Controlling Semaphores, driven by servos? Or tortoises?

[bluestag]

Hey guys,

 

I'm gearing up to build a few semaphores for the layout.    I intend to control them with the same lever switches (they mimic prototype levers) which are fitted with a wafer, which has five sets of single pole double throw switches on them.   The turnouts are driven by tortoises.   They use one set of switches, which feed either positive or negative half wave DC to the tortoises.   I will only consider a solution that uses the same switches.    But they could be wired to be a reversing switch, ie feed in full wave DC and  reverse the positive and negative with a throw on the switch.   There is no turning these off.   They are on-on.   Does anyone make a driver for solenoids that will respond to positive/negative messages in either half wave or full wave DC?   Constantly on?  Or must I use tortoises?   With a Rube Goldberg mechanism?

 

A photo of my frame:

 

 

[Ray H]

Is there a centre off position on the switches?

[Jeremy Cumberland]

SPDT switches can be wired in a variety of ways.

 

For simple on/off, connect the middle terminal to the supply and one of the outside terminals to the device.

For on/off with a device that requires a discharge, such as a capacitor, connect the middle terminal to the device, one of the outer terminals to the supply and the other terminal to return. [Edit: this won't work if they are make before break contacts - mentioned below in this thread - but this is an uncommon configuration.]

For on + / on -, connect the middle terminal to the device and the outside terminals to supply + and supply -. [Edit: again, this won't work with make before break.]

For operating two devices where one is always on when the other is off, such as a two-aspect signal, connect the middle terminal to the supply and the two outer terminals to the two devices.

 

Are your switches wired together in some way, so you can't configure each individual switch as you want?

Edited June 16, 2023 by Jeremy Cumberland

[Chris White]

I am unaware of any solenoid motor designed to use a continuous current. If it's just the switch you want to stick with but are prepared to do different wiring then there are a number of circuits designed to take a simple "on/off" input, i.e. just using one side of the  and based on that changing output the appropriate pulse to operate a twin coil solenoid motor.

 

Your title mentions servos, whereas you posting does not. Here again you would need to employ a circuit that would utilise one side of the switch for an "on/off" input and output the appropriate command signal to drive a servo base on the state of the input.

[bluestag]

Thanks guys.   Yes, there is a solution to drive servos with a board that allows an on-off.   No, there is not center off, just as there is no center position for a prototypical signal lever.

[Michael Hodgson]

1 hour ago, bluestag said:

 

No, there is not center off, just as there is no center position for a prototypical signal lever.

Not true I'm afraid - some prototype signal levers do have a mid-position, although it is relatively unusual and unlikely to be wanted in your model.

 

The NER (until they fell in line with the other main line companies) used to clear subsidiary (call-on etc) signals by pulling a lever half-way, and to clear the main signal as well, they pulled it all the way over.  This was because they didn't like the idea of passing a signal (in this case the subsidiary) at danger even though the main arm was off.

 

Some boxes had what were called Push-pull levers.  The Normal position of the lever was central rather than back in the frame, and you clear the signal for one direction (typically a shunt over a crossover) by pulling in the usual way, while pushing the lever back in the frame  cleared the signal for the shunt in the opposite direction.  Naturally you only used this technique where you would never want both signals off the signals at the same time, and putting both on one lever rather than two had the advantage that you didn't need to provide locking between these two levers.  However I've always thought push-pull levers inelegant, as they do make for an untidy looking frame when there are no trains about.

 

Both approaches saved a lever, and with several of these in the box, both the lever frame and the building structure would be shorter, so it also saved money.  King levers (used for rearranging the interlocking when closing a box for the night whilst allowing trains to run through) might also have a mid position, again something you're unlikely to want to model.

 

Levers could also have a couple of other intermediate positions - not quite fully normal, used for back-locking, and not quite fully reverse, typically used to ensure that an electrically operted device had motored over.

 

Some lever quadrants had more than one notch in which you could hold the lever in the reverse position - this could be used as a way of allowing for temperature expansion in the operating wire.

