# Printing a protractor?

wrote:

With the voice of experience, not so neat!
OK if creating a spinning knob to control something else that has its own position indication, but not so neat if you intended to use a calibration on the rotary encoder itself, as was my intention.
I created a 20-division circle (using Alan Bains' program again) for the flywheel/knob of the rotary encoder, but the problem is that the increment and decrement positions are spaced 1/2 of a division apart (but both within the 1/20 division allocated to that position)
When you move the flywheel one division, what then happens depends upon where you were in the 1/20th space, ie, just before or just after one of the increment/decrement positions.
The net result is that you slowly lose calibration after changing direction several times, and the availability of the protractor (which is where we came in on this thread) becomes an absolute necessity and not just a convenience.
Still, if you don't experiment you don't learn by your mistakes!
A flywheel attached to the rotary encode makes for a very light (ie extremely senstive and not an emergency flare) touch, and it does not take much vibration or passing wind for the dividing head to move on a few steps. Therefore, the circumferential brake now becomes a necessity, and previous talk of magentic braking is redundant.
Again ... Still, if you don't experiment you don't learn by your mistakes!
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
With the realisation that I can now produce accurate protractors using the laser printer, and that it would be a small extra effort to produce a matching vernier scale; and the further realisation that with the power of Excel I can produce numerical tables with little or no effort, do I even need the stepper motor approach?
With a protractor circle of 200 divisions, coupled with a 10 division vernier and an 80-tooth bull wheel, this would give me 80 * 200 *10 = 160,000 divisions of the circle.
Suddenly, and it is a "conversion on the road from Damascus" resulting from Alan Bain's introduction to the joys of Postscript this morning, there is no need for stepper motors (with or without micro-stepping) and no need for the Micro-Attachment of the Versatile Dividing Head developed by George Thomas. (With all due credit to him because he was working before the days of the widespread availability of computers and 600 DPI laser printers).
So, with a vernier-loaded cursor, a protractor of 200 divisions, and a circumferential braking mechanism and armed with Excel-generated tables for any number of divisions (and of course a second protractor on the faceplate so that you can read off how many whole turns you've done), you could divide down to a reasonable tolerance any no of divisions and not be more than a gnat's cock in error.
A further advantage is that you won't need any division plates at all!
This all seems to good to be true. There must be an error in my analysis, but I can't see what it is.
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
On Tue, 4 Nov 2008 14:28:40 -0000, "Amateur Machinist"

If you use the entire length of an A4 sheet of paper at 600di, your best resolution is 7016 points. This assumes that the rollers in the printer can drive the paper with more accuracy than 1 part in 7000, that the paper won't distort by more than one part in 7000. and that the paper length fits the circumference of the disk to better than that accuracy.
The principle works, but don't expect too much from it. I spent a month over Christmas in China in 1995 Using a similar concept (with Nomex tape) when measuring torsional vibration on a steam turbine The raw errors were far higher than desirable.
If you only want the original 200 divisions, you won't have too many problems :-)
Mark Rand RTFM
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
wrote:

1. ISTR that when reading about the drilling of division plates that any angular error in the position of a hole would be reduced by the worm / wowm wheel ratio. The same would apply here.
2. Assuming that you are addressing the paper to be stuck around the circumference of the faceplate; that is only a secondary consideration and gives you a warm feeling (in the stomach and not in the seat of the pants!) about how many whole turns you've had. This removes one of the uncertainties about traditional dividing; how do you remember how many turns you've done? Having a second indication on the output shaft removes that uncertainty.
3. With the worm set up, then it would be a simpler matter to use it to engrave the edge of the faceplate so as not to have to use paper in that part of the application.
4. There is, I agree, a problem with paper stability, especially in damp areas. Perhaps lamination in plastice would resolve this? (I don't know).
5. Once the principle of not needing any division plates has been established, then it is unnecessary to proceed with the paper approach. Traditionally engraved scales on metal would provide all the answers.
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>

That sounds an interesting project. Care to describe how you went about it?
How would you know that the level of raw errors were "far higher than desirable" unless you had some form of superior measuring system, and, if you had such a system, why wouldn't you have used it in the first place?
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
On Wed, 5 Nov 2008 09:11:03 -0000, "Amateur Machinist"

