More Experimenting with Micro Servos as Switch Machines

[C855B]

A package of these arrived this morning:



I had aborted the previous servo project as impractical. It was (correctly!) pointed out to me that the open mechanism was going to be a nightmare once ballast was introduced. I had drawn and tested a mounting box to shield everything, but by the time that was done the size was out of hand and defeated the original objective.

As you can see, this is a fully-enclosed micro-micro-micro servo (nano? ). First test is to clip off the mounting ears and CA it to the ties. Main challenge at this point is how to use the supplied horns (levers that attach to the servo shaft) to offset the throw wire. I'd drill the shaft, but if the servo is centered under the ties, the throw wire will be through one of the points.

This servo thing is now the issue in the way of real progress on the GC&W. Much incentive to resolve!

[ryan_wilkerson]

That's very small Mike! Looking forward to your progress and how well it does in operation. I have 4 Tortoise machines to install this week to get ready for the next train show.

[C855B]

It works!

Not a valid vimeo URL
Installation was much easier than anticipated. Main two kinks were drilling the hole in the servo horn close enough to the shaft to clear after trimming, and the somewhat precision bending of the tiny link.

[Philip H]

Well done.

How much will you charge a$$hats for installation?

 

[C855B]

You wish.

For the GC&W, one down, umpteen-hundred to go.   I'm just relieved it went together this easily, and the end result is not going to require nearly the trenching that the v1.x solutions had threatened.

[Steve S]

Why not do something like the pic below.  It gives you access to the servo if it ever goes bad. 

And those super, duper tiny servos are usually more expensive than the larger 5g or 9g servos.

Steve S

http://www.housatonicrr.com/images/ConstJournal10/ChineseMicroServo.jpg

[C855B]

You would have to understand the original design parameters for this project, Steve. I was specifically designing to not mount servos under the benchwork. I am using 2" foam as benchwork base throughout, and this technique avoids several issues that come with under-bench machines. Problems such as long linkages and the slop, linkages that are custom to every installation, needing a mounting system to the foam, concerns locating machines relative to bench bracing.

The goal is modular standardized construction, with all switch maintenance functions on top. Only work under the bench will be plugging/unplugging control wiring.

IOW, to replace an ailing servo, an identical switch/servo set is in spares, and I back-off the joiners on the failed pair, pop it off the roadbed, replace with the spare, and the failed module goes to the bench. Also, very high priority is standardization and the resulting mass production. I have 400 to 500 to install on my layout. Besides, I have a working hunch the turnouts are going to fail more frequently than the servos. Switch machine duty is very low duty cycle relative to a park flyer.

Cost? $3.73 each. https://hobbyking.com/en_us/hk-5320-ultra-micro-digital-servo-1-7g-0-05sec-0-075kg.html?wrh_pdp=2

EDIT: Updated link and price.

[ryan_wilkerson]

Very cool, thanks for making the video! So have you given details of the control board in another thread? Could these be easily DCC-controlled?

[C855B]

Thanks, Ryan! Yes, I think I discussed it in one of the other threads from the first of the year, but it bears repeating.

The control board you see is a stock Tam Valley Depot Singlet, and it is powered from the DCC track bus. Very easy to use. In the demo video it's connected to a Digitrax Zephyr, programmed as Switch #001, and I'm controlling it from the Zephyr.

The main downside of this arrangement is the Singlet doesn't have a built-in relay for frog polarity. The relay board adds $7, making the total operational cost close to around $28 for a total servo+Singlet+relay, plus the adapter cable for the submini connector on the servo (under a buck). This is versus a Circuitron Smail at $32 street price. On my layout I'm figuring on using the QuadLN, which with relays should make for a per-turnout cost of under $22. There's also the possibility of using an Arduino with servo driver shield, which should drive the per-unit cost down even further.

[peteski]

Maybe you could modify your design to also include SPDT micro switch (for powering the frog) which would be mechanically activated by one of the servo's arms?

[dckuk]

Nice alternative.

I am using a double or triple pole double throw switch, and using the "spare" contacts for power routing and signalling.

The modular system I just discovered will allow me to add relays to the "mix" at a later date while retaining the control boards, at which point I can cut the relay feed into the wiring in place of the toggle.

