Lever Frames

[James Moorhouse]

For a description of a model mechanically interlocked frame see this video: clickable Youtube link

[dcockling]

Perhaps the bod who sees it first could copy to here!

Gavin Clark, who makes the kits, is providing me with updates as I'm in contact with several non-members who made enquiries at and following Warley.

On 7th December Gavin wrote to me: "The lever frame kits are still delayed. The issue is with the spring manufacturer, who is already working mad hours - we are only a small tiny customer for him, once every 18 months or so. If I get them before Christmas I will let you know. I expect the etches before then, but they will not be much use without the springs. I have in the meantimr upped the order to 100 kits."

Then on the 5th January Gavin wrote: "You will no doubt be pleased to know that the etches finally arrived today. I am still waiting for the springs."

I will post any further updates on this thread, and will announce when they have been delivered to Jeremy and are available for order again.

All the Best
Danny

[andrew jukes]

Thanks, David, for the reference to the description of your servo-based locking system. Reading your last para., I should be in the asylum already as my locking includes full track-circuiting and sequence locking. There's lots still to do but that side of things is relatively under control.

The pictures of the lever locks make me think a solenoid operated version should not be too difficult. Instead of the slotted 'twist to lock' stops, I would think a stop sprung upwards and pulled downwards to release a lever could be arranged. The solenoids would probably need to be staggered so they could be a reasonable diameter, with little rockers to move the stops. It's also interesting to see how much electronics seems to be associated with the lever frame. My approach would just need connections (for 20 levers) to 5/8 of an RPC SRI4 card and 5/8 of an SRO4 card.

Of course, the design differences reflect different choices on how the system can be used. My layout will be (almost) totally dependent on the computer and initial operation will use the screen display for the whole layout. A lever frame for Welwyn North will be an add-on, often not used (and with all levers locked out of use) and only able to be used when (in software) control of the relevant points and signals is switched to it.

Andrew

[Pannier Tank]

The pictures of the lever locks make me think a solenoid operated version should not be too difficult. Instead of the slotted 'twist to lock' stops, I would think a stop sprung upwards and pulled downwards to release a lever could be arranged. The solenoids would probably need to be staggered so they could be a reasonable diameter, with little rockers to move the stops.
Andrew

I've had thoughts of a similar system. Lever Frame connected to tappets with two holes (Normal / Reverse) for the solenoid pin to lock the levers. The lever catch handle would operate a microswitch which would be connected to the Electronics / Logic Board (in my case a PLC), if the proposed move is valid, the solenoid would energise releasing the lever. Once the catch handle is released the solenoid would then lock the lever again. One solenoid for each lever.

[grovenor-2685]

This is the problem with solenoids which the use of servos neatly avoids. To avoid having a lot of solenoids permanently energised when levers are free additional contacts are needed to signal the desire to move that lever, ie. your suggested microswitch on the catch handle. Don't forget you still need to know the lever position so need at least one other microswitch per lever so you are doubling up on the mechanical complexity and on the number of inputs needed to the controlling logic.
Regards
Keith

[andrew jukes]

Neat, David - that deals with the worry of having several solenoids always on.

In my case, the solenoid feed from the RPC SRO4 card would simply be taken via the catch-handle microswitch. The mechanical design would need to be able to cope with a catch handle being released with the lever part-thrown - but it needs to be fairly robust anyway.

Not sure what Keith means about doubling up on the inputs to the controlling logic. My picture of how this would work is that the computer interlocking would provide an unlock output to a lever which would only power the solenoid if the catch handle was pulled. The lever position needs to be sent to the interlocking anyway - so one input bit for lever position (normal/reversed) and one output bit for the solenoid (locked/unlocked). There are all sorts of things a real signal engineer would not like about this arrangement (e.g. a part thrown lever could be locked but still able to be moved either way) but most of these will be caught by the interlocking logic which needs to treat a wrong-side input from a lever as no different from how a mouse click on a locked icon on the screen would be treated - i.e. ignored.

Andrew

[Pannier Tank]

David has designed a neat way of detecting the Catch Handle being released on the Scalefour Lever Frame:-
http://www.rmweb.co.uk/community/index. ... ver-frame/

[The_Prisoner]

Thanks for the many replies - I now have much to think about.

