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

[RJSV]

Coppertone2:
  My recall is that a coil, say wound on thick nail, is fairly weak, in voltage output, especially compared with a piezo sensor. Like, 0. 070 volts vs. 1.8 volts, for loud noises.
But, with added X8 OP amp circuit, (Lm324 has four),
not bad voltage.  Always possible, some foil sheild sometimes grounding that, and wrap around, will help, for hf electrical noise, nearby .

   Of course, for scope trigger, could always trigger on electric dc switched in, for toy motor, then, somewhere, for diagnostics, another toy motor used as generator, creates 'pulse', but that's kind of vague, and besides; single shot stuff on scope won't show well, especially is it changes, each trigger time.

[RJSV]

   These 2 diagrams, (enclosed photo), perhaps not related directly, show the kinds of timing, in the milli-Seconds range, generally.
Bottom diagram features a 'SQUARE WAVE' generator, having also those periods of mechanical output 'rest', at standstill while gears still turn (somewhere).

   Upper diagram features the timing (I think...), estimations for the two existing 'OVERFLOW' gear outputs.  The 'L' (left side) gear will turn for a bit, RUN-OUT, until toggle lever TIP gear disengages, approx 90 mSec.
The 'R' (right side) or SET OVERFLOW will run, after TIP gear contact engagement, for as long as input.

[RJSV]

I had a really tough time, on this SQUARE WAVE generator. Thinking it would be easy, you have your ON time, plus another timed 'off' time period...at least in ttl electronics...
   This thing does the sequence, for 1 SQ wave cycle, in 'brute force', setting an output and timing that (T1 in diagram). You get the needed positive contact result, at the end of T1, for an assertive response, which is to issue an 'End of T1' signal, signifying the switch had reached end of swing travel.
That starts up action, to soon pull (switch T3), to end the cycle timed by T1, plus the short wind-down T3 represents.
For the (empty action time) of T2, that signifies the off or stationary period, timed and issuing an 'END' pulse, to next section.
  The next section, not shown, is virtually the same, except that it times things for a CCW or counter- clockwise output, in the SQUARE wave.

   Also not shown, is the reset, or 'RECOVERY' mode, where everything (in diagram) needs to be returned to start, for another timed square wave cycle.

[RJSV]

   As to any errors, there are plenty...for example, that previous 'SQUARE WAVE' Generator layout showed incomplete waveform diagram, it needed the second 'rest' time, that brings that waveform to 1. 4  seconds, or call it 0. 7  hz.
   Now, innovation for avoiding complete self-destruct, by abrupt direction flips suggests: DON'T DO THAT...
But that doesn't help.
   What can do is:
   Double-up your little system, this is Test Equipment, (potentially), and only making a few...there.
So, one impulse timer/generator runs a pulse output, CW or clockwise, say, once per second.
Now, with attention to phase, exact 180 ° (degrees), another separate little gear layout does the exact same thing, out of phase, and going south (pulse rotates to CCW or counter-clockwise).
Build those two simple sources, and, diagram shows
a simple phase sensitive controller (gear-switch), coordinates start times.
  The two, opposing-coordinated signals mix, logical-or style.

Possibilities include un-conventional timing overlaps, doing unexpected things...

[RJSV]

   A brief run through the issues of scale distortion, just common friction being stubborn; I.E. viscous grease + drag...
   Figured, est. that 0 degrees thru 60°, to be way, way slow, at first, at 42 X SLOWER !
Wow, but the stick is static hold, electrons got comfortable stationary, etc etc.
  So, after first 11 degrees or so, figured estimate was:
Go up, decent, and turn curve (see fig.) slope down; Classic 'S' curve, where speed became, essentially without acceleration. Otherwise called peak limited.

   The right side diagram features that same 'S' curve function. Notice, starting the diagram for clockwise, that curve starts with 'jump', to 90 mSec, for only 11 degrees, a bare sixth of total angular travel.
Seems fairly normal. Big dots signify more linear response, travel position, VS. time.

