North East Essex Area Group meeting

Moderator: Jol Wilkinson

User avatar
Jol Wilkinson
Posts: 720
Joined: Mon Jul 21, 2008 7:39 pm

North East Essex Area Group meeting

Postby Jol Wilkinson » Thu Apr 21, 2016 8:23 am

NEEAG March meeting will be held at 7:00 on Thursday 28 April in the usual venue:

The Colchester Society of Model and Experimental Engineers Clubhouse
13a President Road
Colchester
CO3 9ED

Old and new members welcome.

User avatar
Jol Wilkinson
Posts: 720
Joined: Mon Jul 21, 2008 7:39 pm

Re: North East Essex Area Group meeting

Postby Jol Wilkinson » Fri Apr 29, 2016 11:32 am

Last night NEEAG meeting was held in rather dampened spirits following the sad news that Mike Gipson had succumbed earlier in the week to his fight with leukemia.

Mike, who had founded the group, was an retired marine engineer and accomplished model engineer. He had decided to "downsize" and became an enthusiastic S4 modeller. He had started on a model of the WW1 military railway at Fovant, Wiltshire, that branched off the LSWR at Dinton.

Besides discussing various "admin" topics", the evening was also spent looking at servo control of signals and trying out MERG and Tam Valley servo controllers and various servos. The latest test etches of a new London Road Models GNR /LNER loco were on display.

28 April 16.jpg
28 April 16.jpg (141.02 KiB) Viewed 2829 times

User avatar
Weskie
Posts: 35
Joined: Sat Jul 26, 2008 1:12 pm

Re: North East Essex Area Group meeting

Postby Weskie » Fri Apr 29, 2016 8:02 pm

Jol.

Did you come to any conclusion about the signal servo's?
Andy Westcott

User avatar
Jol Wilkinson
Posts: 720
Joined: Mon Jul 21, 2008 7:39 pm

Re: North East Essex Area Group meeting

Postby Jol Wilkinson » Sat Apr 30, 2016 8:26 am

Weskie wrote:Jol.

Did you come to any conclusion about the signal servo's?


Not really, Andy. My reason for taking along all the servo control gear was to get some independent views on the analogue and digital servos as well as to find out if anyone else could understand the instructions for setting up the Tam Valley Octopus.

The Tower Pro digital servo doesn't seem to like the MERG unit and occasionally - when powered up - likes to carry out several rotations in one directions and then the other before settling down. The analogue servos don' do this but start up twitch and excessive movement away from the set up position still remains a problem. This can bend the operating linkage.

Despite following the instructions, no one could work out how to set up the Octopus. While it has simple self centering options which seem to be aimed at driving PECO turnouts, understanding how to set it (and whether you have succeeded) with limited travel and specific speed for signals still escapes us.

I have spent too much time (and money) trying to get reliable servo control for signals. One recommendation following a discussion at S4N with Steve Hewitt is to use the GF Controls system as used on Liverpool Lime Street (probable cost for what I need £150 -£170). One of the NEEAG members has bought the Mega Points system - for use with DCC - and that would probably set me back £150 for what I need. The four post signals and six ground signals are spread about a 6ft and a 5ft board and as most manufacturers recommend keeping the servo leads as short as possible, I would need several servo controllers.

I am loathe to invest that amount without an absolute guarantee that I could fit, set and forget! I am also not convinced about the amount of built in bounce in the GF and Meg Points units, especially as lower quadrant signals apparently have much less (some times barely noticeable) than UQ signals.

Following a couple more wasted hours with the Octopus yesterday I spoke to Clive Heathcote of Heathcote Electronics and ordered one of his two servo controllers. He was very open about the problems he has found with servo twitch, etc. and the changes he made to minimise it. He made several installation recommendations (which I had already discovered). I'll report back when I have received and tested it.

Jol

billbedford
Posts: 615
Joined: Tue Oct 02, 2012 2:40 pm

Re: North East Essex Area Group meeting

Postby billbedford » Sun May 01, 2016 9:25 am

Jol Wilkinson wrote:[...], especially as lower quadrant signals apparently have much less (some times barely noticeable) than UQ signals.


