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

[RJSV]

Focus on the 'A' column the purpose is to set, pipeline style, the NEXT switch box Station, for transfer function, down the Network line.
Basically, 'A' (local output), will exit this box, in the 'Footprint', representing that shaft (see picture).
Notice, a CW (clock wise) turning of the 'chained' signal ends up again CW, by the end of the 4 gears, as each will invert the signal sense, CW or CCW.

  A tough structural item to convey, is the simple bearing and 'head to head' alignment, between two separate shafts: The output here, is using the same footprint location, as the input, although such has geometrical reversal, left to right.  (Shaft footprints are, also, continuous thru the switch box interior: thus reversed depending on viewing direction.

[RJSV]

   One concern is Parasitic Back-Feed, where 2 signals mix, but in ackward fashion, as each separate signal, now together, will 'back-feed' as an unwanted or parasitic signal, that may cause some random harm.
Remember, this is machinery but 'designed' by a Software Expert...(ahem...).
   Remedy is easy examination: one case has the parasitic driving, backwards, the 'A' switch TIP Gear.
However, during that time that TIP Gear is un-connected; just flapping in the breeze.

[RJSV]

Hard to see, in photo the separation, at 90 mSec, of that TIP Gear, supposing gear teeth are, perhaps, 0.08 "
and so that little movement, for disengage, came out to about 7 degrees, (out of the total 60 degrees travel).
A bit of time is needed, overcome the extra heavy 'grease', but also not much acceleration at end of
 travel.
   The full diagram would have, at end, a 7 degrees or so of engagement.  That might be very short, but not a source of (short) pulse.

[RJSV]

   Taking a look at how to use a 'PROXY' set-up, each switch box needs to SET the following switch box, meanwhile applying that to also RESET the currently selected switch box.
So, in effect, the swing travel (should) be same, just mirror reversed.  But, conveniently, the switch here, in the current box, provides a usable 'substitute', and this the term used: PROXY.
   The overflow signal indicates that the typical 300 mSec travel delay, has made contact (that's why you get that rotary 'overflow status' signal.

[RJSV]

   I would seldom consider using the term 'delightfull', to describe a gear train, but here goes:
   Picture shows, (maybe never going to be clear, really)
The gear train runs towards the left, that being the 'A' positioning motor. Two 'alternate' lucky breaks are that a 'backwards' and unwanted signal that 'B' Output sends, as a result of legitimate sending to SET  switch 'A',  that signal also folds back, a parasitic that goes into the 'B' overflow (output).
However, good news is, that RESET Overflow gear is currently in an unconnected state, (when B output happens, the 'B' switch is, by definition, positioned away from the overflow.

   Same good news, for the parasitic signal the other way; When 'A' is being RESET, that signal, from 'B' overflow (while being RESET), will back-feed, to the
'B' local output; but no bother...again, by definition, the 'B' switch will then have been positioned, in the open or RESET condition.
Very confusing, I know! (I had to design that shiii)

[RJSV]

   A chronological treatment is going to help for understanding over-all function, of these pipelined switch boxes:
   First diagram; From a box's point of view, an arriving 'SELECT', arrives, on the 'Local' or 'L-in' shaft.  That gets moved over, from left to right in diagram, resulting in driving the positioning 'motor' element of switch 'B'.
Looking up higher, straight above is the upper layer in this expanded view and that has the little toggle, that follows the lower toggle positioning .
The 'B' shaft is then going to drive anything, but for now the shaft is stopped (dry switching).
That's going to take some 300 mSec, nominal, for toggle travel.

[RJSV]

   Readers should note that it's the 'A' shaft itself, that powers the SELECT action, (to SET the 'B' switch).  That is, from current (new) box point of view, the 'A' shaft actually starts up, rotating slightly before that SELECT starts coming in.  (More in a sec.)
   That's because the 'A' 'generic' shaft signal, when asserted, is there in all the boxes down the network line. (The other boxes are just not in selected mode.)
Note also: I corrected the 'A' switch position depicted, supposed to be in RESET position.
   Looking at the gear chain for the 'A' motor, that's shown with a marker pen set on those 4 gears. Ignore, please, the front little chain, (4 gears, they have a little paper slip along them). That was previously discussed.

