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

[RJSV]

(Diagram shows) cleaned up layout a bit. That's maybe, 3 hours, min, right there.  I think I figured someone, (you...) would have paid out around $20,000 dollar, for this monstrousit..., Er, masterpiece.

   Up top, 'B' channel has that cute little 'TAP', which actually allows for lower friction (build-up).
Each input, and output really, have rt. angle interface, to the external shaft set, (4), that interconnects, box to box down the line.
Down below, in this 3-layer jobber, is the so-named 'MOTOR Layer', that is for positioning all three layers together. The output labeled 'CL' is 'channel C local', and that is your custom user feed.

[RJSV]

Wanted to show, picture has the side view of the 'robot' dog.  There are 3 movements, one is just tail wagging, that driven by an 'eccentric'; by that meaning a moving shaft/axel with wheel and a post, for eccentric action.
So that goes back and forth, regardless of electric motor direction and polarity.

   For walking motion, is one output gear, and for opposite polarity, the output drives the head, 'bobbing' as the little audio board makes audio 'barks'...

   When motor is driving walking, (pretty lame), the front dog leg moves slightly circular: also used is a post, for the leg piece to restrain motion.
Both eccentric drives have that feature, in front
 Actually the two sides are same shaft rotation, just in different phase, for shifty walking.
In back (legs), the leg is just pulled back and forth, skidds or scratches in semi-realistic walking 'shuffle'.
3V motor has the typical 10 uF cap across it.

[RJSV]

Here is view of whole package, wired remote.

[RJSV]

   Gear Ratios matter a lot, and there are a couple different situations going on here.
   For the 'transparent' aspect, you want as much lack of acceleration in the signal.  This is because of the friction that can increase, when rotation rates vary; up and down.  This is the case, even when overall gear ratio is at 1.0, and the reason being the non-linear friction, of various sources.
   In the photo/diagram, the 8 gears form various but average 1.0 ratio.  The fact that some place(s), in the gear chain from input to output, will speed up the signal, like 40 to 20 for a 2X, that speed-up causes disproportionate more friction, during that couple if gears transition (of the rotaeasery signal).
Then, point being, a following ratio, like 20 to 40 (2X gear-down), will not experience a '1/2' degree of friction (reduction) to balance out the increase.

   Current example the two diagrams are the gear ratio changes, experienced along the gear chain.  Probably need a LOG (logarythim) where, 1/2 is same distance up to '1.0' on the graph, and a 2X ratio is, also that distance, up from 1.0.

[RJSV]

   A transmission line question is coming up:  Suppose there are 8 boxes, each sometimes switching, but for now they all just 'chain', serially.  That's 8 gears per box, at 1 : 1 ratios, for a total of 64 gears, probably typical 32 tooth nylon.

   NOW... What happens, when your processor gives out a 250 mSec pulse, to the motor (that's 300 mA little pm motor)?
   I figure an impulse into typical toy gear chain, will take, maybe, 240 mSec. before the third station output starts rattling to life. (Another TECH term).
AND, figure not much flywheel effect; the gear train prob comes to slimy HALT soon, like 100 mSec after the 3 Volts gets stopped.
So I tried to graph that
Please see photo

[RJSV]

This more detailed (see photo), view shows the 8 gears in each station switch box.
Idea is propagation thru the chain is palatable, or at least audiable and a 240 mSec pulse, making way thru 24 gears, comes out to 10 mSec per gear, to sort-of 'wake up'.

[RJSV]

Hard to convey in the picture, here, but question is about the back end, of the curve. That is, while I can imagine a front 'wave-front' powering into each next box, it's hard to characterize the back or turn off where rotation has stopped, in the wake.
   Plus, it becomes important to distinguish: most power stop actions also inherently place a BLOCK on any rotation, down the line as well. So that is, maybe, equiv of sending an (electric) pulse, then shorting out the line!
It is possible, that removing the electric motor connect so the gears in the network line are free, that might have effect(s).
Certainly, in steady state, having a 'dead' motor connected is a huge 'buzz-kill' (uh,...stalls the whole drive chain...)