[RJSV]

Ok FOLKS:
   POST A PICTURE, of your ROBOT. DOG

   (Mine is posing next to some nice wood burl box)

[RJSV]

Hey!  Ha ha:. Jokes on ME !  :
   I figured a lazy break... Show your ROBOT,. bla bla
But, along the way, That DOG exhibits the very same little glitch, in movement, that I feared.  The action of switching the toggle, when you've mixed toggle movement signals, INTERMIXED, with the 'data' being switched, well, that creates approx. 120 mSec. of backwards drive, (to left side gear, in this case).
That's before the formal, desired output direction takes over, so (you've) got to wait...until the TIP gear comes clear, of the former contact with left side output.
   If you recall, that moves, one of the linkages, outside of toy's gear box, that moves the legs, in little circular patterns. But the brief movement is obviously backwards, from point of view of one of the Robots paws.

[RJSV]

This current photo, makes the case against NETWORK propagation delay, at least some.
   You can see, as the signal or 'work' flow is from that electronic Base Controller, having small electric motor outputs.  But the shaft, it simply connects, straight is best, down a 'series' line, on occasional bearings, in each distributed box.
  I'm thinking, say, 20 msec to box #1, from controller, then 400 mSec, for a Station box out 20 feet away.
Then, internal switch swing times of typical 300 mSec plus ..
  So might be, at end of string, 400 + 300 = 700 mSec.
And so, instead of playing '700', against 380 mSec,
I just throw hands up and say: " Do 1. 000 Seconds, for a few percent over what's needed, without case by case delay.
   Often, with a 'Human', we can take 1 second, just pulling arm back, after pressing 'START'.

  Cant see good, thru my intentional 'smoked plexiglass' boxes., But only one box, at a time, is with switch TIP gear engaged.  All those other Stations, with their own main (un-engaged) gear it's just hard, sometimes, to conceptualize it all, so hard, while keeping some computer science function detail in mind.
   But, so anyway, don't forget, the main method is to send, to excess, that one single impulse, of concern, trusting switching (or whatever) will be complete
Meanwhile, down the line, there, the Station switch performs an 'auto-stop', of sorts, self switching upon completion of some local task.
For example, when switch 'B' gets cleared, overflow after positioning completion, then proceeds to reset switch 'A'.  So the box kind-of self disconnects components, for least drag but also so that function doesn't 'Do the UNDEAD' thing, later.

[RJSV]

Here is more detail, on the 'Proxy Gear' similarities with that tear-down video, on " FORD MARK II Range Computer"..something like that.
  Inside, a GEAR acts as a real model in a simulator intent, (partially what I thought a tracking gear could do, for my particular issues.)
The gear position symbolized an aircraft under tracking, while another gear symbolized 'velocity', (I guess, maybe).  Plus, acceleration, (the 'third tier' in calculus integrals) was also in there
Radar input provided 3-D info (? maybe?).
Purpose was to swiftly move the big anti-aircraft guns, but as time brought out the Jet propulsion, well, the FORD MARK II Computer faultered...

('Proxy Gear does similar work, following some remote component, but also provides valuable longer term timing (delays).

[RJSV]

With all this subject jumping around, Readers might appreciate some more depth, describing that SQUARE WAVE GENERATOR.

   (Please see diagram).  With T3 switch positioned in connected mode, the application of 'power' by way of input, starts an Output going
   The idea, for timing this, is using Switch T1 for a 250 mSec delay, before turning off the impulse Output.
(Meanwhile the input continues to supply the power.)

   Each timing switch ends up in a cascade state, passing output rotation continuously after end of cycle (passed to next component).
For the 'OFF' time in this cycle, another switch-timer, T2 will, as the others, run the time, 250 mSec. and then start sending the 'T2END' status Output.

[RJSV]

For the Sq wave: Diagram shows 'T3' at bottom; that's the output you want, timed to 340 mSeconds.
Notice the waveforms  T1, and T2 finish and then just 'hold' there.  Each separate timer can do that, for a cascade that enables the next component, without additional explicit switching.
Don't be fooled, by thinking a time interval is over, when the timer reaches 'Sat.', as there is still a short bit of work, (and rise times), before effective end of this little cycle, and start of next.
Notice, also, turn-off IS faster here, at 90 mSec, than the bigger swing 250 mSec for the 'assertive' contact switch end-field.

[coppercone2]

this came to mind, the 'no go' zones of these kind of contraptions are bound to be interesting. Then you start thinking about flex of the chassis to allow the gears to some how walk over each other or something so they do not break teeth when it has a aneurysm. So the shafts in the gear box and the gear box walls are all spring mounted and it yields and has a (undefined) spasm instead of breaking bones. That sounds like it belongs on a battlemech (from the battle tech universe, how do you make a bipedal tank keep lumbering after it gets hit by a rail gun?). that might be applicable to walking proteins too, since atomic bonds are stiff but often described with springs.