[RJSV]

(Previous diagram).
   Interesting, the 'main' just, of that gear chain, the 4 gears (underneath pen) form a path from the 'B' local output (that's just by turning the shaft, while selected, some custom connect), path to the 'A' motor input.
So that looks pretty 'normal', but hold on...:
   The upper, follower section, of 'B', sends down, and into this chain, but, ALREADY there is connection, to the gear that conveys an overflow...Maybe not obvious, but feels 'reckless', or like a:
'. '. MECHANICAL. SHORT. CIRCUIT ' '.

Uhhh, I think they call that a 'STALL', ...pretty sure.

[RJSV]

   I'm assuming a (rotary) signal gets down the network line pretty quick, like 1 second or less. Assuming, say, 12 stations, each 2 feet apart.
In each station, that rotary signal gets thru in probably maximum 50 mSec.  Funny then, the nominal gear start-up or accelerate time is 10 mSec per gear, : That means significant time, especially with, let's say, 8 gears.  That's an 80 mSec time to start rotating at speed, and signifies how a switched signal can 'lag' in time, behind the 'raw' generic shaft (any of 'A', or 'B', or 'C' series BUS chain.

[RJSV]

Let's stay on this TIMING thing, a bit more:
   So impulse #1, that was put into the 'Box Input', and goes, internally, directly to set (that box's) 'B' switch.
In the string or series string network, impulse #1 serves to SET the next box, from (expiring) current.
   
   Impulse #2, that's where channel 'B' outputs, again internally only, the 'command', as it were, to SET the 'A' switch.
I'm figuring, by the way, sending BASE CONTROLLER knows which station / which delay or pulse length, etc to use.
   At 50Sec. each, 4 stations should be OK at 200 mSec delay, before assuming you can start timing, plus whatever action time, for switching the network components, that it.
   So that is saying, at Station 4, when you want to impulse one of the internal switches, your Electric Motor impulse should last, at least 200 mSec + 300 mSec, for best reliable switching.
   So, STEP #3:). Now that channel 'B' had SET channel 'A' to active, and assuming you've done all of that 'custom' output from channel 'C', now it's time to send out, to next BOX, a SELECT for local hand-off.

Step #3 is kind of complicated:
   It sends, out via the 'LOCAL' shaft, to SELECT next
   It parallel sends, via LOCAL, to RESET the channel 'B' switch, right here in this box, but also,
   It uses the 'Saturation Overflow' to accomplish RESET
back around the loop, as that reset takes away the power that is doing all this (that's switch 'A' being active).

   So, that's a few steps, but, aside from any complexity with channel 'C', the actual impulses are simply, 'B', for 500 mSec and then 'A' for 500 mSec.
If no activity for 'C' output, it takes a pair of those, to get down to each next network box.

[RJSV]

   You get a TON of diverse ideas, using that switch with the second, logic follower layer up top.  Several ideas, at once, can pop up; (if you are enthused), including a very 'traditional' looking 'Edge Triggered' pulse 'circuit'.
   In photo, the diagram has a drive (rotary) signal arrives to the 'D2' gear input.  Not shown is an additional connection option: By connecting the switch motor driver to the switched output, (D1), you get an interrupted impulse to the motor, that will start to dis-engage the whole switch (toggle lever).
That's one layout method. Another, places that motor input to the basic circuit input, this the motor keeps running, strong, even after pulling the reader portion free of any drive input.  That's going to produce a classic RISING EDGE TRIGGER.

[RJSV]

This photo, features the latest shift, from schematic and (endless seeming) diagrams, to a test bed, for getting some timing info.

   Landlord is Elec Tech / Engineer with many nice items, like the TEKTRONIX old school SCOPE.

[RJSV]

   Using the first test bed, I want to try out some of the many assumptions and cautions, (mostly concerned about parasitic signal feeds).
   Probably try a toy electric motor, as a generator of some electrical output, for triggering that '465 scope.
I figure, roughly, a motor shaft spinning at 30 revs per second, that's going to be 1 milli-Second per gear TOOTH, for 30 tooth nylon gear.
I have used the light metal guitar string, for various springy contacts, so that is another option, for generating a little pulse, for diagnostic uses, in the LAB.  (We use a Lab with 'car storage' option, and fresh air).

[coppercone2]

as a side note, did anyone start researching spintronics after looking at all these pictures... just for namesake alone?

