Sharing some project planning phase: A (digital) ELECTRO-MECHANICAL Network

[RJSV]

Here is picture, you can see,  where two metal screws are:. It's possible to use nylon fasteners, for in-service use (prototype has some shortcuts).

[RJSV]

   Getting pretty exciting, watching my prototype switch box get assembled, for first time!
   This 3PDT Mechanical Switch (rotary activated transmission really) is approx 2 X sized, and main 'test' involves that moving piece: the toggle lever.
   See in photo, the little screwdriver on top the box indicates the 'toggle lever'.  These cut-out ports you see are for access in assembly and visual confirmation.  That toggle lever arm might be required to be a different length, or need a wider swing.

   But the box construction itself ends up fairly aligned, as those various shafts, (3 parts, for channels 'A', 'B', and 'C') look aligned, decent.  We are talking very low accuracy, perhaps +/- 1 mm in critical parts...

   The upper shafts, those don't actually exit the box, as those are stationary axels for each of the 3 SPDT arrangements. (You only see 2,  because the 'C' channel switch subsection is behind the other two,  and so it shares the upper stationary shaft, (with 'A' portion).
   Mainly, I'm interested in debug the toggle lever sizing
(in photo, that toggle lever moving part is just a scrap piece, for confirming things fit...so far.
   I've now gotten rubbed oil finish, outside, some extra glue, in loose spots, and some nice, yellow highlight paint, on the sides.  Generally, the 'face' in the photo, is considered the output plate:
   3 outputs are the 'BUS CHAIN'
   1 output is 'local BUS control...i.e SELECT to next box
   and 1 output will be the actual, locally available
    ' CLIENT'  signal, for doing custom things, such as
    winding or un-winding a bobbin, for simulating a
    node, in a serial string of (these) boxes.
 

[RJSV]

Here is another view, of the 3PDT switch.

[RJSV]

...maybe hard to see, in there, without contrasting color, but the mock-up 'toggle lever' is mounted on the right side stationary 'axel' wooden dowel rod.  The toggle shown with downward tip 'selecting' the left side, although incomplete as it's going to actually be (rubber) wheel to wheel contact, selecting either the left or the right side (for output from the switch box).
   Generally, in the serial string network, four output shafts connect, on down the line, to each next 3PDT switch box.  That's considered 'transparent' state, where the local output is not controlled.

[coppercone2]

The friction scaling problem seems frustrating, to get the best alignment you would need to get a fly cut or better plate, drill holes with a jig borer, temper and ground,  then size the holes properly with a jig grinder and use machine pins for the elements, using thermal fitting to seat components.

I am just posting because I am kind of wondering what it would cost to make this design in a high precision variant. It also seems that its something you can get partially manufactured in a gear-box shop, or a spline shop, there are still manufacturers of precision scientific grade gear-boxes BTW. It would be a non standard custom job.. but I was kind of wondering what the cost would be. I know thats not your objective but I find it interesting to think about how it would perform if you had this going with moly-d on precision ground gears. I think this construction would ultimately give the lowest friction that can think of with my limited knowledge of gears.

Here is a 3d-printed 11 million to 1 gearbox someone made (crazy amount of slop, but it does function!)



Those videos might help you size up your friction problem more accurately, if it does exist, since those are 'cheap' designs.

[RJSV]

   Thanks for thoughts, contributing to the viability of some components, in even harsher conditions, as MOLY STEEL has potential. Hypothetical there, has sometimes been the 'SOLAR ATMOSPHERE PROBE", (sounds comical, but that sort of science involves 'skimming' close to SUN, briefly.)

   I'll have to think about your post, Coppercone2.  I guess a 'jig' has a pattern and a 'follower' pin, for guiding a nearby attached cutter. I want a Dremel tool with that...
  As to developers / fabricators cost:
   Latest progress on proto box involved 3 days, total about 16 hours, and so at nominal $20 per hour that's a labor cost (equivalent) maybe of $ 320. However, most formal employment, you maybe need to double that, covering (health insurance) the myriad of formal business expense.
  That's more closer to $ 600, and only a moderate gain in the prototype development.  (Any reader wants to send me that... I can spend even more than $ 600... maybe just for decent jig...)
   I estimated, 12 of those hours spent on serious machine shop tasks:  the rest of that time, 4 hours on just 'dinking' with things: That means staring and scribbling notes, and making tiny changes in the fit and friction, on that prototype model components.
(Also involves minor nasty 'ice pick' stab wound, to finger tip...)

   Now, (thanks again coppertone2), a MOLY-D version has to be adding substantial materials cost.

   Two mechanical issues, yet to be tested / addressed involve:
   1.).  The toggle lever movement needs to have a slip     
            clutch type effect, at the end of travel just to
            guarantee full travel, (but maybe not stall at t
             that completion).

