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

[RJSV]

Filling out the details, here's how a remote station Switch Encoder works:
   (See Diagram), the two variations of (resolved) SWITCH positions are easily encoded; simply 'invert' by using odd number of gear interconnects, and mix the two possible (signals).
Using the rt. side layout scheme, you can see, one path gets through unchanged, while other gets rotary direction reversed.

  Now, also here, please note that the method, for reading and sending back position, to the BASE Station, needs some additional logic, for (time) sharing the 'A' shaft.  Otherwise the box, itself, will try to 'actuate' based on shaft 'A' sourced 'commands'.

[RJSV]

For 'crude' analog, remote, a conversion to Pulse Duration allows for sending a proportional signal (back to the computerized BASE station).
First, the 'C' shaft rotations get looped back down, towards BASE Station, signifying pulse start.

[RJSV]

Now, second photo shows: The timer runs approx from 150 mSec to 250 mSec.  You want to skip the lower range, as that gear start-up is very non-linear.
   After that Timer-Switch has rotated, to contact the Overflow Output wheel, the process is started, to pull the first switch OPEN, that way the rotation drive, 'C' to 'A', is interrupted, signifying the stop of time period, and also has to account for the 'Disconnect' pull, adding another 90 mSec or so.
   Base Station gets pulse (by reading shaft movement) and will time the 100 or so points, between 150 and 250 mSec.  Of course, a calibration helps, and temperature, etc can cause inaccuracy.

[RJSV]

Switching hats, for a sec, this photo covers some current thoughts, on how to plan for Oscilloscope (Tektronix '465, YAY), the response involving getting some contact (or optical), and measuring timing, starts with electric impulse start, and interval would end on that diagnostic temporary electric contact.
  High 'E' GUITAR STRINGs are very delicate and springy, for making temporary switch, (along with 3 V battery holder, and optional pulse indicating LED.

   Another 'street' tech method, has that 'tickle wire' on a regular gear, obtaining ground, or whatever, by way of the metal shaft. That way the little springy wire goes round just sitting in top of moving gear.
Of course, other contacting switches (Microswitch Co.) do that, but I wanted high resilience for fast rotation movements.

[Capernicus]

Are u making a mechanical analogue computer with cogs?
The illustrations are beautiful, nicely colour coded for clarity,   How much paper have to got lieing around now about it?   Your going to need even more,  maybe 15 years worth and then u can probably start a toilet paper factory with it.

CCW subtract,  CW add.  I'm the same.  and that gives u scalar values, (not just logic gates, which is digital.)   you have a range from 0 - 360*  (or 0-70* and dont use the full circle ever) for your minimum and maximum and u allow for half turns inbetween to get the scalars I think that's a good way to do it.

You get multiplies from the radius of the cogs.  and I think doing that is what gets you amplification possible.  U sorta power into the transmission gearing and it gives u amping.

You've got even more in front of you if u want to finish your computer!!! =)  keep going! you'll make it and if u finish it you could be quite successful.

I wish I could help more,  but I'm not using cogs, I'm using cranks. It's similar how its based apon rotation the same as cogs, but its a little different too, because I end up with a different set of problems,  but I bet one thing mechanical does in general is to tend all LOCK up together, if one part of a mechanical system locks, the whole freezes all at once.

A tricky thing (And I think you also stumbled apon it.) Is that if you attempt a gate with two operands,   if you affect it with operand a, it actually outputs and goes down operand b at the same time,  (so it spits back into the input.)    U need some way of separating that,  one thing u can do I think is u can do operand a and operand b sequentially instead of at the same time.

If you cant get arithmetic to work, you can try and get logic gates and do it binary/digitally, may be easier, up to you.

I think cogs get solved slightly differently than what I'm doing,  so its up to you keep at it!

[RJSV]

Capernicus, thanks for review !
   (feels now, like you the Senior Designer, here )
For your points, some thoughts:
   Not directly thinking about ANALOGUE, but just a couple posts back (approx 2/28), I'm discussing how an 'analogue positioned' moveable 'STOP' will adjust switching speed...That is, the distance that TIP GEAR has to swing, for contact with output gear.
So you can get manual input knobs etc. on a range of 0 to 99 for a somewhat nonlinear behavior.
(That's converted at BASE Station, being a Pulse Duration, where, for one example build, it's a max pulse of 250 mSec, where usable range is kept at 150 to 250, this you have a 100 point range, to some accuracy.

