(report on) David Barham's servo presentation to the NLG

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

(report on) David Barham's servo presentation to the NLG

Postby Russ Elliott » Thu Oct 04, 2012 1:14 pm

David Barham was the guest speaker at the September meeting, and gave an excellent presentation on servos and their application in turnout, signal and lever frame operation.

After a brief overview of previous solenoid and limit-switch or stall-based slow-action mechanisms, the latter now becoming increasingly expensive, David described the differences in wiring required for the small and cheap servos now available, at typically £2-50 each. Servos are 3-wire devices, and need to be programmed to tell them which way to move, how fast to move, and how far to move. A servo's control signal is a regular series of digital pulses whose length is varied by controller circuitry. Controllers are available quite cheaply, the MERG 'Servo4' unit being about £6 each, which can control four servos. The programming of the servos is from a dedicated programming unit or can be set from a PC, where software can also programme signal arm bounce. David illustrated methods of mounting servos, and their links to TOUs etc, using simple plasticard structures. He also addressed the issue of servo 'twitching', and ways to avoid it.

The second part of David's presentation concerned the bus system developed by MERG as a means of reducing the amount of wiring needed where many servos are being deployed on a layout. This 4-wire bus system allocates identities to each servo, and carries signals from a master control panel board, which can take inputs from up to 128 switches, through to various types of output device boards on the layout itself. The programming of 'events' and 'actions' can be achieved either manually or via PC software. Although the individual 'controlling' and 'consumer' devices are relatively inexpensive, their total cost does add up, and there is obviously a tradeoff between the cost of such a bus system and alternative more conventional means. The MERG software available for controlling the bus was I thought somewhat arcane and not particularly user-friendly for technophobes, but there's no doubt the MERG developers have done a very thorough, solid, and flexible implementation.

In the context of a simple layout, I wasn't entirely convinced about the "servos simplify wiring" message. Unlike many conventional actuators, e.g. Tortoise or Fulgurex, servos do not have built-in auxiliary switch contact sets, so crossing vee polarity switching for example has either to be provided by extra switch/lever poles at the control panel or effected by installing microswitches adjacent to the servo arm. The former approach will add considerably to the number of cores between a control/lever panel and the layout baseboards. Moreover, any auxiliary control panel switching, whether used for vee polarity changing or forward routing or sequencing of other indicators, will assume a servo has completed its movement successfully.

The third section of David's presentation focused on the locking of lever frames using servos. Using a conventional SHAG frame, David used some very thin (and slightly more expensive!) servos that would fit within the 10mm pitch of the frame levers. These servos rotate an ingenious collar to lock and unlock each lever*. By this approach, all the logic of the locking chart can be implemented 'off-frame', and developed flexibly and gradually. The comparison with the finality and mechanical complexity of a precision bar and tappet implementation was self-evident.

I think some of us were probably slightly brain-overloaded at the end of the evening, but many thanks to David for covering a lot of ground, and making an unfamiliar and highly technical subject understandable.


* Some of this is shown in http://scalefour.org/forum/viewtopic.php?f=4&t=1772

User avatar
Mark Tatlow
Posts: 696
Joined: Tue Dec 02, 2008 11:24 pm

Re: (report on) David Barham's servo presentation to the NLG

Postby Mark Tatlow » Fri Oct 05, 2012 8:43 am

Russ,

I have started to use servos and I am sold on them.

They only simplify the wiring a little bit in truth (you need one wire to each servo plus one extra; as opposed to a pair of each TOU's). It is true that you need to come up with a microswitch to control crossing polarity but again this is the same as a TOU (agreed TOU's often have this built in but servos now come with mounts that can etc).

What sells them to me is that they are so adjustable and can be set to operate gently/realistically. This is of benifit for turnouts but it is when they are used for signals that they really come into their own.

I am worried about the servo twitch that some are reporting though, I have not experienced it though.
Mark Tatlow

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

Re: (report on) David Barham's servo presentation to the NLG

Postby Russ Elliott » Fri Oct 05, 2012 11:45 am

Mark Tatlow wrote:you need one wire to each servo plus one extra; as opposed to a pair of each TOU's.

Hang on a mo, I'm getting confused now. Displaying my ignorance of servo matters, I have two elementary questions to ask:

servo4-query.png
servo4-query.png (11.46 KiB) Viewed 4679 times


Hmmm (post diagram edit): Question 3 on the above - can the lever input be a single-wire split potential one?

It is true that you need to come up with a microswitch to control crossing polarity but again this is the same as a TOU (agreed TOU's often have this built in but servos now come with mounts that can etc).