   2.).   See photo diagram: pulleys need to be operated
             by either of two torques, with the other pulley   
             not adversely affected.  In simple view, turning
             one pulley will cause the other to turn, and this
             will also put torque 'backards' into the other
             driver pulley (3/4 inch dia. here).
             I need to observe, before tackling that, as
             my mechanical expertise is not 'expert'.

            By the way, eevblog hosts a great video,
     Dan Gelbert discusses prototypes, in machine shop.

[coppercone2]

yeah your design has alot of unique challenges over a gear box, that was just my first idea of a starting point. I never even made a gear before, the best I  could do is resin print one. Interesting to read that though. Maybe I can make a brass one on my proxxon mill (I do have a kind of indexer), but it would require alot of learning. I know gears are very difficult to master.

And of course doing wax prints and casting gears is an option to make them metal cheap. I would think zinc cast would be easy, brass would be best. Zinc is stronger then aluminum and I think melts at a lower temp, if you get the right formulation, and it does not need you to engage in anything especially hazardous like molten brass (that is a big step for me). I think it might be possible to pour a very good zinc casting.. but its only conjecture

[RJSV]

   Here is, a nice picture, (from inventor perspective),  showing the so-called 'BRIDGE' drive arrangement:

   The toggle lever is shown, left position in this view, where that is the 'selected' state.  The foreground input shaft has that little 3/4 inch mock-up pulley driver wheel, for that blue rubber band, the band going up, to the pivot axis. The bridge is intended for bringing the input rotation, ultimately, up, over, and back down to the 'tip wheel', that is the moving part for output.

[RJSV]

...Now, this second picture shows, look for continuation of the input shaft, (thru the half-wall, look towards up, and left, for the OUTPUT shaft, continues that shaft direction).
   That's the output shaft, and it's almost as if the input could just be going straight through the switch box...
BUT what happens is the rotation is transferred up, over and back down, and for the non-selected case, will exit out the OUTPUT FACE as if continuous.
That's the 'data transparent' mode.
   Of course, my modeling abilities not so stellar; in this project I tried using heavy heavy paper (tablet backing) but it is so difficult to cut using Xacto knife, I'm going back to just saw cutting my boxwood, for 3D mock-up.
  As to why fuss over aligning the shafts, to some yet-nonexisting NETWORK rules, where input and output align, I figure why leave that major design feature out...heck, that's a ton of work to have to modify later.

[RJSV]

...Those last couple posts help view one section, of the three section 3PDT switch, generally using 3/4 inch pulleys for driving the band. The arrangement first acts to bring the input rotation into alignment with the toggle arm PIVOT axis: that way the toggle arm can swing the 20 degrees or so, between stops, without causing tension changes, in that drive band.
   In the photos, the left is 'SELECT' and swung to the right is the un-selected or transparent switch position.

   That input shaft also must be located going 'through' the other main section, keeping clearance from any interference. That issue is why this 'working mock-up' is valuable, as a sort-of 3D design aid.
   Anyone have suggestions, PC software for modeling to find such interfere spots ?

[RJSV]

   NOW, taking a self-critical viewpoint:
   Getting this '4 signal BUS' to work is nice project milestone, but also establishes challenge:  To make that series network to function with less.

   To that end, consider a 3 shaft SERIES BUS:
   You would 'encode' one of 4 (outputs), using two shafts, plus a needed clock (mechanical shaft).  The two 'addressing bits', using CW for '1', or CCW  (counter clockwise rotation, to a stop), for positioning a multiple selection switch, then pulse the mechanical 'clock' line, producing (rotary) output, on the one of four.
  This won't work, in extension, as each new access, further down the series string of switch box stations will require yet more and more 'bulk' of switching activity. Or, say, it can work but ridiculously burdened.
Each access, let's say you are at station 7; now to control and access the next, station 8, you need station 7 multiple times, for twirling those 3 inputs to station 8.  But for that, you have to revisit station 6, multiple times, and for that you need station 5,
...and for that...station 4, etc, etc etc.
   Gets to be a factorial number, of sorts, and the result is a (very) non-linear pile-up of required access/outputs, in the rotary encoded digital switching scheme!

   In current design ( 4 shafts total travel down the network line), a station count, like 30, is going to mean TWICE as many pulses issued by base control, VS having 15 stations, for example.

   I also, here POSTING A PICTURE, OF A BUG, as I have no picture of CAT.