I hadn't thought about using servos to do interlocking. I was thinking of interlocking as a separate subject.

My main idea was to get away from the on/off (or crash/bang) approach of most solenoid-operated electrical systems and provide a cheaper solution than tortoise (and similar) motors. I have no intention at this stage of trying to model signal bounce - I'm looking to produce something that could be built by the average P4 modeller or I could easily provide ready built.

I seems like a lever pitch of about 1/2-in is fairly standard. I now need to resolve the problem of the various throws.

Peter

[barhamd]

I can't see how you can do any better than a servo as a replacement for a solenoid for much less than the price of a tortoise equivalent. But then I can't see how you could use servos in a method much better than the MERG servo4 board which drives 4 servos each from an 'on-off' switch and allows setting the throw and speed of each separately. To be honest it sounds like you are trying to reinvent the wheel.

For details on how I've used servos (and I know there are plenty of other ways to do it) see http://www.rmweb.co.uk/community/index.php?/blog/186/entry-8275-fitting-tous/

I guess I could just see a market for
[*]a friendly, non-electronics scary device that takes a single input and then controls a single servo,
[*]in reality that would have to be a +v 0v and signal v so that the servo had some power to turn it both ways
[*]maybe with a pair of variable resistors to set the throw with a screwdriver so people don't have to worry about a 'setting box' or PC
[*]of course you would need to get it CE certified before you could sell it,
[*]and you would need to be able to price them for under say £2 each.

Good luck!

David

[andrew jukes]

Going back to the idea of using little solenoids to lock the levers and so simulate a properly locked lever frame, it seems finding suitable solenoids could be difficult. Small solenoids, or coils suitable for solenoids, seem to to have largely vanished from the lists of such as Rapid, RS, Farnell or Maplin. What I would need is 30-odd, suitable for 12V or 24V, probably at least 250 or 500ohm respectively - any suggestions on other possible sources?

I suppose I could tear apart some surplus relays....

It would be ironic if I ended up using servos simply because suitable solenoids have become extinct!

Andrew

[Pannier Tank]

Going back to the idea of using little solenoids to lock the levers and so simulate a properly locked lever frame, it seems finding suitable solenoids could be difficult.
Andrew

http://uk.rs-online.com/web/p/dc-d-fram ... d/0347652/

A bit expensive, also as it's a 'pull' type solenoid, it will need some pins turned up to complete.

[Jol Wilkinson]

Going back to the idea of using little solenoids to lock the levers and so simulate a properly locked lever frame, it seems finding suitable solenoids could be difficult.
Andrew

http://uk.rs-online.com/web/p/dc-d-fram ... d/0347652/

A bit expensive, also as it's a 'pull' type solenoid, it will need some pins turned up to complete.

A rather similar solenoid is available from Andrew at MSE/51L, and is somewhat less expensive (although it doesn't include the spring). The pin is not captive and could possibly be turned down to suit your needs.

Jol

[Russ Elliott]

We used that RS solenoid on Green Street's ground signals. Not sure what the resistance is, but I don't think it is as high as Andrew wants. It is 100% continuous rated though, which would be essential for the lever frame application.

[LesGros]

For a description of a model mechanically interlocked frame see this video: clickable Youtube link

Well worth a look if you are thinking about a plan to build interlocking. The presenter has an easy conversational style as he describes the layout, tables and charts. A note pad will be useful to capture relevant gems of information.
The slide lever mechanism he employs has the virtue of compact mechanical simplicity, with the complexity of the locking stemming from the layout requirement being relatively straight forward to follow
Also worth noting is the teamwork element, in that the locking chart was prepared by an expert in the subject, and the layout builder is, clearly, a skilled and imaginative 4mm modeller.

one question arises, who is the builder?

[paultownsend]

The pin is not captive and could possibly be turned down to suit your needs.

If you have a lathe to do this it would be possible to make your own solenoids at very low cost.
Bods used to do this often and easily. It is also feasible with mini electric drill clamped in vice and running at slowish speed.

I made some yonks ago for AJ uncouplers this way (strictly not solenoids as armature was fixed, but that is only a matter of clearances in the coil former).