Didn't LQ signals tend to bounce when they were pulled on?
Bill Bedford
Mousa Models
http://www.mousa.biz

User avatar
David B
Posts: 1001
Joined: Sun Jul 20, 2008 5:30 pm

Re: North East Essex Area Group meeting

Postby David B » Sun May 01, 2016 10:23 am

As one of the Missenden Abbey Weekend organisers, I am following this thread with interest as we have a course on signal operation (making them work on a model) this coming October run by Martin Lloyd, a pioneer in the use of servos. This is a subject that does cause problems for modellers not familiar with servos and the associated electronics. Our electronics courses are run by the people who design the MERG kits, who will help with any problems modellers have with electronics.

With this in mind, we have contacted Mike Bolton who has been instrumental in the creation of many of the MERG kits and the CBUS system. Mike is one of the tutors at the Missenden Autumn Weekend. This is what he has had to say:

We spent some time sorting the startup twitch on the Servo4 boards. Where it could be a result of the Servo4 firmware and hardware, we have totally fixed the issue, at least in my Servo 4f. We did extensive testing of these. Earlier boards may still have issues.

However, the main problem where twitching is concerned is poor power supplies to the Servo4 boards. This we have no control over other than to make recommendations.

I cannot speak for the Servo4 code by others which has the bounce in it.

Where servos twitch due to passing DC locos, that is interference from the locos directly to the servos. It suggests poor loco suppression. Often the case with handbuilt P4 locos where no suppressors are usually fitted and where bouncy pickup from the wheels is common. One reason why DCC locos are better and why it has been reported for some locos and not others.

We have no control over the design of the servos themselves or whether some makes / types are more prone to pickup. That will be true of any servo controller, not just the MERG ones. The suggestion that using a micro per output is better than one micro for 4 outputs is a total ‘red herring’. Also screening or twisting the control wires to the unit is highly dubious as the micro code has interference rejection and debounce in it. Both the Megapoints and GF designs are based on the MERG one. I was speaking to the GF designer last week at a Gauge O show and he wouldn’t answer me when I asked where he got the design from, but it was pretty obvious to me.

However, I have not used the Servo4 myself (ever) but loads of the CANSERVO8 boards, including on Slattocks which is P4. We have no twitching issues with these, even with longish leads to the servos and different makes of servo although it is a DCC layout. That said, the club ‘Dewsbury Midland’ (OO) uses many servos for its signals including a gantry of 11 signals, all powered by Servo4 modules with SG90 servos and DC (OO RTR locos). This has worked perfectly for years and at many shows. No twitching issues here. Any twitching would destroy a beautiful gantry.

I know there are many layouts using Servo4 boards that have no problems so I am pretty certain it is not a MERG issue, but one that may occur with all servo controllers. I do know that any based on a 555 timer are prone to interference, hence the move to using a micro in the first place. A reason why I would not recommend the MERG PMP servo controller.

Maybe worth asking on the MERG Forum who has had problems and who has not, and what the circumstances were.

User avatar
grovenor-2685
Posts: 3000
Joined: Sun Jun 29, 2008 8:02 pm

Re: North East Essex Area Group meeting

Postby grovenor-2685 » Sun May 01, 2016 10:27 am

Didn't LQ signals tend to bounce when they were pulled on?

Signals are pulled off, gravity sets them back to on.
Granted with LQ the pull is only as far as the balance lever with a rod from the balance lever to push the arm off.
Bounce is more likely when going back to on as the wire is then slack, but the combination of arm and balance lever rodded together has more inertia than an UQ arm and the bouncing slower and less obvious. And the further from the box and hence more wire to pull back the less its likely to bounce.
The double pull by the signaller when pulling off is, however, more likely on the more remote signals as the lever will be heavier to pull.
Regards

dal-t
Posts: 571
Joined: Mon Jan 11, 2010 8:06 pm

Re: North East Essex Area Group meeting

Postby dal-t » Sun May 01, 2016 8:49 pm

Just watching Paul Merton's 'Secret Stations' and must admit I was surprised by the amount of rocking/bouncing as a l/q semaphore was pulled off. Mind you, it was an arm without a spectacle plate - the clip was too short for me to spot what the lighting arrangement actually was, but someone with younger eyes and/or video replay might have pinned it down?
David L-T

