Large Scale Central

Switch Machine Project

Track and Trestles
#121

John Caughey said:

An early fix was to have a return spring on the far side (of switch) handle the push (by pulling) part and the servo the pull.

Is it done yet? yer bud,

John

Howdy John, and thanks. Here’s what I said to Skeeterweazel who brought that up on MLS:

Hi Marty, if you’re suggesting that a far-side spring would help keep the cable taught and therefore reduce binding, that makes sense. Not sure how that would affect the driver board’s auto-calibration thing, but it’s worth checking out, thanks for the thought. I don’t see an easy way to get a spring in there, so I’ll hold off and see how well the smaller tube does tomorrow.

I was just looking at the setup, and no matter what springs you put in, the auto-adjust will keep adjusting the servo position to a certain torque value and then stop. So it’s hard to judge what’s doing what. So I’d like to get the “flexible throw rod” (cool term, Dan!) moving as freely as possible, if only to reduce motor strain.

#122

Dan Padova said:

I can see alot of potential for your creation. I never thought of a flexible throw rod as long as the one in your photo. Great idea.

Thanks Dan, that’s encouraging! And I like your term. Ideally, it could be 10’ or more in length, to permit a clustering of drives in a single box / structure, with a common power source, and perhaps a shared RC relay control board.

#123

I received the smaller sheathing tube, and cut a 10’ length of that and the 1/16" SS cable. Too much resistance still. 4 or 5 ft would work, but I really would like to get that friction down as far as possible.

So, as some prognosticated, I’ll get a coil of spring wire, and mess with that next. Actually 3 coils, in different sizes for comparison. McMaster has it in SS for a great price, but as always, the bummer is the shipping charge.

More later…

C

#124

A little update.

Last weekend I compared the cable to spring wire (.031, .041, .051 dia’s, with and without spray teflon lube), run thru a 1/8" od x .073" id nylon tube. The .041 spring wire worked best, but was still too sticky for the switch machine to happily deal with. Still, that’s what I used for subsequent testing.

By “happily,” here’s what I mean. The controller board has an auto-adjust routine that, when it senses the servo having too much resistance, calls that the end of travel. Then it goes the other direction, and does the same. If there is too much resistance in the cable, the board will think that’s the end of travel, and the points won’t be thrown all the way.

I tried 8’, 6’, 4’, 2’ and 0’ cable lengths, with straight, and (as possible) 180-degree, 360-degree, and reversing 180-degree paths, using a digital scale to read the peak force needed to throw the turnout. The more extreme the cable path, the more resistance, and force required, as you’d expect.

The threshold resistance for the board (Tam Valley Depot “Singlet II”) is about 1/2 pound. The lengthier / more extreme cable paths required more force than that, so the board didn’t auto-calibrate for those.

So, I ordered 2 sample push-pull cables from RC product manufacturers, to see how the pro’s do it. Today I compared everything, as follows (dimensions are approximate).

  1. Sullivan cable: .032 SS plated cable (7x1x.014) in nylon tube (.070 od x .040 id)
  2. Dubro cable: .06 nom steel cable (7x7x.006) in nylon (?) tube (.125 od x .095 id)
  3. Spring wire: .041 SS spring wire in nylon tube (.125 od x .073 id)
  4. McMaster cable: .06 nom SS cable (7x1x.020) in nylon tube (.125 od x .073 id)

Here’s the setup, using 4’ runs, no turnout attached,

Average push/pull forces were:
1: .10 lbs
2: .14 lbs
3: .41 lbs
4: .32 lbs

Long story short, #1 was too flexy; #3 & 4 (the ones I’d messed with last weekend) were too sticky. #2 seemed the best, but the tube is unmarked, so I don’t know where to get it in bulk. So I tried the Dubro cable and my prior tube, and it worked as well.

Here’s a brief vid of it working.

https://www.youtube.com/watch?v=-d6h2Xk_sWE&feature=youtu.be

I learned several things through this, which I’ll type up a bit later.

Thanks for viewing,

Cliff

#125

Cliff, can’t you set the TAM unit to have a higher “force” threshold?

I think it’s CV50

Greg

#126

Greg: YOU’RE RIGHT! Funny, the on-line instructions (vs. those that ship with the board) have an extra page 3, with the DCC info. Hadn’t seen that, so thanks, I’ll definitely give that a workout today! For a non-DCC app, the good news is that the auto-adjust can be manually “helped” within reason to stop at the proper point (see conversation w/ DM in my next post), and things should be fine.

[edit] From the TVD Singlet II DCC chart:

CV: 50

Function: AutoAlignCurrent

Default: 0

Range: 0-255

Notes: The auto adjust current. A zero means use the default of 24. Change to ~10 if alignment current is too high.

#127

I learned a few things last weekend and yesterday, starting with raw output force of different drives.
By “servo drive,” I mean the thing I’ve been messing with.

Piko (similar to LGB, but I don’t have an LGB): 0.22 lbs causes spring to flip; motor put out 0.40 lbs
Aristocraft: 0.20 lbs
Train-Li: 1.02
Servo drive (resistance when auto-endpoint-setting kicks in): .43 lbs (@ crank radius of .47")
Servo drive (stall force): 4.80 lbs (@ crank radius of .47")

This heavy-duty servo puts out a lot of force, more than the other components (e.g., the servo mount) would want to see.

The automatic calibration of the board, however, was causing me problems. I set up several wire & tube configurations last week, and all but the most simple had more mechanical resistance WHILE THROWING than the board’s resistance threshold. So, while going through the calibration routine, the controller would see this resistance, and assume an endpoint was reached.