[RJSV]

   Maybe providing 'too much detail' but trying to write towards beginners, and plus that's actual kind of my own level of expertise; maybe similar to a 'diary'.
   I did work in a physics / engineering R & D situation, doing electronics development and testing, in development of high vacume, high voltage electron beam 'accelerator'. For a medical related application.
They insisted on every engineer keeping entries in a 'department LOG BOOK'; keeping track of project details. Plus, that was a 'shoestring' budget.  I got to view, first-hand, the dynamic of a little tiny 'shop', about 7 folks.
   We had the 'old guys': a couple  doctors investors in the venture. (They wanted to come 'tinker' with their Shakey hands.). We had, luckily, 'Hans', a mechanical genius with German accent so thick... For whatever reason, seems like the Swiss / German region has been a real center, for mechanical contraptions.
Now, perhaps India producing innovation...

   Picture showing my pivoting 'switch lever', right now I am 'adjusting' that 'mock-up' component, for best interface, with the two receiver wheels that are the switch 'poles'.  This pass, adjusting the cardboard piece for shortening (radius) of travel, from 2 1/4 down to 2.0 inches, pivot to 'tip wheel's.

   Mechanical and assembly problems include:
   1).  Perhaps use of 'oil finish' causes problems with
          glue strength:. I try not to put oil finish near glue
          areas (inside switch box).

   2).  Using wood dowels (1/8 inch),  as 'axels' is kinda
          lame, but works without massive friction.
   3.). Assembly issues must be included, in the
          packaging:. a consideration way way way outside
          my background, if you consider solely, an EE
          degree, with mainly software experience (on
          paper).

    But, innovation is a basic way of life, living in SF bay
    area,  and (these sorts of technology centers)
    attract all kinds of new developments.
       One university fellow noted:  Something like one
        out of every nine start-ups 'might' suceed.
        That's 11 percent.

   Other mechanical related issues involve needing simple one-way drive wheels, that don't turn when pushed 'backwards', with actual implementation using simple features, in plastic pulley wheel(s).

   PLUS, completion (of useable 3PDT switch), creates easy path, for other, ideas to follow (this).

[RJSV]

Here is some more progress: In the photo the mock-up components show the main 'cross', or 'X', of signals. The 'A' output acts to reset the 'B' segment. The 'B' (chain) output acts to reset 'A' segment.
The diagram / photo helps for visualizing and avoiding interfering parts.

[RJSV]

This next photo also shows 'X' shaped control paths, within the 3PDT switch box.
Many details of (this) partial paper test build were obsolete before construction finished...It's tons of work, thankless, but...can't see it THAT way.

   The basic idea, is a driver pulley at 3/4 inch diameter, going via band, to drive a 1 1/2 inch pulley, attached to the pivoting lever arm. The lever arm has the 'tip wheel' for driving one or the other, of output wheel pair.

[RJSV]

   You can see the (picture) organization, shown back to front, is 'C' (green toggle bracket in back quadrant).
In front, is 'A', the red components, and 'B', the yellow components (shown on left-hand side of photo).
   Each 'switch' section takes up a quadrant of the whole box.
Each individual of the 4 sections has a compartment, so the full box has, from back to front, 'C' inputs, 'C' outputs, 'A' inputs, 'A' outputs, each taking roughly equal depth (1 inch each, out of 4 total).

[RJSV]

Viewing photo, straight down, is seen the 'half-wall' that divides and supports the (front) 'A' compartment. (Other 2 partial walls omitted in this photo ).
   The red toggle bracket is seen (back rt corner) must straddle the half-wall as it acts as a bridge: rotary signal goes up, over, and back down, to the movable TIP Wheel.

[coppercone2]

the rubber bands are a nice way to transfer energy in a prototype because its torque limiting, so you cannot hurt yourself too bad. A prototype is usually a good place to hurt yourself because you are working with measuring exposed mechanical systems most likely.. so its a good way to go about things. It would suck to get a pinky crushed in some gear chain. Just watch out for them snapping and nailing you in the eye.. you can glue them together with super glue to make sure they are never super taught (custom size), so even if they are less reliable because of the glue joint, they should not have alot of force when they break. *

Are those broccoli bands btw?

*I am just saying this because it looks like something you might show to common people because its interesting and mechanically impressive. If it works then it would be a success at a science fair etc.

I kind of feel like your mechanical intuition is like.. much greater.. then most people on the forum  . I never really built mechanical objects that are this complex/dynamic (moving parts are hard). I tried to build like a micro chop saw that I can put a dremel flex shaft on and lets just say that 2 days later I decided to buy a proxxon lol. Nothing came together quite right and it turned out to be a enormous amount of work designing a seemingly simple pivot. I could probobly do it now after a few years of watching people work on youtube. Also the hole alignment and layout is tricky..

but my mistake was also not making a cardboard prototype.. i went strait for the aluminum pieces

[RJSV]

   The picture, reminds me of a 'square Porcupine' but those fragile wood dowels are a pain to replace, and so not too good idea, to locate strewn across. .
Boss would say " if gonna have vulnerable, at least minimize that."
So, a better footprint, is three pairs, each of A, B,C channels. Channel 'A' local is only used inside, so that is the short stub output.
   Not so bad, a chain of 2 BUS signals, plus a local control,  gets you access, albeit minimal.
And so, with another BUS signal, (i.e. channel 'C'), basically along for the ride; that rotary channel can be used for any misc general output.