[RJSV]

Here is diagram, probably still imperfect, but to show the 'cascade' sequence, at the end everything's streaming out...
   But like before, the good Output you want is from 'T6'
and there are some special details, keeping rotary direction proper, for the negative half, of Square Wave.

E.O.C. (see right side of diagram), means 'End of Cycle' and that would start up a process to 'restore' the various toggle levers, for next plus / minus cycle.

[RJSV]

   This looking, 7 gears -switches, course they're TIMERS, and do special jobs, like issue overflow / underflow status...(like some 1982 z-80 CTC beast).
And so being TEST GEAR (no real PUN), that's no sweat build.
   The arrangement sends out Clockwise, 500 mSec, then 500 rest stationary time.
That second half is simply blank timing, as the zero part of Square waveform.
   Same for the CCW or counter-clockwise Output, you get 500 milliseconds on the CCW labeled Output, then 500 of stationary rest.
Notice it's two fragments of a SQUARE WAVE, needing a mix.

[RJSV]

Coppercone2, I'm going ta need you on that...WTF
DIVIDE BY ZERO ? ...HUH ?

...what happens, if I press 'PLAY' ?
   Will it: PLAY another video ?
    ...maybe it will 'PLAY that video...'...

Huh, ?...

[Bud]

Can we nominate this thread for the longest Solo thread of the Forum ?     

[coppercone2]

I just mean that the failure modes of timed mechanical systems are interesting, since you are getting into pretty precise timing. When you start counting miliseconds thats getting precise. I like how the calculator loops harmlessly rather then getting jammed up/broken by incorrect input. It could have probably been built so that it explodes gears when you divide by 0.

[RJSV]

Actually, I had been thinking; What about a video, that can run any program, or PLAY any video...
Was this an old-school prank play?
   Spatt:
   Run Splatt;

At any rate, you get that (120 millisecond) glitch, when even the follower stage has CCW coming in...
You get a 'pull-away' transient output, very un-planned, but surprisingly benign.
Like, when an un-designed feature sends that 'backwards traveling' signal, but to un-selected output of something...Not even touching...

[RJSV]

   Here is (photo), almost buildable, and that's really only two gear assemblies...
   To start your SQUARE WAVE cycle, a 500 mSec impulse, going CW or clockwise rotation the INPUT comes up, feeding through to the Output TD-3.
TD -3 is kind of a 'Normally Closed' switch; when it is turned it disconnects, typically in 90 mSec.

First, TD- 1 starts the swing, over to contact the next (TD-1), and, notice that TD-1 times through a bigger swing, than 60: It swings about 85° for a delay of 400 mSec, before starting the disconnect to be 90 mSec 'fall time', of the impulse.
It was fun to go ahead and skip any 'intermediate' gears that would connect the two separate timers!

[RJSV]

That last schetch featured a 'disconnect', at 90 mSec and a good place to point out: that 'TD-1' timing set has a, literally, positive assertive time-out, that being the little impact, and connection to overflow Output gear.
Looking at 'TD-3' that's not the case, as the Output just becomes unconnected, and so does not work any loads, as it slows to stop.
   So the task is to time the next part, of the SQUARE WAVE, that's the OFF or stationary, for 500 mSec.
But first, to get that start, best way is use another gear, for timing a specific, assertive end to the pulse time, regardless of exact match, or not, with disconnect and stopping.
   Of course, any TEST EQ like this, adding some components is no big deal.

[RJSV]

   My first choice is to continue my work today, amid bad news/war, for many.
Yes, I watch, and argue among friends...(we disagree).
So, please forgive, I see the troubles, but work is work, so...

=========================
=========================
=========================

   Going into a modeling / prototype phase, for the (Mechanical) Data Network, but also interested in mechanical LAB DEVICES, instruments such as
Square Wave Rotary Source Generator.