The advantage of brew your own is that you can make the oomph to be just what you need from your chosen volts. If you use commercial units, even if you can find some, they may be too energetic like old fashioned solenoid point motors. For Lever frame locking you should need very little force as the solenoid needs only to move its own core and do very little "external work". If I were to try this, I would be looking at winding solenoids with thinnest enamelled copper wire that I could handle reliably and use a dedicated cheap 5v supply. Armature would be bits of nail or better... soft iron flower wire. Options would be vertical solenoid where oomph must compete with gravity or horizontal solenoid and a simple crank to give vertical motion if required...maybe design locking plunger to operate directly horizontally...just thinking aloud at this stage.
Once you choose the wire on practical grounds you dont need to be very accurate with precise laying in or counting of the turns because this application is undemanding.

If anyone asks I could do a prototype or two and come back with a design recommendation. If no-one asks this will still happen but be way down my modelling task list!

Before Keith jumps on me, I am also trialling servo interlocking but suspect the solenoid method may be cheaper, smaller, simpler. Both should be viable but the solenoid option could be driven directly by ACC8 without needing servo4f s too. and yes I am aware of the version of Acc8 with servo firmware to eliminate servo4f s.
I suggest different users will like the choice of solenoid or servo.

[Russ Elliott]

...as the solenoid needs only to move its own core and do very little "external work".

I'm not so sure about that. If a lever is locked by a wedge-shaped wedge thing (see my earlier sketch), then a reasonable amount of force is needed to keep the wedge up against the sides of the lever sufficient to let the user know that it shouldn't be pulled. (That force could be gravity, or the solenoid in its energised state, depending on what one's view is on what the default/'fail-safe' condition should be.) The solenoid approach appeals because it is instantaneous (compared to the time of a servo), but I suspect the price to be payed will be in amps. A rack of old H&M SM3s would do the job, but the real estate required would be large.

[paultownsend]

We used that RS solenoid on Green Street's ground signals. Not sure what the resistance is, but I don't think it is as high as Andrew wants. It is 100% continuous rated though, which would be essential for the lever frame application.

RS at £10 and MSE variety at £7 both quote 12v at 1/4 amp so 48ohms. RS quote 35gram force

Doubtless both are suitable for the real work of signal operation, but are I think still over powerful for frame locking.
Andrews requirement for higher resistance could be met by DiY coils.
Dave's double hole arrangement eliminating "nearly always on" solenoids makes the solenoid design even easier as no overheating worries.

...as the solenoid needs only to move its own core and do very little "external work".

If a lever is locked by a wedge-shaped wedge thing (see my earlier sketch), then a reasonable amount of force is needed to keep the wedge up against the sides of the lever sufficient to let the user know that it shouldn't be pulled. (That force could be gravity, or the solenoid in its energised state, depending on what one's view is on what the default/'fail-safe' condition should be.) The solenoid approach appeals because it is instantaneous (compared to the time of a servo), but I suspect the price to be payed will be in amps. A rack of old H&M SM3s would do the job, but the real estate required would be large.

I agree if you need a wedge action, but Dave "Pannier tank" suggested;

Lever Frame connected to tappets with two holes (Normal / Reverse) for the solenoid pin to lock the levers. The lever catch handle would operate a microswitch which would be connected to the Electronics / Logic Board (in my case a PLC), if the proposed move is valid, the solenoid would energise releasing the lever. Once the catch handle is released the solenoid would then lock the lever again. One solenoid for each lever.

I suggest the only "external work" required of the solenoid would be overcoming friction and possibly a light spring.

My point is that H&M type solenoids are overkill for lever frame interlocking.

Approxiamate calculation suggest a DiY coil of 900 turns of 40swg en Copper wound on a 3mm core with a coil lengthof 20mm and OD of coil = 10mm say Overall OD =12mm could deliver around 15gram force from 10v supply and the resistance being around 240ohms, current draw will be about 40 mA. ( max for this wire)
That force is around half the quoted figure for the RS coil mentioned by others at a £10 hit. I expect the Wizard/MSE one is similar strength at £7.
The DiY core could be a piece of c. 2mm nail or better some soft iron flower wire running in a brass tube. Varying some dimensions would allow an overall OD of <10mm to suit SHAG frame lever spacing.
Such a coil would be very cheap and quite easy to make, ideally on a lathe but I have done similar for A-J uncoupler magnets yonks ago with just a vice mounted minidrill at slow speed.