 

Whilst lever switches do bear a visual similarity to prototype levers, they don't replicate the "feel" of a lever frame, because most prototype levers have an additional catch-handle and the interlocking physically prevents you from moving the lever when it is inappropriate to do so, which is something you can't easily replicate with switches.  When using switches, electrical interlocking has to be more complex to stop you making conflicting movements, as it has to be done by not acting on inappropriate operation of the switch - the switch can still be moved but its associated signal doesn't clear. 

 

But interlocking is something that usually isn't modelled, it's seen as overkill because our models don't represent a threat to the safety of the travelling public.

 

 

[Ray H]

My question about centre off was intended to ascertain whether the switch was either ON (when switched one way) or OFF when switched the other, or whether the two travel extremities of the switch's travel were both ON in which case there had to be a mid point in the travel where the switch changed from one ON state to the other ON state and thus had at least somewhere in the travel when both ON states were ON or was momentarily in an OFF state.

 

A switch that doesn't change the ON state to another state isn't technically a switch as it might do something mechanically but it doesn't electrically.

Edited June 15, 2023 by Ray H

[MartinRS]

What about using a small (older style) encapsulated relay? Just glue an actuating arm to the armature. Experiment to see if the relay has the power to overcome friction in the system and can move the (low) mass of the arm and linkage system. If you are switching polarity, rather than switching the voltage on or off, place a diode in series with the coil.
 

Edited June 16, 2023 by MartinRS
Fix typo

[bluestag]

Well, despite the rare instance of prototype leavers not being binary, MY switches are.   As I bought them, they had two wafers with a total of 13 spdt switches on them.   What in heavens name they were for, don't ask me.    I take the outer wafer off and have five spdt switches in the end.    To control Tortoise switch machines only requires one.   Two common wires carrying either side of half wave DC, and a wire to run to the Tortoise, which gets a common return back to the transformer.  

 

Now I have learned that one spst is all that is necessary for controlling servos.  A common wire and one more wire per servo.   Good.   Few wires is good.

 

Soon (in the larger scheme of things, say a year to build the signals) I'll have some red levers in the frame as well.

 

I have seen mechanical interlocking on a layout, and it looked attractive.   And time consuming.   I have PLENTY of other modelling projects, I decided to do without.

[DCB]

Tortoises are too slow for any realistic representation of mechanical signalling or point operation.  Solenoids are generally too quick blink of an eye, 1/5th of a second, 1/10th maybe when you want about 1/2 a second.   I used a motor in my experiments operating through 270 degrees to move a Ratio ready made semaphore operating wire about 1/8th inch,  There was too much stiction. I had to use too much voltage and it went too fast, better than a solenoid, far better than a tortoise but too fast Plan b will be a geared motor operating an operating arm working through about 270 degrees which will again raise the operating wire about 1/'8th inch.   Getting that 270 degrees in half a second is the challenge.  Putting two capacitors one for up one for down in inverse parallel would allow on/on switching but would put a huge load on the system at start up. My solution is that the up cycle can only start with a pulse when the signal is down and vice versa.  It's all good fun but it's no accident that Triang etc chose passing contact switches for signals and points when they wanted the levers to be on/on

Edited June 16, 2023 by DCB

[bluestag]

DCB.

 

I don't have time for that sort of building.   Apparently servos have been proven.  And when driven by proper circuitry, run slow enough.   I am trying to get a finished layout while I still can.   And I have a three arm bracket to build.   I am told I can get the mechanism under the signal such that I can build it on the bench, and drop it into a hole in the layout, rather than crawling under the layout.   Going under the layout should have ended a decade ago.

 

Kevin

[Grovenor]

And there are a number of sellers of suitable servo mounts for this application,

Wizard models/MSE for one and Dingo mounts for another.

[Longhaireddavid]

Surely one Arduino could control a few servo/ signals with a bounce for each arm built it?

[bluestag]

58 minutes ago, Longhaireddavid said:

Surely one Arduino could control a few servo/ signals with a bounce for each arm built it?

I likely will use a MegaPoints product.   They have a twelve channel all singing all dancing board for reasonable money, that does not require programing as such.

[Harlequin]

MTB MP1 point motors might fit the bill nicely.

They are slow acting and have internal switching to turn themselves off at the end of their travel. That means they can be connected to a SPDT switch for simple operation or controlled electronically if required. Megapoints do a board for them.