   As to the timing: I'm looking for what I like to call:
   90 / 250  ...which means typical expected is
   250 milliseconds, to travel the moving switch TIP,
   and 90 mSec; that's how long you get rotary output, when 'pulling' the switch gear out of contact / engagement.
The 250 mSec timing possibilities are nice. All without any explicit 'gear train', at least not fully
For estimating: The thing obviously takes some (wild) fraction of a second, and most certainly not 'zero' elapsed time, ... 250 mSec has a darn good chance !
  Audio sounds, like quick loud hand-clap, are fairly close to 50 msec, and a 'slap-back' from (concrete wall) nearby, that's around 50 to 120 mSec.
That Toy I'd mentioned, dog's tail moves about 4 to 6 times in one second, I figured.  That's approx 170 to 250 mSec.
By the way, in case wondering, the TOY runs at rate of 2 hz, at the later stages, into the directional switch.
?? Numbers don't match, above, but will figure things out, with that 'open face' special model (see photo of prototype, with transparent face, for testing.)

[RJSV]

Taking a look at speed parameters, and the increased voltage pulses,  the numbers come out to a shockingly LOW LOW 1 / 17 MPH ...
Yes, that's about 6 % of 1 MPH  !

figured like this:
   Second half of (toggle lever) travel, at 150 mSec point (of 250 mSec total), things are up to speed (in viscous grease), say that 1/10 inch takes another 100 mSec of travel:
   That's an easy 1 inch per second (0.1 inch in 0.1 sec).
Now, that rate can be increased, by using DC pulses higher than 3 V, say 9 V, for 9 times the power (SQ law),
that the 3 V runs at. Maybe, est 300 mA X 3 V is only about one watt.
Now, 400 mSec might even be too much ( heat gen), in a 500 mSec ON, 500 mSec duty cycle (50 %).
Maybe the 250 mSec pulses, issued thru transistor buffers, would be tolerable for shorter runs.
Heat sink, on motor always possible...

   With shorter swing ARC, right now, could get switching times down, along with that motor pulse Voltage increased to 9 V, 600 mA (transient 1/2 watt).
Perhaps, times reduced, to 90 mSec 'rise delay',
  and also reduced 'fall time', or 'disconnect duration'
(Actually, more like a little delay, then gear winds down).  That's easy some 35 mSec...with any luck...
   Some 4 Hz potential, without too much unusual construction.

[RJSV]

This enclosed diagram lays out typical switching time durations, (to accelerate and move the switch toggle lever).
   Using Google, the typical EM relay (mechanical moving contacts), operates at several milliSeconds travel time, with specials doing 100 uSec travel times.
So, that is a benchmark curve set; SEE LOWER CURVE, please.
   Up top, with an easy, demonstrated timing clustered near 200 mSec, readers should realize; that kind of time, basically one fifth Second, is on similar time scale, as a quick arm and hand movement, (in sports, etc).  Going under 1 mSec for a mechanical device, that's just awesome.  ...Using previous thoughts, that 'speed' would be:
   250 X the 1/17 mph calculated.
So...that's 14 MPH average for the 'TIP Gear' ARC swing.  (If I did that correct).
  GOOGLE 'Small signal relay' for more history info on that older technology.

   Some scientists get bored with this sort of (endless seeming) narration, covering a single switch action, blah blah blah: BUT:
   I've also heard that, in the past, many SEMICONDUCTOR INDUSTRY people spent lots of time, on 'pesky details'...too.
(Lol, TIC). (That's 'Tongue in Cheek).
-- Rick B.
   

[RJSV]

   For a bi-direction data interchange, couldn't be simpler:
   For reading a Switch, let's say it is a thermostat, or something...this (scetch in photo) shows a few boxes in series.  To read your special switch, (of course, when at box # 14, or wherever), BASE Station sends, via shaft 'C'; that goes thru sensor-switch and, thanks to BUS method, is able to DRIVE the 'A' BUS shaft. That, of course, goes to every Station, but this action is done while box #14 is in 'SELECTED' Mode, meaning some drag on the 'A' shaft, but only one Box at a time, anyway

[RJSV]

Continuing on, with a second photo, filling out that mini-Network; And now shown, on right side, from Station #13, That box is doubled-up, with separately addressable Station #14, for doing that M1 'motor' function, to move the Switch that can be read, later.
So the two stations would be in same box, and cable of sending the 'single bit', of data, really, and BTW encoded 'rotary', that is CW = 1, CCW= 0.
   Notice, maybe, you can store '1 bit' that way,...right there in the remote box.  (And simple, too!)