[RJSV]

   Picture showing:
   The graphic has main TIME axis, moving downward as time flows.   On the left, each series string (of gears), is for the rotary power flow as left to right; in other words, an electric controller / motor drives the series string Network.

   Looking at the 'red' gears; those represent 100 % of total speed has been reached. See first line; at 200 mSeconds from origination pulse (that's set at 250 mSec.)
   Next, going down, the (electric) power is gone, by time= 250 mSec.  So the gear train speed starts to decay. Scanning further downward, viewers can interpret this diagram to show the 'evolution' from active rotating, downward in speed, to 50%, then 20%, before full stop.
   The graph is a little harsh, in that the residual rotation(s) likely persist more like a couple hundred milli-seconds..., Not 100 mSec.

[RJSV]

   The reason the graph (previous) helps, is to get a glance at how 'FAR' down the line that momentum / inertia will take the gear start-ups, after formal STOP is made, and will those extra interactions cause problems.
   Well, according to graph, there shouldn't be much residual transmission, (down the line to other switch boxes), maybe one box would be accessed.

[RJSV]

...But what (that last) graph starts to show, how everything in that gear train line-up just, maybe, goes into a 'free fall', or at least 'free wind-down'. What I mean is, there is a sort-of ACOUSTIC like connection, from gear pressure applied, down the line.
Think of it this way: With gear tensions or 'run lash', all back lash gone, try poke or tap the gear.  Acoustically, that TAP will progress thru your train, enabling you to feel the taps, some number of gears forward.  That's what I'm calling 'acoustic' and that moves at speed of sound (in metal).  By taking some (even slight) pressure off, by cutting the motor for instance, you change the mode, of this gear train slow-down, to something like, every gear just slowing down by itself.
(Using speed of sound est 1 foot per millisecond, very roughly)

[RJSV]

The provably, better approach, changes the rotary switch configuration, of the chained or 'Local', to a situation where the 'chained' signal is constant; never switched.
Going box to box, there could be as little as 2 gears, interfacing the input to the output.  Right now the gear count is eight, and even worse, for avoiding friction, the 'speed-up' then 'slow-down' aspect quickly adds to the total drag.
  This way, each of the columns are not switched, there are three 'TAPs' switched in, which is locally similar to the (current) state of ...:
   Revision 2.0.11.49532, Feb. 4th
haha

[coppercone2]

what if the stop applied a break to all the gears? like a whole bunch of pins with rubber on the tip that press into everything to hault it.

[RJSV]

   Coppercone2:
   I don't think I wanted that individual stop: but it's just that I realized, each individual gear, maybe, gets into a state where there is no 'immediate' forces but rather just coasting down, along with everything else, along the chain.
   I heard about gears, many years back, dude was saying that ' gears aren't what's doing the work'...that it's really forces on (gear metal teeth) rather than some kind of 'spin' energy...
Something like, he saying ' Gears are just a convince, round in a clever way to bring those 'working surfaces' into contact...


   At any rate, stay tuned, please...more breaking news...
   I've figured a basic change that gets rid of massive gear count (and DRAG!).
  The new idea has the 'chain' from station box to station box, with, essentially NO gear, that being just one continuous shaft (although using couplings of course).
   See photo, there is ONE gear, incidental as the TAP switch being open will leave that gear unconnected.

[RJSV]

   The 'switched-in' signals don't make it, back onto the serial network.  For those components in 'switched-in' chains, a couple of gear-downs don't hurt, as this is one instance, not chained.
For example, a 1 to 2 ratio, followed by 1 to 3 ratio giving a total gear-down of X6, could then position the toggle 'motor', with geared-down ease.
As to the question: Why not switch off all of the box, at once?
   That's a (frustrating) circular argument involved.  It's like, channel 'A' cancels 'B', turns out the lights, and goes home..., but who is left, to turn off 'A' s lights?