This is just a first stab at rough coil design to assess orders of magnitude. Solenoid design for a required force is quite complex and must be backed by trials as some dimensions relating to the core and stroke are quite critical.

I will try and do a trial of this within in a few days, while trying to finish a Lo-Rider build!

[andrew jukes]

Thanks for all the helpful input, especially Paul's thoughts on home-made coils.

My first sketches of a way of doing this would have a locking peg, lightly sprung upwards against a stop, which would prevent movement of the lever just like the servo arrangement shown earlier. It would not be wedged against the lever, so to some extent Russ's concerns about the unlocking force needed don't apply. It would still be possible to jam the stop and prevent it unlocking by holding the lever against it but even this could be prevented (with extra complication) if the catch handle latched the lever mechanically at each end of its throw, so the solenoid would be energised by the catch handle fractionally before the lever became unlatched.

I thought a good arrangement might be to mount the coils on the base of the lever frame with their cores projecting through the base, each pulling up one end of a small rocker when energised. The rockers would be pivoted on a bar mounted underneath the base. The other end of the rocker would then pull down the locking peg against the resistance of the spring. I wanted an arrangement that would allow the coils to be staggered, assuming the coil diameters would be greater than the lever spacing. Choice of spring, the arm lengths of the rockers and the design of the coil would all needed trying out and modifying, probably using a single lever test rig.

On coils, I do have around 20 DPDT relays with 12V 250ohm coils. Pulling one to pieces could be a good starting point for such tests (trouble is, there is much else to do of a far higher priority....)

Andrew

[James Moorhouse]

For a description of a model mechanically interlocked frame see this video: clickable Youtube link

one question arises, who is the builder?

The builder is Peter Jordan. He's very active on the Signal Box Forum.

[paultownsend]

Paul's thoughts on home-made coils.

My first sketches of a way of doing this would have a locking peg, lightly sprung upwards against a stop, which would prevent movement of the lever just like the servo arrangement sh

On coils, I do have around 20 DPDT relays with 12V 250ohm coils. Pulling one to pieces could be a good starting point for such tests (trouble is, there is much else to do of a far higher priority....)

Andrew

So I finished the High Level Lo-Rider and spent today on first coil trial, details below but before that I respond to Andrew's 2 points above...
1. locking peg needn't be sprung, see my trial using gravity, but mechanical detail needs quite a bit of work to optimise.
2. 12v 250ohm coil ex relay may be useable but probably not as a solenoid 'cos it is unlikely to be easy to separate the armature to become a moving plunger. However use of relay via an extra lever is feasible but adds complexity and size.

To my trial of today...just a proof of concept rig has been built to check my arithmetic re enough force from a small DiY coil.
The simple answer is yes it can work so I offer the trial info. with piccies etc so anyone who wants to follow it up with a more engineered mechanical arrangement...my coil mount and operating arm with holes need to be etches for consistency.

The coil made has changed in size a bit from yesterday's proposal to
a) make diameter less than 10mm to match Shag frame lever spacing and
b) to use my available but sub-optimal materials in stock.

Coil is wound with 40swg enCu wire soldered at ends to 24swg stubs to give a little robustness to the prototype. It is on a former of nominal 1/8 " Plastruct tube, 30 mm long with cheeks superglued in place. The cheeks are built up from 3mm fibre washers and 9mm OD nylon washers of 4mm bore. They are 25.4mm apart on outside length giving 23.5mm inside the cheeks for coil winding. Piccy 1 shows the assembly ready to transfer to the lathe where the approx 3.8mm snout was lighly chucked and the outer end supported by tailstock centre. My lathe was set to run at about 360rpm. for reasonable fingere control of the wire layup.
Because this was a quick test I didn't take much care with even layers, broke the wire once and had to splice and sleeve the joint. In practice I got much less wire on than I had hoped. I forgot to take a photo of the lathe setup so after completion put it back to simulate, thus piccy 2 shows some outer insulation and the second stub wire which were both actually added off the lathe.