User avatar
David Thorpe
Posts: 744
Joined: Tue Oct 21, 2008 11:13 pm

Re: North East Essex Area Group meeting

Postby David Thorpe » Sun May 01, 2016 9:20 pm

Jol Wilkinson wrote: One of the NEEAG members has bought the Mega Points system - for use with DCC - and that would probably set me back £150 for what I need. The four post signals and six ground signals are spread about a 6ft and a 5ft board and as most manufacturers recommend keeping the servo leads as short as possible, I would need several servo controllers.


For ten signals spread over 11' of board, I don't think you'd have to worry about your servo leads not being long enough. Megapoints state quite clearly in their manual that "servo cables can easily be extended with a servo extension cable or by soldering cable to a recommended maximum length of 5 metres". It seems therefore that one centrally placed servo controller (which can serve up to 12 signals) would be adequate for your signalling leads.

DT

User avatar
Jol Wilkinson
Posts: 720
Joined: Mon Jul 21, 2008 7:39 pm

Re: North East Essex Area Group meeting

Postby Jol Wilkinson » Mon May 02, 2016 8:57 am

David Thorpe wrote:
Jol Wilkinson wrote: One of the NEEAG members has bought the Mega Points system - for use with DCC - and that would probably set me back £150 for what I need. The four post signals and six ground signals are spread about a 6ft and a 5ft board and as most manufacturers recommend keeping the servo leads as short as possible, I would need several servo controllers.


For ten signals spread over 11' of board, I don't think you'd have to worry about your servo leads not being long enough. Megapoints state quite clearly in their manual that "servo cables can easily be extended with a servo extension cable or by soldering cable to a recommended maximum length of 5 metres". It seems therefore that one centrally placed servo controller (which can serve up to 12 signals) would be adequate for your signalling leads.

DT


In theory David, but as it is a portable layout I want as few servo leads that have to be plugged in each time we set up and want to minimise any extra cabling beyond the umbilical cables that we already have between the control panel and the boards. So a controller for each board is the minimum. Another downside of the MegaPoints Controller (and others) is the need to do the set up with push buttons and flashing LEDs on the controller. Great on a sales stand at shows, but not ideal when it is mounted under a baseboard (especially at my time of life). The separate " remote setting box" is one of the things that attracted me to the MERG system in the first place.

The length of servo cables seems to be a bit of a variable, some manufacturers recommending it is kept to the minimum.

Mike Bolton's comments make interesting reading. The Mega Points website says that "If you must not have a power on twitch (2mm fine scale and delicate work)" and quotes a supplier reference for a Digital servo, the implication being that servo twitch on power up is not unusual. Mike Bolton states that "We spent some time sorting the startup twitch on the Servo4 boards. Where it could be a result of the Servo4 firmware and hardware, we have totally fixed the issue, at least in my Servo 4f" but then goes on to say "However, I have not used the Servo4 myself (ever)" I am a bit confused by this as the MERG unit I use have "Servo 4F" stickers on the chips. If his claim comment that the MegaPoints and GF Controls units are based the MERG design, then it would seem likely that I would suffer the same problems with those.

The comment about power supplies is interesting. The MERG site quotes "The power to the board can be either a DC supply of 10 to 15v, either smoothed or unsmoothed, or an AC supply of between 9 and 12v RMS" which would seem to indicate that the Servo 4 is pretty supply tolerant. Other manufacturers usually just quote a DC input between X and Y volts, although the Tam Valley Octopus has it's own on board power supply (regulator?). I use a 12V, 2 amp regulated and smoothed supply which should meet MERG's requirements.