Something kinda makes me feel like eventually someone is gonna figure out how to use monoatomic? hydrogen as a gear

[RJSV]

   Here is a pretty wild example, using commonplace analog concepts...(and YES, you can squeeze some analog, even from 'digital' switches).
   Using your custom 'C' channel, suppose you have a moveable 'STOP', that can be adjusted, in some separate action / components.
   Now, that picture shows, the temporary STOP is now at 48 degrees, travel from zero. Course, you lose any right side gear function, as that requires full travel, to insert gear teeth, of moving TIP rotor.
   You can arrange, for PULSE duration match, you would have a 'PROXY', that's just a useless looking, no output switch, that provides tracking info, regarding being in step with another gear layout; that being in the next box, or wherever.
   But back to function: The arrangement is to run the two gear trains at same time, and cause any lingering output, in the compare, to be 'ERROR', for feedback.
AHH: but not feedback to gear train run, but to cause the slight adjust, of moveable stop.
That way you can issue repeatable impulses, even though the actual components are transient involved, or, said another way: The process, for a single time interval, is DESTRUCTIVE, in that you cannot repeat, without an 'image' source somewhere. That's done by that moveable stop, that does not change, except by another, separate action.

[RJSV]

All kinds of so-named 'RUN-OUT'  schemes can be tried, for generating (surprisingly short) impulses.
First of all, just a simple Run-out can be done, shutting off motor power (the toggle pulls clear), and subsequently the toggle stops it's arc of travel.
   But, importantly, an output is tracking that short bit of inter-gear 'drama', uh...and that means Output stops soon, too (Output is essentially in parallel with the TIP gear).
  Took a sec. to illustrate some possible Spring Detent, there...
   Now, I say that switch Toggle travel, in that 60 degree ARC, takes, est 240 mSec. (divides out to 4 mSec per degree, average), and, showing in the picture 11 degrees of swing ARC, why isn't that, 40 mSec ?
Well, the 11 degrees of start-up travel are probably the slowest, due to huge GREASE viscousity.
Plus, please note that those three gears, in photo there, don't turn, or at least AREN'T WORKING.
They could just as well be sculpsures (my favorite),
or...heck... little toy 'King Kong' figurines, those being transported / dragged upon the disk of their fate (The regular big motor gear disk)

[RJSV]

Looking at (this) photo, Readers may wonder: That looks a mess...how could such a wild
   ' A LION, STANDING ON ANOTHER LION'S BACK'
...how could that, you know, work ?

(It does the compare you'd need, in a PW servo.)

[RJSV]

(double-decker)

[RJSV]

...That double layered, does some fancy footwork: first, the lower half will position the upper, within normal 0 to 60 degrees pivot.  Then, second, the upper layer (yellow components), is simply spun, a little, and is designed for that to be irrelevant.
   Notice, way up top, the red gears bring in the 'minus' or negative quantity, for 'backwards' subtractive effect, on this compound toggle.

   This would do, for example, 200 minus 190 = 10 mSeconds.  Now, really, the actual quantity is gear tooth count based, and some slip/ drag.

  Anyway, the result is an 'error' signal, to be fed to the static PW Proxy.  Hopefully, as a servo, next time the error comes down.

[RJSV]

  Regarding that Pulse Width copy process, for SERVO use, the included diagram features a 126 milliSecond count, against a 118 mSec duration 'Test' case, for determining SERVO response:
   Sorry, the arc of travel, of the moving TIP gear should be shown as 'running' in, (or back), not as 'running out'.
That's running out, to a moveable stop, on the right, but no overflow.  Could be seen as a prep move, for next impulse, to be, again, with ARC of travel back towards the left overflow gear.
(SEE how easy, things gets bass-ackwards....
The 'subtraction' process is quite adroit: coming up with an 'error', as the first (top) timing is longer.
   Process is to then feed that 8 mSec 'error', back to the adjustable stop, in this one, repeatable SERVO action.

[coppercone2]

You know what makes this more interesting now then 50 years ago, is that you have a high speed camera in reasonable smart phones. When I started seeing timing numbers pop up, in this time range, I kind of wonder if its within the realm of using a iPhone slow motion mode to prove this design.. this used to take serious money I think. Well, or it was a huge pain in the ass with a bunch of encoders and electronics. Talk about rapid prototyping improvement... when I think about this measurement problem and the capabilities of a phone. Timing a whole bunch of gears with a phone video is a gigantic cost reduction for a prototype. I think if you got a tripod or put the phone on a micrometer arm (and maybe some mirrors/periscopes on another one) it might be able to give reliable and reputable scientific data on a complicated gear train so long its not shooting oil everywhere.