Piccy 3 shows the lever frame with locking attachment thrown onto the front, here simply because it is on my Cbus/servo/bouncer test rig and I had no space behind the frame.
Piccy 4 shows the components....the locking bar has 2 2.5mm holes spaced about 5mm intended to match the lever throw. I had to elongate one so 4.5 mm would be better...this bod measured the throw at lever tip but the Shag shoe rides a little way up and reduces the throw a tad. The coil armature is half a panel pin measuring 1.6 x 19.2 mm and drops by gravity into the holes. This basic idea came from Dave above in this thread. Operating the coil lifts the weight of this pin up into the coil cleanly thus unlocking the frame. No springs needed but some way of stopping the locking bar rotating on it's 0.5mm wire thru' the frame is needed. The clearances on this trial version are generous, the 1.6mm pin operates into the 2.5mm hole so when locked the lever can move a little, but not enough to operate the microswitch, and feels locked. The coil mount is only plasticard on woodscrews and spacers so is not rigid enough. As I said above etches for the coil mount and locking bar would make this satisfactory and the pin/hole clearance could be reduced.

As this was primarily an electric test here are the operating parameters achieved.
The coil is 75ohms, I ran it on 5.2v causing a current of 63mA. This is 50% more than the continuous rating for the 40swg wire so the coil is only suited to intermittent operation, but that is what is required here. I tried 3v to limit the current to the correct max.of 40mA but then didn't have enough oomph to lift the armature pin as used here. I have no way of measuring such a small force but the pin weighs about 0.3 grams. Even tho' the pin has clearance into the locking bar holes and is a good sliding fit into the coil former, friction is probably still the greatest load! A puff of graphite lock lubricant could be useful.

I count this crude test a success for proof of concept and demonstrating that a cheap small coil is sufficient, but would need these improvements in a production run...
1. Etches as mentioned.
2. Method of inhibiting locking bar rotation
3. More coil oomph to give a bit more force margin....a fatter coil would be better magnetically so perhaps stagger alternate coils so the OD can exceed 10mm. Use slightly fatter wire, aiming for a load current toward 50- 100mA without exceeding wire rating so inadvertent continuous powering would not burn it out. Improve design of core with thinner walled tube so fatter armature. Aim for 200ohms +/- 50ohms.

Although I could try other coil designs if anyone asks I won't be pursuing the required mechanical improvements for ages as I have other priorities for my modelling.

[andrew jukes]

That's really helpful, Paul - and much simpler than the way I was thinking. I like the way the work done by the coil is kept so minimal.

I'm in a similar position in doing more. Much as I would like to get on and build a locking lever frame, getting my phase 1 track-laying finished is slightly more important!

I think staggered coils will end up being necessary. For my MERG RPC-based system, for electrical simplicity the coil current needs to be kept below 50mA and I suspect a bulkier and higher voltage design (12V would be ideal) would work best. I have pulled apart one of my relays - the 12V 250ohm coil is just under 10mm diameter but its frame is 10.5mm wide and there is still the problem of a solid core.

Seems to me, though, that a solenoid-locked lever frame can be assumed to be feasible, using one input and one output bit per lever from the RPC system - which is good news.

Thanks again.

Andrew

[paultownsend]

for electrical simplicity the coil current needs to be kept below 50mA and I suspect a bulkier and higher voltage design (12V would be ideal) would work best.

In the MkII prototype trialled above ( MkI design was scrapped before winding) I was surprised that I only achieved 75ohm resistance when the design predicted 240ohms.
So I checked my calculations for the MkII 9mm dia coil as built and found a x2 error (whoops) so theory actually predicted 120 ohms for a perfect coil with no insulation. In practice one allows for 25% lost space due to enamel insulation and scrambled layering so my MkI should have got to 90ohms. The unwanted wire break/joint/sleeve takes up some 15% of available space so recalculated predicted R= 76.5 ohms Thus theory and practice agree now...phew!

Recalculating for a MkIII coil of 15mm OD and 20mm length has a predicted resistance of 270 ohms allowing for the practical departures from theoretical design. This implies two staggered rows of coils required for the Shag frame or ...
MkIV coil has OD = 20mm ( 19mm winding dia. on same 3mm core ), using the same 40swg wire and coil length reduced to 15mm gives a predicted resistance of 295 ohms allowing for the practical departures from theoretical design, thus 12v at 44mA for MkIII or 41mA for MkIV, both meeting Andrew's requirement for RPC and both would have more pull than the MkII tested.
I will give MkIV a whirl when I have found some time but not today. 20mm cheeks will probably be squares of .040 " Plasticard for simplicity giving overall coil assembly length of 20mm including the snout. As this was only used for the chucking and could be sawn off after turning ( or turn on a mandrel so snout is redundant) a length of 17mm seems achievable.