I fully accept that most kit built locos with Mashima or similar motors don't usually have suppression fitted and this may be a source of the problems I initially experienced with twitching caused by locos but have now almost totally resolved. It would still be worth doing so can anybody recommend the best way to do this? I've done an internet search but this produced a variety of different answers and component values. However, Mike Bolton's comment about bouncy pickups on P4 locos is also a red herring, as he likes to say. Compensation/springing gives better current collection than achieved with rigid locos.

The more you look into this, the more confusing it becomes with various designers/manufacturers claims about what their system can/does/doesn't do. It seems that I have had a fair share of problems with servo operation of semaphore signals. Perhaps such issues would be less evident with servo point control, where the mechanical items are more robust, travel/movement limited and erratic movement probably less visible.

nigelcliffe
Posts: 462
Joined: Mon Jul 21, 2008 8:31 am

Re: North East Essex Area Group meeting

Postby nigelcliffe » Mon May 02, 2016 1:56 pm

It is my experience that the source of Servo motor is a significant factor. I have a number of "9G" size servos (the common cheap size) from different sources, and get different amounts of sensitivity to electrical noise. Some work just fine, others are easily twitched by either certain locos in the vicinity, or operation of electromagnetic uncoupler coils.

I find it simpler to just replace a servo motor with a different maker's motor than faff around with additional pull-up and pull-down components on control boards.


Though I've built a number of the MERG Servo4 boards, and changed the firmware on Jol's in a quest for reliable running, I don't currently use any of them. The currently used boards for Coldfair Green are the LocoServo from Hans De Loof, which interfaces nicely onto LocoNet, which in turn talks to Coldfair Green's new control panel. With those, I have replaced some servo motors which were twitchy. The LocoServo board turns off its data pulse to the servo motor after a few seconds, so any residual servo movement/buzz ends within a few seconds.


For signals, I wonder if a simple solution could be produced for Jol using Graham Masham's methods with stepper motors and Arduino processors. These can have programmable (and random within limits) bounce, etc. It is cheap, guess at £5 per motor with control system, and could be arranged so it is a one-wire to operate (on/off) plus power supply to all the boards on the layout.


- Nigel

User avatar
grovenor-2685
Posts: 3000
Joined: Sun Jun 29, 2008 8:02 pm

Re: North East Essex Area Group meeting

Postby grovenor-2685 » Mon May 02, 2016 3:41 pm

The LocoServo board turns off its data pulse to the servo motor after a few seconds, so any residual servo movement/buzz ends within a few seconds.

So does the MERG servo4 (at the users option).
Graham's stepper method could well be a good alternative, i was sorry to miss his demo at the last Scaleforum.
5 pounds per arm is pretty well the same as Servos.

Incidentally the MERG Sema4 firmware has been updated to make it easier to include a 'double pull' when setting up the bounce.
Regards

ColinMcC
Posts: 6
Joined: Mon Jun 18, 2012 9:13 pm

Re: North East Essex Area Group meeting

Postby ColinMcC » Wed May 04, 2016 8:46 pm

I have been reading this topic with interest and offer the following thoughts.

I suspect that the problem is caused by the back EMF generated by the motor (this occurs if the motor is turning and no voltage is being applied) and is more likely to be caused by controllers that use PWM to drive the motor. The PiCController (I believe) uses PWM and that the Pentroller may also use this method (I don't have access to a scope to prove this). Older type of controller, e.g. KPC hand-helds, have some sort of back EMF detection and alter the voltage appropriately but I don't believe use PWM. Whilst at Raymarine, I was part of the team that designed auto-pilots and we to process this back-EMF signal and it was found to rather noisy. This noisy signal is returned via the track-work to the controller.

Therefore I believe that this back-EMF signal interferes with the quality of signal used to control the operation of the servo. So, in line with this thinking, a simple cure maybe to wrap both the servo and its connecting cable in aluminium foil or self-adhesive copper tape. Additionally, it may be prudent to place suitable capacitors to suppress the signals at the track feed.