You can read about the product line: http://megapointscontrollers.com/megapoints/

[C855B]

Maybe you could modify your design to also include SPDT micro switch (for powering the frog) which would be mechanically activated by one of the servo's arms?

Yes, this solution pops into my head every so often, but three things work against it. Most significant issue is a second mechanical alignment to get right. While the sample works well, it's a bit of a trick to get the points centered, the servo centered and the linkage attached, all while keeping fingers out of the wet CA on the back of the ties. Doing this or something similar twice is sort of asking for it.

Second issue is throw distance vs. activation force. The tiny servo is good for only 0.075kg/cm of torque. Compare that to a super-cheap regular "micro" servo, at 1.2kg/cm. My impression is there's not a lot to spare to press on a microswitch button. Bear in mind the throw is limited by the points throw, the linkage is solid, not sprung, and I'm relying on servo force and a certain amount of precision adjustment to keep the points against the stock rails without stripping gears.

Third is the added size. I thought this through with the original idea, and by the time you resolve placement and the mechanical dictates, the concept is defeated. This direct-mount thing works best with the smallest hole cut into the roadbed. When there's an opportunity to move an electrical support bit out of the critical mounting area, take it.

... You can read about the product line: http://megapointscontrollers.com/megapoints/

Intriguing. These folks are new to me, too. Everything (servo, servo controller, DCC interface, relay boards) added together comes to <$20 per turnout. The controller supporting 12 servos is a great economy, tho' a downside is centralizing massive bundles of wiring. Something to think about, definitely. Thanks!

[davefoxx]

Yes, this solution pops into my head every so often, but three things work against it. Most significant issue is a second mechanical alignment to get right. While the sample works well, it's a bit of a trick to get the points centered, the servo centered and the linkage attached, all while keeping fingers out of the wet CA on the back of the ties. Doing this or something similar twice is sort of asking for it.

Second issue is throw distance vs. activation force. The tiny servo is good for only 0.075kg/cm of torque. Compare that to a super-cheap regular "micro" servo, at 1.2kg/cm. My impression is there's not a lot to spare to press on a microswitch button. Bear in mind the throw is limited by the points throw, the linkage is solid, not sprung, and I'm relying on servo force and a certain amount of precision adjustment to keep the points against the stock rails without stripping gears.

Third is the added size. I thought this through with the original idea, and by the time you resolve placement and the mechanical dictates, the concept is defeated. This direct-mount thing works best with the smallest hole cut into the roadbed. When there's an opportunity to move an electrical support bit out of the critical mounting area, take it.

Intriguing. These folks are new to me, too. Everything (servo, servo controller, DCC interface, relay boards) added together comes to <$20 per turnout. The controller supporting 12 servos is a great economy, tho' a downside is centralizing massive bundles of wiring. Something to think about, definitely. Thanks!

Go with a Hex Frog Juicer, and you can prevent this from becoming too complicated.  The turnout throw and frog powering would be two totally exclusive systems.  Besides, using a Frog Juicer makes frog polarity automatic and gives some short circuit protection.

DFF

[C855B]

Yes, there is much to be said for using Frog Juicers. I was concerned with the cost per turnout ($12 for the Mono Juicer vs. $6 for a relay, in quantity), but it does vastly simplify wiring, another objective. I'm having to avoid the Hex Juicer because of detection blocks, where nearly every mainline turnout will be its own OS block. The only place(s) I could use the Hex is in yards and other locations with no need to detect individual turnouts or crossover pairs.

Once you consider detection within the turnout, track wiring turns a little complex, even consideration for which bus or sub-bus the Singlet attaches to. Also begs the question - will a Frog Juicer affect detection? It has its own current draw when idle. A relay off the Singlet would not affect detection because the Singlet would be powered by the main or distribution DCC bus, bypassing the detection sub-bus.

EDIT: Looking at Duncan's site, yes, the Frog Juicer has to not be in the detection circuit. I don't quite know how I feel about that yet; it does add wiring complexity.

[rodsup9000]

  If you hooked the frog juicer to the main buss for the power, it wouldn't be in the detection circuit. If I understand how it works right, it works like a auto-reverser and you don't need detection for just the frog.

 Really like the way you have the servo mounted. Are going to mount the driver under the turnout too??? or some place close.