I thought it time to look at costs as the incentive for DiY coil was to avoid silly £7 or £10 prices.
My MkII was wound with stock wire left over from a previous life, so effectively free, but that is cheating!
Annoyingly the three major electronic distributors that I deal with ( R-S, Farnell, Rapid ) now only stock en Cu wire of 36 swg and fatter.
A quick calc. for MkV coil using this shows for size 20mm dia and 20mm long a resistance of only 60 ohms. This will be overpowered and current too high.
Cost is about £20 incl Vat and delivery for 500gram 1850M reel which would do for 18 coils of this unwanted 36swg.

I have looked at wholesaler wire specialists who can supply 40swg wire and due to volatile copper prices they don't show list prices any more so you have to go thru' the boring quotation process which I am currently too busy to do. However the manufacturing cost is mostly copper weight so I expect it to be around £20 for 500gm which would yield 4800M and make 24 MkIV coils
Thus material cost is predicted to be below 50p per coil.

Of course there is a fair bit of work in making these coils...about 3 hours for me to make MkII which would reduce to less than 1 hr each for a production run.
Looking at the time to assemble a Shag kit I suggest the additional 5hours for coils is acceptable and seems to reduce the cost of 5 MSE coils at £7 each from £35 to £2.50.

Servos cost about £3 each so £15 per frame and unknown build time for linkage.....Dave Barham tell us please?
Size of servos is 30x30x10mm = 9000cu mm against 20x20x20= 8000cu mm so nowt in it.

Both servos or solenoids assume use of CanAcc8 to drive them with firmware available for either version and avoid use of servo4f boards.

Having gone through this exercise, it seems my suggestion of DiY solenoids certainly minimises cost, but adds an hour/lever minimum to build time. How does that compare to servo linkage build time Dave?

Perhaps the best option for simplicity of linkage, minimal cost and build time is for some hero to locate a source of naked coils of similar spec. to my designs but without the framing, springs etc that make the RS/MSE versions ludicrously expensive (IMHO) for this job.
Is there any mileage in getting a manufacturer to quote for making a few hundred?

[andrew jukes]

If the pitch of the SHAG frame is 10mm, best to aim for 18 or 19mm OD, to fit with a staggered arrangement.

My relay coil looks as though it may be possible to push out the core after drilling out the rivet at one end. That way, I could end up with a close to ideal 12V coil. Something to try when I'm back home - and if successful I will do some experiments with it with a locking pin.

If that doesn't work out, I would eventually be interested in 30-odd coils (or material to make them with myself).

The sliders could be etched, and if they were it would probably be worth etching a simple frame to keep them flat and in line and provide mountings for the solenoids. It would risk starting to look like an interlocking frame........

One other thought - the microswitches that indicate (to the computer) the normal/reversed lever positions would best switch at mid-travel of the levers to avoid some odd asymmetry in the response of signals and points. That might mean the easiest way would be to make the sliders work the microswitches.

Andrew

[barhamd]

The servos I used were DYS0201 from Giantcod http://www.giantcod.co.uk/dys0201-micro-servo-torque-p-402575.html at £2.83 each.

The method of locking with a two lengths of of K&S tube, say £2 each for the length which did 20 levers, I made all the mechanics for the locking of 20 levers in an evening. There were also two lengths of 1/4 angle which had the short lengths of the tube in which the slotted lock rotates.

I reckon the total cost of the ironmongery was well under £10.

Hardest job was replacing the plugs supplied on the DYS0201 with molex type connections to connect to the MERG board. These probably cost < £1 and were done in a Pub garden during a meeting of the CHEAG.

David

[Alan Turner]

Hardest job was replacing the plugs supplied on the DYS0201 with molex type connections to connect to the MERG board. These probably cost < £1 and were done in a Pub garden during a meeting of the CHEAG.

David

??????????????????????????????????

But standard servo plugs (sockets strictly I suppose) fit the MERG board directly?

Alan