some recommendations, I'll report on them shortly. Meanwhile it would help if you can confirm the version of firmware in your modules. It should be written on the PIC label.
Regards
Keith
Regards
Keith
Grovenor Sidings

User avatar
Jol Wilkinson
Posts: 1113
Joined: Mon Jul 21, 2008 7:39 pm

Re: Twitchy servos

Postby Jol Wilkinson » Wed Dec 12, 2012 4:30 pm

Hi Keith,

the labels on the Merg servo drivers are;

SRV4G for the three later updated versions and SRV4E for the original older version.

Regards,

Jol

User avatar
grovenor-2685
Forum Team
Posts: 3918
Joined: Sun Jun 29, 2008 8:02 pm

Re: Twitchy servos

Postby grovenor-2685 » Wed Dec 12, 2012 9:44 pm

Below is the latest investigation report, should help somewhat.
As it will probably be the new year before I finish the tests at the moment the best suggestions (collated from all the various posts and not just dreamt up by myself) seem to be:
a) Run traction and servo circuits in separate looms, and keeps these at least 5 cms (although this is an arbitrary distance taken from what I happen to have on my work bench for the tests) apart where possible;
b) Fit 10K (although this is an arbitrary resistance selected at random) pullup of servo "drive signal" to servo +5V, as close to the servo as possible;
c) Use "Servo4h" (I'll be updating my "Sema4" to go high on timeout in due course);
d) Select "timeout" option for "drive signal";
e) Use twisted pair for servo "drive signal" and ground;
f) Use twisted pair for traction supply and return.
I've put the list in a presumed order of importance but I don't have any evidence for this order at the moment. I'd suggest definitely do "a)" and "b)", preferably do "c)" and "d)", then if you can do "e)" and "f)" (try and at least do "e)" if "f)" isn't practicable).

It might not be necessary to apply all the suggestions. Some might like to apply them in order until the problem disappears. Also not everyone will be in a position to update firmware in which case try "e)" and "f)" before worrying about "c)" and "d)". As always if anyone finds that just applying some of the suggestions clears the problem for them please post to the forum as this should help clarify a simplest minimum, "least effort", set of fixes.
Chris White (M819)


My suspicion is if they select the "power off after timeout" option most people won't see any problems when using Mike's firmware, particularly the latest "Servo4h", or my latest "Sema4e". However I will be putting out a newer version, "Sema4f", to pull the outputs high when timed out as Mike now does.

As to the list of suggestions. Well I'd reiterate they are just that, suggestions. If you're starting greenfield and can get things like twisted pair and separate looms in from day one it won't hurt to do so. If, like our club, you have a finished project and have just started to witness the issue then try the latest firmware with the time out option first, it could be as much as you need. Next step would be to get the traction and servo circuits into separate looms, then if possible get twisted pairs in there.
Chris White (M819)


It would seem that the first thing to do is to get the latest firmware installed (Servo 4h) and see if that is enough.
Contact me off list if you need help with that.
Regards
Keith
Regards
Keith
Grovenor Sidings

nigelcliffe
Posts: 747
Joined: Mon Jul 21, 2008 8:31 am

Re: Twitchy servos

Postby nigelcliffe » Wed Dec 12, 2012 10:03 pm

I'm reasonably local to Jol, and can do servo firmware updates if that helps.

- Nigel

User avatar
Jol Wilkinson
Posts: 1113
Joined: Mon Jul 21, 2008 7:39 pm

Re: Twitchy servos

Postby Jol Wilkinson » Wed Dec 12, 2012 10:07 pm

Keith, Nigel

Thanks, I'll be away for a couple of days and will get in touch when I return.

Jol

User avatar
Jol Wilkinson
Posts: 1113
Joined: Mon Jul 21, 2008 7:39 pm

Re: Twitchy servos

Postby Jol Wilkinson » Tue Dec 18, 2012 8:56 am

For those interested in the outcome of my experiences in the probable cause and cure(s) for this problem, several things are underway on which I'll report back in due course.

Jol

Mark Tatlow
Posts: 898
Joined: Tue Dec 02, 2008 11:24 pm

Re: Twitchy servos

Postby Mark Tatlow » Sun Jan 27, 2013 7:40 pm

Gents,

My servos twitch appallingly when nearby Fulgerexs are used. I know that this is to do with dirty signals coming off the fairly crude motors that go into these Fulgerexs and I know that it can be smoothed with a suppresser.

But how do I go about doing that exactly...............???
Mark Tatlow

User avatar
grovenor-2685
Forum Team
Posts: 3918
Joined: Sun Jun 29, 2008 8:02 pm

Re: Twitchy servos

Postby grovenor-2685 » Sun Jan 27, 2013 7:49 pm

A set of those capacitors you normally cut off locos when converting to DCC should do the trick.
Refer http://www.pololu.com/docs/0J15/9

from Rapid electronics http://www.rapidonline.com/Electronic-Components/100nf-2-5mm-Y5v-Dielect-Ceramic-Capacitor-08-0275
Regards
Keith
Regards
Keith
Grovenor Sidings

User avatar
Jol Wilkinson
Posts: 1113
Joined: Mon Jul 21, 2008 7:39 pm

Re: Twitchy servos

Postby Jol Wilkinson » Fri Feb 01, 2013 8:35 am

Prompted by Will's piece Forme, ForYou, Forum in Snooze 181, the problem with servo twitch on the extension to London Road is not yet unresolved and still under investigation. The cold weather hasn't helped as it isn't too comfortable in the workshop. and with my dearly beloved's now annual trip to New Zealand taking place, modelling moved into the kitchen so I have been doing other things mainly involving plasticard.

However, an update will follow shortly.

Jol


Return to “Electrics”

Who is online

Users browsing this forum: ClaudeBot and 0 guests