Print THIS 3-D Mechanical Memory !

[RJSV]

Neglecting 'VINTAGE' aspects, so to explain:
  Partially a study / speculation of CODE writing dynamics, ALA 1890. Contempory 1890's could potentially include these little (systems) at least physically buildable!
  Think about retail plastics, and epoxies available today, down the street. But also, for doing business, where do you go, to hire your people ?
And, don't forget, foisted bribery and organized crime were present, with those 'feet' in your door.
   No cozy 'Coding Camps' there, 1898 !
===========================

 NOW, info details on large latching ENCODER:
Photo shows one example shaft, for simpler 1 digit latches, very similar. For larger input width, an Encoder uses a reduction and latching process, creating 2 regular sized digits, (00 thru 99).

  Try split that 'monster' input into decades, of 10 hoses each, for a 'split and extract' process. You have one 'Pivot shaft', say '4X'.
Using the photo-diagram understand there will be a setting tab on EVERY column, for extraction of any BUS lane (incoming ball). So, you have your high digit, which can later be read out, only that it's much more numerous in managing, still essentially simple.
NEXT, you are free to combine like signal lanes (below the decade capture device). This combining process is going to destroy your BUS signal, considering the 100 lines are now ten lines (low digit) ready for more 'conventional' ten-state latching.
 
'CONVENTIONAL' Ha!

  Then, the Encoder sub-system has only to latch, say, one of X0, X1, X2, X3, etc up to X9.  So that completes the (Row, Column) extraction.
Usually, that's an answer, stored as a table entry. For multiply, 4 X 2 needs an answer; '08'.

  Clever readers note: an alternate method would transpose, that ending up as '24' resolved, again, to an answer, '08'. That's simply reversing to use a (Column, Row) transposed method... All simple, really.
 
The Mechanical Teaching Computer has a student-welcoming feature, named as a ' SCIENCE' micro-instruction... What does it do ? I don't know: That, waits for the student to think up and build !
This ain't rocket science, in a 'pretend' year; 1888

[RJSV]

" SO... Who sold you this CODE, then ? "
- actual Monty Python related Service Engineer training video
DAVE: We need sub-headings !

  When ya gonna write some code, there ?
  Here goes, let er rip:
 Please also see picture

COSINE_Q:
 Generated  Inst.    Comment
  03, 03.   LDA (Ptr1) get high degrees
  07,.            STA.         send
  09, 04,.          Ptr1++
  03, 03,.    LDA (Ptr1) get low degrees
  07,.             STA.         send
  09,.                 Ptr1--    point back to string start
  05,.           READ.          answer high
  03, 04,.        STA (Ptr1)
  09, 04,.           Ptr1++
  05,.           READ.          answer low
  09, 07.      CALL PRINT_R5.   Print Cosine
=========================

  Imagine a museum patron, drop a ball (Caution, Choking Hazard).  Almost 'immediatly' the printer starts up... All mechanical...
  (I bet the printer will cost more than other, entire project!)

[RJSV]

Errors: The museum patron using Demo machine needs more 'tubes' for an input, getting COSINE (165).
  Could set up a 200 tube version. At any rate, the code is assumed to handle each 'quarter', boiling a full Cosine degrees input, to 75 degrees, and a minus or negative answer, and the decimal point...
  So that is a hybrid, as the table lookup gets a well-resolved answer, while the code does messy customizing, for printing a '-', here, for result of '-97',. That actually is better shown as '-.97', or even better would be '- 0.97', for the COSINE trial coding.

  Now, I seem to be 'teaching' string / floating point introduction, everyone forgetting the Mechanical Computer (not too fast).
  For a glimpse, of why CODING is genuine interest grabbing, try this:
   How far does a de-orbit SpaceCraft move, during one clock, of a vintage Z-80 early 8 bit CPU ?
Speed is 17,000 MPH.

  Looking now, to ROM-CORE, shown in photo, a user has 100 sequence spots, each for a short hose line, down. That 'driver hose' can be resolved, as two digits of fetched instruction.
   "No chance of Self-Modifying Code...(heh heh..)'
 
  A look at my fake object code (up in last post) you can see there are numbers like 05, 05 ; that should really be 55 as there are some single and some doubles, in any case the Cosine sub-routine has some messy inaccurate when looking closer, like even and odd addresses in the 'classic' fetch.
  It's all been done, in that sense. I count approx. 17 program pointer 'steps'. The ROM-CORE outputs are in digit pairs.
  Now, this Sub-routine, COSINE_Q, is for one quarter, the 00 thru 90 degree input range, with implied decimal pt. I think the 'PRINTER' could produce INK right away, plus, the fuller Cosine, will wait, for that dropped ball... Supposedly dropped into bin '165', following that, will CALL my COSINE_Q.
   Fairly quick, PRINT_R5; is CALLed for doing a little string, hopefully sticking a decimal point in there. Maybe, or maybe not ASCII.
You can agree, perhaps less a challenge than BASIC !
Expensive though.

  Ok, Any answers to Space Shuttle (tm) ?
   Answer: 1/4 inch forward motion, each Vintage Z-80 CLOCK duration, ( at 1 MHZ)
RJHayward

[RJSV]

 Errors:. For clarity, the COSINE(A) function needed to show how Cosine needs to print a minus sign, then.
 For details; The math is to find 15 degrees, as the requested Cosine simply 'reverses'. Then, the code doing the answer string puts a coded sign, like integer 11, in a BUS set with those (extravagance), but it does pay off, to watcha buncha grown 'students' argue about strings... They forget, it's so retro...
 Photo has a clear Cosine.
 
  The fuller Cosine object code will first test, if the (museum) thingy is idle. If integer wanders, off of '99' then the following lines start:
  Set PATH = MATH
             Set PATH = COSINE
             CALL LOOP_Block1
///// now printer is initiated///

  The test, of quadrant, is roughly,
  PATH = MATH
     PATH = Subtract d - d
    //// now a test, etc, to do either '-' or '+', etc ///

[RJSV]

  This is a bit of detail, the so-called 'Code Stack' can be summed up, C-64 native speed is 1 micro-sec.
So, using that loop, 160 and the 'run' being 10,
 You have 2 plus 160 X 34. At end is another 4 bytes, for (your) decrement / branch, that is already counted.
  All total, the two ways in comparison, est about 2 times faster, (than typical single action in loop).
  More speed up is gained, using Y indexing downward, that way the loop test is quick.
   C-64 graphics blocks can be Tower, of 24 (lines of repeated instruction).
  It helped, to 'race', the real-time video sweep. Gave the best chance, of avoiding inducing 'flicker', for one thing. Imagine starting, on the 'tail' of that beam-scan, precisely, now it must go round, the timed video, and includes the sweep back (time).

  At a point, the hardware beam 'catches up', but enough time was had, earlier.
  Most likely, both enclosed topics, ... boring as all H377. Some can gain a bit, if you own a working copy, that is nice.
  I, we, used the CART, HESMON gave diagnose-able.

[RJSV]

Oops, sorry, meant last reply post for another topic...
(You don't want me pilot your airpl... Lol)