[RJSV]

   So it's almost like a 'cartoon caracature' of a machine design...dictated by a Software expert, fabricated by an artist - satirist...(sigh).
Looking HOKEY...very HOKEY!
   But...next version gets rid of some complaints (the footprint), gets even larger, more small parts to fab (3/4 inch pulleys, need, like, 50 of them.

   And YES, coppercone2, THOSE are 'brocoli' bands; I left my professional rubber bands (in my other car).
The next version puts the 3 switch segments, 'A', 'B', and 'C', in a 'stacked' arrangement, all sizes being equal and uniform, across the three.

[coppercone2]

Well the cartoon prevents you from making a large heavy expensive time consuming mechanical piece only to be foiled by some kind of clearance issue...

Do you have a 3d printer or lathe to make pulleys with? if not..


 You can cut wood disks out with a hole saw quickly, I would try to wedge them on a shaft and use a upright drill press as a wood lathe with some round files/rasps after some sanding... but you need to be good with your eyes to make 50 identical ones, there are no stops or settings doing it this way... but it should be fastish.. so long you choose the right base wood (easy cutting). I only ever did this for chamfering wood disks, I never tried to cut a groove in it.

Another type of pulley suitable for a band could be made like this:
1) get big piece of plexiglass or polycarbonate
2) cut a bunch of disks out in two sizes, one for the area where the band runs, and one for the 'bobbin'
3) center drill and stack disks on a shaft to make the right size pulley and glue together with acrylic or polycarbonate cement, so it wicks in from the side
4) put it in a drill press and polish the outside of the ring using a dremel with a series of scotch brite buffs on it and then solvent-polish it with the appropriate solvent to make it glass smooth (well, at least smooth enough, really polishing further it would be ALOT of work).. if you don't want it clear on the top and bottom, just scratch it up with a medium grade scotch brite buff.. perhaps texture would even help with grip (not sure how band interface components work).

But its a very.. geometric.. pulley, I am not sure if it would work well.

I guess you can combine the two techniques and try to premake the geometry with flat disks then use files on it to round it out.. maybe you can even use wooden disks but the problem is, I never saw thin wood sheets that are too flat.. I expect it to bow. Don't know what the hell would happen if you try to file plexiglass on a drill press either! probobly a bad idea, but it might be doable with a dremel with a sanding drum on it if you are gentle. It will overheat and crack the plexiglass most likely, don't sand it aggressively

I don't misuse drill press much BTW, if you do this alot you might find that the chuck falls out or something.. need to be real gentle, i recommend gently tapping on the open chuck after you are done doing this with a rubber mallet, to make sure the jacob's taper(re?)seated into the drill press, just make sure the chuck jaws are hidden when you do this.

[RJSV]

Here is photo of next phase, box number three, actually...
   Previous box gives model, for exact panel sizing, that's always a pain.  Things like which panel overlaps the other two, etc.
Each segment has 1/3 the whole, as a book shaped compartment. That's 2 inches depth, giving roughly 1 inch for inputs and 1 inch (depth) for outputs.
In photo, see the 'stairstep' arrangement, 'A' outputs at the highest, of the 3.

[RJSV]

The advantage, this box layout has simple, one behind the other. By offset downward (1 1/8 inches) there can be 3 separate segments or sections, 'A', 'B', 'C, all fit within 7 inch height.

[RJSV]

   In photo, I wanted to point out, the 3 'music boxes' that are shown side-view.  The stair-like arrangement puts the 'A' segment up higher, and in front.
The 'C' compartment is on the (photo) left. The two horizontal panels are the 'floor' for each separate compartment: floor is for each so-called 'half-wall'.

[RJSV]

In this photo view, is another version, of the Rotary Switch: this one, perhaps, more suitable for miniaturization. That is called a 'SLEEVE SWITCH' and involves coordinated movement, with sleeve sliding on-off, and with a flexing or shifting of the drive (cable).

[RJSV]

Here is a nice, 'Christmas-y' color photo: the green compartment is like 'START' or SET on a flip/flop.
The red is for clearing or 'RESET' as a process
The blue, that's for what ever the application wants, using the one dedicated output shaft ('C').
That's also called the local output shaft, or 'Virtual Motor', that being addressable.