Thanks for your thoughts, on what I've been calling 'spurious back-feed', (or 'parasitic back-feed).  Nice to have somebody watchin mybaac, even if it's finding faults.  Numerous detail errors in there; not real bad, but I DID the MOTOR ELECTRIC POWER calc wrong:
   It's up to 6 Watts, transient motor pulse, using 9 V pulses, approx 600 mA.

Ever listen to Dave Blank?

[Capernicus]

Are u talking about moving the cog itself?  That could be useful,  maybe adding cranks (Like on the wheels of a train.)  to your design might help.
Cranks will add nonlinear behavior, tho.

I'm Just watching a Dave Blank thing now, on nano-technology.

Resin 3DPrinters could print some really small cogs, but it would be a bitch to have to assemble it all together with tweazers.

I'll be back, I'm a bit busy atm.

[RJSV]

   Capernicus, coppercone2; MK14 : Here's more details:
   The 'COG' moving ?  Not sure what that question is?
I'd say, the 'GEAR's move, sometimes only, like 1/ 4 turn.  Probably OK in (any) machine under light stress conditions, and modest working lifetime.

   The gears do not, however, do not DIRECTLY have analogue encoding; most logic in each switch box / Station are for FULL actuation-switching, thus considered 'DIGITAL', in academic sense.
I've left open though, possibilities, especially with analogue easy conversions to pulse width (or duration if it's just a single impulse).

   This morning, posting this display of one suggestible method, for generating electrical contact action, for SCOPE analysis.
  One thought was to simply, symmetrically, connect another gear train and switch. It will be 'backwards' as you are sending into the second gear train's output.
Goal is just to 'wiggle' the motor, obtaining approx 1 V pulse, for timing studies.
Need caution as the 'motion reacting' gears probably take a lot of time, maybe too much, for valid timing measurements of the first gear train (that I'm interested in characterize).

   In photo, the left side has electric motor to stimulate with a plus or minus direction impulse, say of 250 mSec (as the device likely switches in that 250 mSec time).

[RJSV]

Here, in photo, shows a supplemental support, for 3-point support, of the little switch module, on that plastic base.
   Picture is upside down, you can also see where the 'left side' output gear goes. That gear location is on rt. side, in this photo.
   Goal here is to have a stable test stand, and provide (various) ways to get a wire-to-metal contact result, for measuring timing.  SEE: in photo that eccentric mounted 'link piece' is underneath and pokes out a bit
for the temporary trials, to obtain timing and transit details, in the ARC swinging toggle with TIP gear for commutations.
The little cludged support was cut from a typical plastic parts slide drawer.
That 'push button' panel you see is from the Robot Toy;
With one push, you get + 3 V for CW rotation, or, on alternate button you get the minus (- 3 V) for CCW output, from motor, into the Direction Sensitive path switch.

[RJSV]

   A little self-awareness can often help.  Looking back, (please see Nov 24, 2021 post, reply #18), and thinking,with the (mechanical logic switch tools, presently could attempt a binary-coded addressing scheme.
   It might be lot like: Imagine telling the boss, as investor rep listens; "Fred, ...We gotta dump this whole line, ...our design has been 'OUT-INNOVATED'!
"Well, ...that's bad (sez boss)."
"Yeah, and it was me, disrupting my own technology..."

   Oh, well: Just trying, to get the thing out the door, before another round of 'improvements' disrupts it.
It reminds me of similar post, there was a culture of spinning the pc board fab, too early for proper testing and bug fixing.