It is likely with DCC operation that this noisy back-EMF signal would be confined to the decoder (within the locomotive) and so the problem is less likely to occur.

I have used the "Bouncer" system designed by Tony Witts (of the SHMRC) successfully on Ogden Fold for its signals. It has also been used on Hope-under-Dinmore for its signals however its use on the turnouts was abandoned because of the twitching of the point blades. I am aware that the operation of the Bouncer was reverse engineered by MERG member(s) to modify its mode of operation. I was a member of MERG for a while until they stopped me paying subscriptions by direct debit.

I hope that these comments help.

Colin McCallum.

User avatar
Jol Wilkinson
Posts: 720
Joined: Mon Jul 21, 2008 7:39 pm

Re: North East Essex Area Group meeting

Postby Jol Wilkinson » Thu May 05, 2016 7:54 am

Colin,

shielding of the servo cables is one thing I haven't tried but as spasmodic twitching (no, not occasional bird watching) has gone or at least we aren't aware it still happens following the steps I have taken. These include the installation of the Tam Valley "No-Buzz In-line Servo Quieter" in the servo cables - don't you just love American marketing speak - so I don't think I need to try that.

The "failure" at S4N affected one MERG board only and was unlike anything else experienced as it was impossible to reset the servos using either the control panel switches or the remote settings box.

The addition of capacitors at the track feeds hasn't been suggested before. As the layout has what amounts to section busbars to feed the droppers, where should the capacitors go and what type/rating should they be.

Some posts ago I asked for recommendations for the circuit/type /rating on loco motor suppressors but so far nothing has been suggested. In days of yore I believe Hornby Dublo used a choke and capacitors but I don't know how they were configured. An internet search produces a variety of suggestions, including the use of ferrite beads. So for someone that understands electrics well enough to wire a layout and can assemble MERG kits but isn't an electronics expert, I find myself rather lost.

Jol

ColinMcC
Posts: 6
Joined: Mon Jun 18, 2012 9:13 pm

Re: North East Essex Area Group meeting

Postby ColinMcC » Thu May 05, 2016 7:53 pm

Hi Jol,

The problem at S4N with the MERG controller may have been a genuine failure and that's why we always carry a spare! By the way did we power off and on the unit as it may have been a software failure?

In the last century, whilst living in Sutton Bridge, I used to help to build layouts and we always used to install a suppression capacitor where the power was fed to the track. I would guess that it would be sufficient to place a capacitor at the feed(s) closest to the signal on platform 1 and any other twitchy signals. 22nF or 47nF disc ceramic capacitors (as they're non-polarised) could be used. Rapid Electronics in Colchester have part numbers of 08-1065/11-3447 and 11-3448/11-3453 respectively. There are about 5p each (including VAT). Ferrite beads should not be necessary.

The choke / capacitor used on HD locomotives is probably a bit on overkill. The 22nF capacitors described above would be sufficient in locomotives with Mashima or similar motors [I probably would NOT fit them to Portescap powered locomotives].

However, it is my opinion is that the problem lies with the cable between the control unit and the servo and part of the solution is cable management and keeping the track power circuits away from the servo interface cables. If they need to cross then they should cross at right angles. It may be beneficial at these crossing points to shield the servo cables.

Maybe we need to gather more information. We know that, for example, the signal on platform 1 twitches and so we could determine which locomotive(s) are causing the problem; what motor they use, which controller, which servo etc. There is obviously a problem with this method of signal control and that the model railway fraternity is trying to solve the problem without understanding the root cause!

I agree with your posting on setting the various units and I believe it should be possible to use a serial interface to set-up these units.

Also, I did manage to confuse the MegaPoints controller at York / S4N and it took several seconds to sort itself out. I will admit that what I did was a bit unfair but its software should have been able to cope.

My skills are writing embedded software but I know sufficient electronics to be dangerous and wire layouts, etc. (currently rewiring Pampisford).

Colin McCallum.


Return to “North East Essex (NEEAG)”

Who is online

Users browsing this forum: No registered users and 1 guest