 

[bluestag]

2 hours ago, Grovenor said:

And there are a number of sellers of suitable servo mounts for this application,

Wizard models/MSE for one and Dingo mounts for another.

I like the dingo.   Compact, and they have a triple servo mount, as I need a triple arm signal.

[Ray H]

Nice, easy, etc., the various suggestions may be but the way I read your first post you were looking for a way to operate whatever system you choose to actually operate the signals.

 

It needs to be something like a switch, something to change an electrically state either directly or indirectly. You can, for example, have an all singing means of make the signal arms move but you still need something to initial the system (like servos) that will make the arms move. As good as AI is at present, they haven't yet got a system that works without some kind of human intervention (even if that's just initial data set up).

 

You indicated earlier that your switches/levers were ON/ON with no centre off. If there isn't even the slightest moment when the switch changes from one ON state to the other, the switch isn't actually doing anything. There doesn't need to be a centre off state that could pause the lever's movement momentarily and that the operator can detect, but there has to be a point in the lever's travel when it changes state with the possible exception of a make before break device which, itself, can still be electrically detected.

[Grovenor]

Ray,

I think you are adding confusion here, Bluestag has clearly described his switches, (they sound like typical telephone switchboard items) and that they comprise a stack of SPDT contacts. Whilst these types of switches did occasionally have make before break contact sets it doesn't matter if they are being used as SPST to operate servo drivers such as Megapoints.

 

[bluestag]

2 hours ago, Grovenor said:

Ray,

I think you are adding confusion here, Bluestag has clearly described his switches, (they sound like typical telephone switchboard items) and that they comprise a stack of SPDT contacts. Whilst these types of switches did occasionally have make before break contact sets it doesn't matter if they are being used as SPST to operate servo drivers such as Megapoints.

 

What he said.

[bluestag]

On 14/06/2023 at 01:00, Jeremy Cumberland said:

Are your switches wired together in some way, so you can't configure each individual switch as you want?

One spdt on each is tied together, with the plus and minus of the half wave dc feeding the tortoises.   That leaves four spdt switches that are not committed yet to any purpose.   So for the signal levers, I can have the signal wire and the return on a spst.   Easy peasy.

[Ray H]

12 hours ago, Grovenor said:

Ray,

I think you are adding confusion here . . . . . 

 

Apologies, I missed the "which has five sets of single pole double throw switches on them" bit of the initial post, possibly thrown by the reference to "(they mimic prototype levers)" in the same post and the reference to ON/ON in a subsequent post.

[Stephen Freeman]

Personally I prefer the MERG Servo4 board using SEMA4 firmware. I have used a few of the others but find the results push me back to them

https://youtu.be/kGUqtGuV7_k  or https://youtu.be/7K3NuoTng2Y for instance.

[KingEdwardII]

I'd certainly recommend slow acting motors like servos for this purpose, especially if you want stuff like bounce in the action.

This then requires a servo driver board - people have mentioned MERG, but there are also some commercial offerings available.

 

Yours, Mike.

[Chris White]

1 hour ago, KingEdwardII said:

... people have mentioned MERG, but there are also some commercial offerings available. ...

MERG is a group of enthusiasts so whilst the Servo4 is a much lower cost than many commercial offerings you do have to build it yourself from a kit and support from the community is readily available but there's no warranty. The Sema4 firmware will allow a bit of bounce to be simulated but some find it challenging to use. For those not keen on spending some time down rabbit holes best go with one of those many commercial offerings now out there.

[Jol Wilkinson]

On 24/06/2023 at 16:14, Chris White said:

MERG is a group of enthusiasts so whilst the Servo4 is a much lower cost than many commercial offerings you do have to build it yourself from a kit and support from the community is readily available but there's no warranty. The Sema4 firmware will allow a bit of bounce to be simulated but some find it challenging to use. For those not keen on spending some time down rabbit holes best go with one of those many commercial offerings now out there.

I have built a number of MERG kits, including Servo4, Servo1 , Servo1R and the relevant setting Boxes. I have had no problems and prefer the remote setting boxes, rather than having to press buttons and count LED flashes to set things up.

 

If you want to use ready assembled units, also take a look at Heathcote Electronics. The Retford signal servo mounts from MSE are also very good.