The magnifier app, seek thermal cam, slow motion cam, magnetic field meter (careful with this one, might be a gimmick).. and you can even hook up a sdr, high sensitivity microphone (acoustic signal analyzer) and external optics to it (macro lens, microscope lens).. not to mention it makes conferencing easier with photo upload

 I wonder how far it is from being a real life tricoder, useful for science.

Tried the slopro app for iphone @ 1000 FPS, quite impressive.

[RJSV]

  Hey, a distraction for a minute:
  That R/C car has steering, and reverse, from remote.
...oh and was $ 9. 90 drug store chain. 
Likely great motor power.
  Make a good SCARE -CROW !
   We got CROWs, bad.

[RJSV]

  Hey Coppercone2, how are ya ?
Thanks, man, for, especially the micrometer mount.
I do have kid-style microscope.
Reminds me, I met a person, INDIA borne, that young person demo'd his 'Tech Rig', a soap box on wheels.
But that kid welded, soldered, fixed any electronics, on the streets, of (Bombay?).

Anyhow, I bet I would enjoy a good, mechanics shop supplies, for specialized, ultra-accurate shit, (that is, a catalog for lathes, etc. instruments.

[coppercone2]

I had a look at spinning some high RPM stuff (like a kitchen blender) and it looks like the iphone app on a iphone 10 does a good job of looking at it.. a bit blurry but maybe the light is not enough.. if you are serious about high speed video you need very powerful lights. If you put a bright color on the gear it should be able to look at those response times, 1000 FPS is 1mS.. so you should get 100 frames of your object in slow motion. I don't know if they do some kind of digital magic that makes it less useful for actual science..

I had these old gear boxes I found before, small ones, about the size of a film canister, that were ultra precise with reduction ratios, brass precision gears and a slot for a photo diode break beam encoder.. I imagine that's the kind of thing you would use to 'test' the system you want. When I looked on ebay A while back they wanted like 200$ for each of those gear boxes manufacturers website was like $600+


Right now I am experimenting with using the FaceID scanner for 3d scanning. I think its quite impressive at making 3d models considering how poorly I took the pictures sliding on my ass around the floor to map a drink can... I was thinking about making a turn table for 3d scans and improving my home made iphone tripod that is too light (got some aluminum round plates from the trash that I can glue it to, and Possibly use two more plates to make a rotary base for it.. just by press fitting them with some retaining compound on a old hard drive motor. Great bearing, no cost, and can be computerized if desired later on (I think they are variable speed so its no problem to make it rotate slow?)

Anyway.. if you never tried to harness the full power of your phone sensors its pretty fun.. gonna look at the lumen sensor and maybe try a sound FFT app also.

I kind of wonder too if you can make a frequency counter for gears from a phone's magnetic field sensors for cheap RPM measurements. I think I had a app that graphed field strength on an old phone.. maybe someone made a counter.. great way to suck your expensive phone into a giant gear box too

For gears, if they still had an audio jack on the phone, it would be trivial to fit a inductive pickup probe into it in order to get RPM recordings too.. that would be pretty high quality industrial grade stuff. I think they have a break out cable (I modified my inductive pickup probe to a BNC cable, so I would need a BNC to phono to iphone adapter).

https://electrical-engineering-portal.com/wp-content/uploads/2014/04/inductive-sensors-purpose-working-principle.jpg

It's probobly possible too to modify one of those ultra-rugged iphone cases with some 2450 cells (or flat LiPO) and power LED to make illumination for high speed video work

[RJSV]

   Thank you!
   Believe it or not, COMMODORE 64 has /had awesome real-time performance.  You build a framework, like, say, 100 uSec per click.  Course, your 50 Hz or 60 Hz video is going to be 16 milliSec per field for 60 fields per sec rate.

   Anyway, much of this is looking like 1 milliSec thru about 999 milliSec range, like 300 mSec typical.
I thought,  gear TOOTH took up 1 mSec, each, to pass by...

   Right now, I'm distracting myself with design for output 'SQUARE' waves, roughly:
  A mechanical 'Square-like' impulse features 450 mSec on time, CCW (that's counter-clockwise), 300 mSec off or coast-down time, and the same for CW, another 750 mSec.
   Total repeat time, of 1.5 seconds.
and with decent stationary time, for less stress when change directions of rotation

   This thing would have like 3 or 4 special gear / switches, but one of a kind test ew. so not multiples.