I'm not convinced. There is a considerable difference in setting up a microswitch that will actuate at some point in a long-travel (say 8mm) pre-TOU movement from a screw-driven machine than setting one up on a servo arm throw of say only 1.5mm. Many microswitches will be a sod to set up over that kind of range. The inclusion of reliable changeover contact sets (note plural, it's not just the vee polarity we might be concerned with) within Tortoises and Fulgurexes remains a pertinent factor in their favour IMO.

What sells them to me is that they are so adjustable and can be set to operate gently/realistically. This is of benifit for turnouts but it is when they are used for signals that they really come into their own.

No question about those benefits. David showed those in his demo bits at the NLG presentation. (For signals though, yer can't beat a Dutch gravity bouncer for replicating the decay curve, but it did strike me that a better characteristic from a servo would be to use it in long-travel mode and lever-ratio it down, which could disguise the stuttering.)

I am worried about the servo twitch that some are reporting though, I have not experienced it though.

I've no experience of servos, so can't comment. The phenomenon seems to be very rare (see servo bounce thread).

nigelcliffe
Posts: 492
Joined: Mon Jul 21, 2008 8:31 am

Re: (report on) David Barham's servo presentation to the NLG

Postby nigelcliffe » Fri Oct 05, 2012 12:26 pm

Russ Elliott wrote:
Mark Tatlow wrote:you need one wire to each servo plus one extra; as opposed to a pair of each TOU's.

Hang on a mo, I'm getting confused now. Displaying my ignorance of servo matters, I have two elementary questions to ask:

servo4-query.png


Question 3 on the above - can the lever input be a single-wire split potential one?




The inputs are on/off, with "on" connecting the contact on the input to the common ground wire. So, wiring of a Servo-4 to a lever frame is five wires, one common ground wire, and four wires to four on/off switches. Wiring of two Servo-4's to a lever frame is nine wires, one common ground, and eight to on/off switches. etc..


The outputs to drive a servo motor can be commoned up as shown, they are common on the board itself as drawn. However, in practise its sensible to have the servo motor leads as short as reasonably practical, and then using the supplied three-wire lead on the servo with the supplied three-pin plug works with the lowest hassle. Sending a single servo across a baseboard join isn't usually the most sensible thing to do (though I have done it recently, breaking my own advice on best practise!).


It is true that you need to come up with a microswitch to control crossing polarity but again this is the same as a TOU (agreed TOU's often have this built in but servos now come with mounts that can etc).

I'm not convinced. There is a considerable difference in setting up a microswitch that will actuate at some point in a long-travel (say 8mm) pre-TOU movement from a screw-driven machine than setting one up on a servo arm throw of say only 1.5mm. Many microswitches will be a sod to set up over that kind of range. The inclusion of reliable changeover contact sets (note plural, it's not just the vee polarity we might be concerned with) within Tortoises and Fulgurexes remains a pertinent factor in their favour IMO.


The trick with a servo is to NOT set the mechanical links so the arm moves only 1.5mm. It needs to move a decent distance, then the electrical contact setup is much simplified as you indicate. Alternatively, if the servo control board had an additional output to control a relay, then the electrical switching of crossings (etc) would not need microswitches. The Servo4 doesn't offer this, but some alternative servo control boards do.

What sells them to me is that they are so adjustable and can be set to operate gently/realistically. This is of benifit for turnouts but it is when they are used for signals that they really come into their own.

No question about those benefits. David showed those in his demo bits at the NLG presentation. (For signals though, yer can't beat a Dutch gravity bouncer for replicating the decay curve, but it did strike me that a better characteristic from a servo would be to use it in long-travel mode and lever-ratio it down, which could disguise the stuttering.)


Agree on all counts, mechanical bounce still looks best if you can put the time into building it. Using longer travel and levering the ratio downwards is better. Its also necessary to have a spring in the linkage because there is a small risk of a servo deciding to temporarily go bonkers and swing its maximum throw before recovering. This often happens on power-up or other power glitches.


- Nigel

User avatar
grovenor-2685
Posts: 3136
Joined: Sun Jun 29, 2008 8:02 pm

Re: (report on) David Barham's servo presentation to the NLG

Postby grovenor-2685 » Fri Oct 05, 2012 12:31 pm

Russ Elliott wrote:
Mark Tatlow wrote:you need one wire to each servo plus one extra; as opposed to a pair of each TOU's.

Hang on a mo, I'm getting confused now. Displaying my ignorance of servo matters, I have two elementary questions to ask:

Servos come with a 3-wire cable and attached connector, these wires are +5V, 0V and a control wire, whilst you could just run the control wire back to the driver board and connect the +5V and 0V some other way you would just be making extra work, the driver boards come with the 5V supply and a set of standard 3 pin headers so that the servo cables fit straight on.
servo4-query.png


Hmmm (post diagram edit): Question 3 on the above - can the lever input be a single-wire split potential one?
The lever input cannot be split potential, its just a single contact closure to connect the relevant input pin to the 0V common, you use 5 wires for 4 servos.