   But now, using SPST switching, plus that hot-hot 'overflow' special function, it's more reliable, to implement self-disconnect, (hopefully without needing some ghastly flywheel contraption!)

[coppercone2]

abrupt interruption of rotation between gears is certainly interesting to think about. When you think about how the deformation from the impact or the 'lurch' from the break might look like something is 'spreading' from the edge of the gear which puts a real spot light on the 'surface energy' as you described.

When a rotating gear has a strong magnetic field applied to it as a break its certainly interesting.. uniform field around the gear when the different parts are at different KE's. Makes it seem like the field should NOT be uniform to give a 'softer break' to the gear.

[RJSV]

Wow, nice:
   Enter the ' PROXY GEAR '.

   I remember, recently saw that great, mechanical computer video, featuring a ' Range fire Computer '.
That, would track an early airplane, effectively 'doing' calculus, for the tracking, using RADAR input feed ...

   Well,... anyway, this ' Proxy' whatever, it's really a;
   1)  ' Proxy Motor ', as it runs, almost exactly, to the tooth, in step with the 'real' motor, the one in the next switch box (or station).
   2) well, uh, it's not.
      It's really a PROXY of a VIRTUAL MOTOR.   Er...
   3). How about: it's a virtual motor, a 'Proxy Gear', and it's a proxy for another 'virtual motor'.
(or gear-train).

   Plus, it's looking like it is possible to make a 'self-extinguishing' switch, using the features in a virtual motor right there, to basically indicate the 200 to 400 milli-Seconds it takes, normally, to fully switch, thus start to an 'overflow'.
   Yup, extremely valuable and simple, for timing, which is actually more literally 'sequencing', at whatever timing it took.  Just being carefully, to match gear tooth counts, for the 'Proxy' right there.

[RJSV]

   It sounds cryptic, but that's the input face, including the 'L in' or local motor input signal.  That is responsible for setting the 'B' switch (TAP).  (See photo).
   Let's start with that: The switch section #16 is sending, activating the next box, Station #17.
   Now that we can focus, on Box #17, we can send as needed, the custom stuff, (custom rotary output from Channel 'C').  Then, when it's time to move on, to Box #18, Channel 'B' can activate the Channel 'A' TAP Switch.
   Notice that any switching activity does not involve switching to an already active shaft, 'A', 'B', or 'C'.  That's called 'DRY SWITCHED'.

[RJSV]

Here is a related photo, showing the 'footprint', for the 4- Shaft mechanical BUS.

[RJSV]

   It was too tempting to use this (Photo), as that's a lot of work.  The main shape, essentially after a 'Y' shaped path switch, gives a 'form' of sorts: outlining where new components can best fit.  In this case, picture shows the 'B' column variations don't use any of the 'right side' gears, for connect by the moving tip gear. So that space becomes the (empty) place of interest.
With the 'A' switch to the right (next sheet), the intermediate shaft can fit, to the right of the 'B' partition, where it sits, in the (output) 'Footprint', as discussed.
   Up to, notice no 'motor' features, just 'reads' the turning there, of the 'B' Shaft into that commutation toggle.  The 'Overflow', by the way, is best if located around, (including the double riding gears, the TIP, and the PIVOT gears), plus the little driver wheel.
All very detailed and messy...AND boring I'd bet, not for me tho....

[RJSV]

   Folks; SORRY, I made major mistake, but easy to remedy...
...But first, a brief look at the layout / locating of the 'LOCAL' output shaft:
The diagram showing simple shaft connects, only local as one box, only to the next switch box.
For layout, that 'L' or Local shaft, in the 'footprint' of the input face, should simply be spaced a decent amount, from the 'A' shaft, and 'B' shaft.
So, half way between, and that same distance (down) makes for equilateral triangle.