At any rate, the '''''''new''''''' scheme would involve, say, 4 shafts, binary coded:
   First 3 shafts are address, in a traditional 3-bit code, while the 4th signal shaft is the 'STROBE' input, off the BUS. Typical 1 of 8 decoder (technically 'demux').
This is leaning towards addressability, of elements internal to any particular switch.
  Those local available signals could do the logic functions needed, those being explained here, as the main 'A', 'B', and 'C' shaft signals.
Notice that I'm still not suggesting binary addressed (switch) boxes, or 'Stations',  which actually might be something like 5 bits, for up to 32 Stations.
No, it's the control actions, locally...
   With some of the aquired skill set, it might be simpler and, indeed faster,  to use the conventional random access type (selection action).
But that way won't do 'unlimited' numbers of serial connected Switch Box Stations...
Hmmm gotta think on that overhaul....

   Can anybody, see, the potential, and engagement such a project, purchased in kit form, brings to the classroom...

[RJSV]

Ok, now, (I'm afraid this latest convolution, is going to alienate / lose some READERs...).
Don't worry, pipelined stuff frequently gets VERY difficult, comprehension-wise.
   It's looking like the (BC) grouping, in the scheme where the 'A' on the left side (of schematic), and 'B' shaft and 'C' shaft are grouped as 'right side' type, that (BC) group can be replaced, calling the new section the BINARY section, of the (serial) BUS.
   What that means, (I'm thinking), is that the 'Binary' section doesn't REPLACE the older way, but rather it SUPPLEMENTS with binary encoding, for expansion to have more than the ONE custom 'C' output.

   (Also see post #134, as it relates to this whole SELECTION, and partial box split-up, into logical left and right 'entities'.)
   So, it's very similar; The former (BC) subsection is replaced by a set of binary-coded functions (local).
   Here's the run-down:
   With 1 shaft, you get 1 function,
   With 2 shafts, you could try encoding, to Clk, A0, but those 2 shafts simply decode, to 2 outputs, (so no gain).
   
   With, 3 shafts, typical binary, you've got CLK, plus you've got a 1 of 4 decode, so 3 shafts giving 4 output.
   And; With 4 shafts, you get CLK, plus you get 1 of 8 decoded outputs.
Etc. etc. for a doubling of local available OUTPUTs for each shaft signal included in the BOX signal 'footprint'.

   Soooo...   ... It's looking like, BINARY decode is a nice expansion, of capability, but not until you go to a '5 Shaft' system, of BOX I/O shafts.

...(some) Potential for error, here...
...VERY HIGH PROBABILITY, of READER 'confusion', here!             Sorry, comes with the territory...

[RJSV]

   I will be supplying the layout diagrams for doing Binary Decoded function expansion (5 BUS signal shafts, now).
   Looking back at the posts involving more than 1 local output, internal to a Switch Box, now: with the binary decided scheme, can put more functionality, in one BOX, relative to the NETWORK of serial-connected individuals.  That means, for example, a  2-bit memory cell, can be written to, without going to an additional Switch Box, as the 'decoded' 1 of 4 will give an additional 2 functions.
   So, a  NETWORK Switch Box Assembly, 5-Shaft type, will then have an additional 2 un-committed shaft outputs, for doing actuations like, for example, 'RING a bell', or, an example set the LOCK to a safety hatch door.  Previous example just used 2 combined box's logic, into 1 box, for getting 2 custom actuators.

OK, Readers, time to start 'bleeding from the eyes'...
(Shucks, I'm just a Fatal Engineer...)

It's gonna take a kid, ages 3+, to really accept all this...
...'DATA BUS'    
...'Mechanical logic Sequencer...''
   ...'Data / Address Decoder'  ?

...'Dude: Where's the chips, there '

[RJSV]

   With numerous changes, here is some help, for understanding structure of the NETWORK:

   Folks could ask; Why not just apply the standard 'decoder chip' I/O lines, and just do the Transparency thing some other way?
Good question. By keeping the 'A' side switch, OUT of that process, of enabling the 'NEXT' network station, you avoid the act of switching a 'LIVE' signal.  That's because when any signal is issued, from electric motor(s), that signal is present all the way down the line, the whole series.
So, any Switch On Select, would otherwise cause a spurious 'surge' as that moving switch contacts the moving 'A' shaft, at completion of switch travel.
So to avoid that, you only want to switch on, the so-named 'right side' components, first, (from the prev box).
That's the down the middle 'SPLIT', in the architecture here.  Next, the 'B' channel (new box now) the B channel performs the 'trivial looking' act, of Selecting the 'A' left side switch.  Sounds too simple, but result helps avoid that 'switching HOT' type transient.
   This time, for that switch box 'clean up' or 'un-Select' the 'A' side, like in the (BC) case, on the right side half, NOW it is the 'BINARY DECODER' substituted, simply for more capacity and features, by decoding 1 of 4 outputs, to actuate 'Locally' in this sw Box, or nearby external (rotary) JOB to perform. Example might be to put a couple of 'impulses' into some string winder BOBBIN.
   Wading through this design process involves some 'sorting out', trial and error...and COFFEE...lots of coffee.
Paying attention to standardizing a 'footprint' helps impose some structure.  Some old assumptions persisted, despite fundamental design shifts...
(We've all seen THAT persistent dysfunction, in work and personal).
   I got lucky, and found some 'architectural' direction, out of that switch connection TOPOLOGY...

[RJSV]

(part 2 of review)
   The sequence, then, is to enable the next box, partially, then turn over the action, to that next box.
The 'new' box then enables it's own 'left side', channel 'A', and starts to go to work, using the binary decoded output to manipulate things.
That would be, first that 'A' switch enable, and then, second, the 'BASE' Control Station will use the previous box's 'A' channel switch. Remember, that is still in active mode...You have a kind of 'Smeared' locality, of what's 'on-line', and what's transparent.
Now, using the pivoting switch 'OVERFLOW' status signal, (described previous), that lagging Left side portion can be made to transparent mode, so whole box is transparent, to the binary decoder signals, while the next box is FULLY selected.
   Each excursion forward, down the network line, works that way, a complicated 'pipelined' situation, very hard to digest, newly...
That last, self-extinguishing action, is 'A' clearing the right side section (the decoder and user custom outputs), with eventual overflow, that loops back for 'Self-Cancelation'.
    I gotta come up with a more optimistic term...
Anyway, that issue, of having no ADULTS in the room, at the end, leaves a BOX in a sort of 'Limbo', but I'm confident a definitive turn-off, to transparent mode will not be impossible ..

[coppercone2]

I like the shape of the latest drawing, it reminds me of old banana plug crystals, and I have a matching transformer that kind of looks like that. Feels like it belongs in space.

I recently diagnosed my old lambda supply with a bad capacitor, they sure like axial parts. That part would look great inside of that chassis as a electrical component.

Its amusing to see the concept art.

[RJSV]

Haha, coppercone2, it's very NICE to have a 'fan',
...besides the stray cats I tend to, around here (lol).
   That rounded case, well that's copied from a Nesquik Chocolate container, yellow plastic, very rounded.
Space-Ready ?  Maybe me too.

   You (must) become a 'Packaging' guy, if you want to survive, in the 'Eddukasion field'.  ...uh, thats 'Edukation'...uh...
Anyway, I tried pretending to be an ARTIST, that didn't work.  Its a struggle, against basic Uglyness, and aging...(But I usually don't have to look at myself, lol).
   I use the plastic container, out back, picking up my POT smoking remnants, so KIDs can't get to it.

[RJSV]

Finally, back to work, it's maybe mundane, but necessary...necessary for the fabrication phase to be STABLE, if not accurate. (My machine shop skills are pretty sloppy: + or - 1 mm....terrible).

   But at least stable, someone could 'stand on' the contraption, so things, (Scoping the sub-second mechanical action), at least repeatable.

   That clear plastic column gives a 'third' mounting, and the other two utilize pre-threaded mounting posts, from the 'walking robot dog' toy, I've commandeered.

   Much of the worktime consists of shuffling around (the workshop), looking for glasses, or, especially:
   "...WHERE's my little Phillips ?...?....".     (screwdriver).
We all get that.
   Right now, I'm trying to remember:. Somebody POSTED info, on conversion of TIME, to Voltage, for checking things...that might help me, using two wires, stone knives, and wooden sticks, to measure LAB stuff during product development.
   That blob of plastic, on a green base, is my ROCK STAR, for the season...(guess I don't get out, much, lol).

   Getting close, to measuring crude start-stop pulses...
(On the generic Mechanical Commutator).