It is true that you need to come up with a microswitch to control crossing polarity but again this is the same as a TOU (agreed TOU's often have this built in but servos now come with mounts that can etc).

I'm not convinced. There is a considerable difference in setting up a microswitch that will actuate at some point in a long-travel (say 8mm) pre-TOU movement from a screw-driven machine than setting one up on a servo arm throw of say only 1.5mmMany microswitches will be a sod to set up over that kind of range. .
It will not be satisfactory to use a servo arm throw of 1.5mm, that would be very poor mechanical design. I always try to arrange the mechanics to that the servo moves through at least 45 degrees which would be 10mm plus at the end of the arm. Its simple to set up microswitches in such a case. Although in many cases I use a relay with one contact set switching the frog polarity and the other energising the servo.
The inclusion of reliable changeover contact sets (note plural, it's not just the vee polarity we might be concerned with) within Tortoises and Fulgurexes remains a pertinent factor in their favour IMO.
What sells them to me is that they are so adjustable and can be set to operate gently/realistically. This is of benifit for turnouts but it is when they are used for signals that they really come into their own.

No question about those benefits. David showed those in his demo bits at the NLG presentation. (For signals though, yer can't beat a Dutch gravity bouncer for replicating the decay curve, but it did strike me that a better characteristic from a servo would be to use it in long-travel mode and lever-ratio it down, which could disguise the stuttering.)

See comment about mechanical design above.
I am worried about the servo twitch that some are reporting though, I have not experienced it though.

I've no experience of servos, so can't comment. The phenomenon seems to be very rare (see servo bounce thread).

There are two seperate twitching phenomena, the first is essentially a bit of random movement on power up, this has been practically eliminated by tweaks to the software and board design. The second is a more severe movement during operation which has been reported by a few users, it appears to be caused by induction of spikes into the control wire being recognised by the servo as a request to move. We aim to investigate this at MERG and find ways to overcome it. So far this has not progressed very far as we have yet to get enough information to reproduce the problem.
Regards
Keith

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

Re: (report on) David Barham's servo presentation to the NLG

Postby Russ Elliott » Fri Oct 05, 2012 6:45 pm

I'm slightly amazed that the servo supply rails can be commoned - the servos are obviously cleverer than I thought, but I take the point about the pragmatic convenience of the standard 3-pin header. Glad to hear of the single-wire input, which will save valuable cores between boards and a control panel. So this presumably is the basic schematic:

servo4-query2.png
servo4-query2.png (7.02 KiB) Viewed 4617 times


I can't suggest anything on the twitching issue, except the obvious one of using coax for the control wire, but I'm sure MERG must have considered that already.

User avatar
grovenor-2685
Posts: 3136
Joined: Sun Jun 29, 2008 8:02 pm

Re: (report on) David Barham's servo presentation to the NLG

Postby grovenor-2685 » Fri Oct 05, 2012 7:44 pm

Russ Elliott wrote:I'm slightly amazed that the servo supply rails can be commoned - the servos are obviously cleverer than I thought, but I take the point about the pragmatic convenience of the standard 3-pin header. Glad to hear of the single-wire input, which will save valuable cores between boards and a control panel. So this presumably is the basic schematic:

almost, the common to the switches should come from the 0V not from the +5V, and the Servo4 board has its own on-board 5V supply so should be supplied with 9 - 12V dc or around 9V ac, higher ac could be used but would need a bigger heatsink on the 5V regulator.

I can't suggest anything on the twitching issue, except the obvious one of using coax for the control wire, but I'm sure MERG must have considered that already.

The primary objective is to use the cheap servos out of the packet, so we don't want a solution that involves changing the cable or rebuilding the servo. This is not a very prevalent problem we have sold hundreds of these and there seem to be under half a dozen having problems, but that is still to many so we do need to come up with a fix.
Regards
Keith

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

Re: (report on) David Barham's servo presentation to the NLG

Postby Russ Elliott » Fri Oct 05, 2012 9:04 pm

Like this?

servo4-query4.png
servo4-query4.png (6.51 KiB) Viewed 4592 times


Btw, what's the supplied 3-wire cable length?

User avatar
grovenor-2685
Posts: 3136
Joined: Sun Jun 29, 2008 8:02 pm

Re: (report on) David Barham's servo presentation to the NLG

Postby grovenor-2685 » Fri Oct 05, 2012 9:48 pm

Diagram now OK.
The little micro servos have 10 - 12 inches of cable. They are meant to be used in model aircraft, plugging into the radio receiver, and they would not want to much spare cable to hide. You can buy extension cables or just make up your own when needed.
Regards
Keith


Return to “North London Group”

Who is online

Users browsing this forum: No registered users and 1 guest