[RJSV]

   My mistake, (Sorry !!), was that the 'B' channel gets CLEARED, not SET !
For that you need the right- hand shown 'Overflow' gear.  Plus that ends the use of that space, previous not occupied...

   Photo shows: The Local output from each switch box will connect, by shaft, and end up powering the Virtual Motor, the type that positions each toggle.
With ratios like; 1:3 with 1:2 and another 1:2, that ends up being a 1 to 12 gear-down, and heck; I don't have any clue what I need there.
   Those two intermediate gears are, mainly, there just to transport the rotation over there, sideways.
The ratio 1 to 12 sounds OK, as I figured some toggle gears only move, like 5 teeth during the mode of switch transport, (expecting 200 to 400 mSeconds normal).

[RJSV]

   Taking a minute's distraction, here is a mechanical
   ADJUSTABLE.  SIGNAL.  GENERATOR.

   Thinking about that 'Toggle' switch, you know: How about a 'feature' to stop in the middle, of travel of that TIP Gear; that's a no-load place, where the switch components are not pushing a gear load.

  Anyway, no time to think in-depth, but you could do some time divisions, and without (obvious) ratio gear-downs.   Some pretty nice Timing and (small) distance potential.
For example, your processor, smart AI, will trial out that a 240 milliSecond rotary impulse will cause a 300 mSec travel time (before the switch toggle lever drags to a stop).
   So your handy processor, ( like an Arduino) can keep a little table:  Maybe Station #7 has a glitch where it needs extra time.  Perhaps, then, Station #10 had been found to switch extra easy / fast...
   Temping diversion !

[RJSV]

   There are various ways to have oscillation, the switch, in SPDT arrangement (double-throw) merely has an inverter in one path, both paths then connect to the switch positioned 'motor'.
With 300 mSec to complete movement, plus about another 200, for loop; that's 500 mSec at very best.  But then, you need 'recovery', that's where the machine puts itself back to starting condition, by returning the SPDT switch back (to ready state).
The way to get faster, is by using an alternating
'either-or' where there are 2 pulse sourcing bits of (gear) logic. Each could do a 150 mSec pulse, every second, so with both, you could have a 2 HZ CLOCK.

   To make even shorter, but controlled, pulses, the switch is 'prewound', which simply means that, during recovery time, some portion of the characteristic 150 mSec can be input: say '50 mSeconds' worth.  Then, the pre-conditioned timing can be used, soon, completing the full, disconnect that a single, 150 mSec
pulse is going to do.
Of course, you also don't want the sort-of 'tooth-grind' that may happen, at very tentative gear separation.

[RJSV]

   Working on a pretty good wrap-up, at least this phase.  The layout has 3 segments, 'A', 'B', and 'C', and so is (ridiculous) very wide. It is a 2 layer deal; the lower layer of a switch is for positioning (i.e. 'Motor').
The upper layer, that's called a 'Reader' or 'Follower', as it can read out a toggle position, out of 0 or 60 degrees rotation from within stops.

   For fabrication plans, comparison shows one of those toys, has about 7 gears, roughly, in that little gear arrangement.  This Mechanical Logic Switch design will use approx. 14 gears, in the upper, reader stage.

   Plus, this Network Gear-box has, about 11 gears, in the lower, toggle positioner spaces.
So:...  That's approx.25 gears, compare with 2 toys, that's at approx. 14 regular type gears.  Point being, you are not in the 'quicksand', not yet anyway.
Also, notice I left out the 'C' channel, another 9 gears.

   That's 34 against what 2 of those robot toys have;  14
Not that bad, those gear boxes are tiny.

[RJSV]

Picture showing:
   That's the 2 TOY gear layouts, with the path switching to demonstrate a very accurate look / size occupied, etc.  The 'Y' shaped layout, of both switch assemblies, is apparent.
Plus, there is a watch band set there as a pointer in photo: that gear, the band points to, on left, is a second gear, just placed there, to simulate roughly what the other layer would take up, in thickness overall.