This is exacerbated with increased complexity of the cable path. I found that additional length and curvature, especially reversing curves, really drove cable resistance up. At ten feet, with reversing 3-bend, the wire-in-tube simply locked up, and spray teflon lube didn’t help.

Additionally (and this is a little hard to describe), a small resistance at the throwbar (say, sticky points) caused a sort of traffic jam back along the cable path, as the cable went from one tube wall to the other.

Point being, there are a number of variables in the tube / cable, so simpler the better.

Having said all that, I received some good news this morning from Duncan McCree, designer of the controller board. Here’s the jist of the Q&A:

CJ: Can I manually set the ends of travel beyond what the automatic method thinks they should be? The unaided auto method feels the resistance in the tube, and stops too soon. But if I help the servo with my finger, the positions end up ok, and the servo still stops properly.

DM: Yes this is a perfectly good way to set the end points - wait until the servo arm gets where you want and then push back to get it to accept that position.

CJ: Even if the servo does stop where I want it, does the above sound damaging in some way?

DM: No it won’t cause any harm.

CJ: Is there a way to increase the auto resistance amount?

DM: Every time it runs it ups the resistance level a bit - so if it stops too soon try it a again. [edit: Greg, it’s funny that Duncan didn’t mention CV50, but maybe he was assuming I’d want to stay more operationally generic]

CJ: I’d like to try manual setting with the pushbuttons, but I couldn’t work out the instructions. Line 1 of your chart says: “Hold Button | LED1 flashing | Set the endpoint for Button 1.” Which button? Do I force the servo into position before this, or after, or ?

DM: You hold down one of the buttons until the LEDs flash. Then you use the 2 buttons to move the servo arm to the correct position. You don’t move the arm before you start.

There are a number of additional things to resolve in this project, like final mount design & final cable / wire selection & attachment method. Also, I’m coming back to the idea that track-side spring is needed.

Thanks for viewing!
Cliff

#128

Fun is HARD!

#129

It’s a cool little gadget, I am using a number of them in Z scale. I had some of the early ones, and the hardware and firmware has gotten refined over time, but they are still “tweaky”.

I have found that the auto calibration was very touchy in Z scale, with very small rails, it is easy to apply too much power.

I also found situations where the calibration ended up with the server “buzzing” in one location or the other.

Greg

#130

You bet, Andrew, haha!

Right Greg, I did the auto-adjust a number of times today, and sometimes the servo kept buzzing, as you say. Usually, it shut off after about 5 or 10 seconds. But when it didn’t, I’m assuming that the adjustment needs to be backed off. And per Duncan’s instructions, I was able to tweak that manually (via buttons).

Main thing, I was in fact able to up the auto-adjust current (40, vs. the default of 24, seemed to work well) via DCC, and the process works far better now. This was with a tightly convoluted 4’ cable.

Though I’m using DCC, I’m interested in maintaining flexibility for other folks. So, since “you can program all functions with the buttons,” I’ve asked DM how to do that for CV50 (adj current), CV35 (speed) and CV37 (direction).

BTW, one little thing that helped was putting tag lines on the jumpers. This is nylon button thread, melted at the ends and crazy-glued at the knot. Fluorescent orange or red would probably be better, if it were dropped while programming in situ. Anyway, it saves having to use needle-nose pliers.

Cliff

#131

If you get those programming instructions, please send a copy to me. One thing for sure, with the buttons, you will not be able to set an absolute value for things like position or current.

I’ve also not done a lot of experimentation on trying to read back CV values, which I think would really help here.

Greg

#132

Will do, Greg. I’m not really counting on the phrase “you can program all functions with the buttons” as referring to ALL functions, but I thought I’d poke Duncan anyway about it… (http://www.largescalecentral.com/externals/tinymce/plugins/emoticons/img/smiley-wink.gif)

FWIW, I used the “program on main” / loco #1 method, which allowed pretty quick settings, and reliably showed the existing CV settings.

Cliff

#133

Actually you cannot read back when programming on the main, it will show the last value entered for that CV, it does NOT read back the value. It is electrically impossible to read a CV back when programming on the main, only during service mode can values be read back.

Greg

#134

OK, thanks Greg.

#135

Greg Elmassian said:

Actually you cannot read back when programming on the main, it will show the last value entered for that CV, it does NOT read back the value. It is electrically impossible to read a CV back when programming on the main, only during service mode can values be read back.

Greg

Greg

You are correct when you use uni directional DCC. When using bi-directional DCC you can easily get the current CV value back from the device when using programming on the main.

This works both for stationary decoders as well as mobile ones.

Stan

#136

I know stan, so doesn’t the decoder and the system need to support bidirectional DCC?

So your comment is helpful how?

(this thread is not supposed to be advanced DCC topics on a capability that is not common, and surely not in his decoders nor his command system)

Greg

#137

I did some final testing this morning, and have come to my personal opinions on best push-pull cable products to use on this project, going forward. This is after working with 3 kinds of spring wire, 5 kinds of cable, and 6 kinds of nylon tube.

My criteria began with corrosion resistance, availability, the ability to easily cut it to length, cost, and a minimum length of around 4 feet. These final selections are based on smoothness of operation and flexibility. The bronze and ss cables worked equally well, except that one can solder the bronze (should that become important). The bronze cable is slightly more flexible.

Here’s the clickable links,
http://www.mcmaster.com/#8908t32/=wwm6pq
http://www.mcmaster.com/#3462t53/=wwm6gi
http://www.mcmaster.com/#9685t1/=wwm7ly

Thanks for viewing,
Cliff