That particular one used nomex tape cut to form a stencil with windows in it. Said stencil being exactly the circumference of the coupling (give or take too much :-|). The stencil was attached to the coupling, then the whole lot was spray painted with mat black paint while the rotor was turning with the barring gear. The stencil was then removed. An optical pickup was used to read the marks.
Working out the raw errors can be done by monitoring the torsional vibration as the machine is run down from a trip. It can be assumed that there will be little natural torsional vibration with an unloaded 500 ton rotor. the datum run showed that this was the case. Signals that stayed constant all through the speed range could be assumed to be due to fixed errors in the paint marks These tests allowed me to map out the errors on the raw signal that could then be subtracted from the signal during the actual vibration test.
The proper test was done by running the machine down whilst the generator had two phases shorted together. This makes it a single phase machine and causes a torsional vibration at twice running speed. I had argued for the generator to be excited during this, but the factory wouldn't allow that and I had to use just residual excitation.
Later calculations ( It took me 8 pages of maths) showed that the factory limit on single phase short circuit current was only of any relevance when the other phases were still at rated voltage. The torsional excitation we were creating was only 0.15% of rated torque rather than the 3% the designers had told us we'd get. Oddly enough, none of them had done the calculations from first principals:-(
Mark Rand RTFM
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
On Tue, 4 Nov 2008 14:28:40 -0000, "Amateur Machinist"

I guess my thinking went in the other direction - I could get all the resolution I needed with my home-made VDH-style dividing head plus assorted division plates, but the BIG problem I found was that even with a written list of positions for each division, it was easy to lose track and end up with a thin tooth on a gear, or whatever. Fitting a stepper motor and then controlling it from a PC or a PIC chip completely transformed the dividing problem as far as I was concerned - you didn't have to mess around with spreadsheets and then still get it wrong; the electronics took that problem out of the equation.
Regards, Tony
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
wrote:

That is resolved by having the second protractor on the output shaft to tell you how many complete turns. With a numerical indication both on the input shaft and on the output shaft, you can read off the position at all times and cross-check. If you're on the 71st tooth, say, and the tables tell you that you have to be on 57 turns and 193.3/200ths then you can just read off your position from the vernier directly and correct accordingly. (Figures above given for illustration and not calculated for any real configuration)

If the power fails on you, you have no guarantees. You especially have a problem with an electronic controller that you cannot guarantee to come back on with the same phase excitation that it had before the power failure.
Anyway, with the second protractor, or other means of (mechanical) whole-turns counting, you wouldn't have your posited problem.
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>

If the power fails the machine stops anyway so no further machining. I think the errors on start up again must be very small compared to the biggest error element at least in my case, the human error. I think Tony is right on saying the "electronics took the problem out of the equation" But I don' want to be divisive on this issue Alan
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
On Wed, 5 Nov 2008 08:34:27 -0000, "Amateur Machinist"

Well come to that, you could have a thrombie while you were winding the handle & then all bets are off ;-)
I can't remember the last time I had a power failure while I was in the workshop. If it's a worry, there is always the UPS option.
Regards, Tony
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>

Not so much an error in the analysis, more an error in the execution. Having now got the required 200-step protractor (plus a brass pointer) to indicate the rotation of the stepper motor, I resolved to use it to mark up the periphery of the faceplate using some indelible markers (after degreasing it from several years' accumulated grime.)
All went well until I arrived at the last division, numbered 79 (counting from 0) only to find that there was one blank gap remaining.
Cue ... Victor Meldrew's, "I don't believe it!" ... I'd managed to miss out the number 14 despite being prompted by the previously written "13".
Then I found that the ink really was indelible and resisted WD40, paraffin and white spirit!
It was now quicker to remark the first 15 than to correct the remaining 65.
So. I now have a neatly marked protractor scale on the periphery of the face plate, with the exception of the first 15 spaces which are messily corrected
Perhaps, after all, there was something in Tony Jeffrees' preference for driving the thing round electronically!
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Amateur Machinist wrote:

Try IPA or Hammerite thinners to clean off the messed up bit.
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>

I was thinking India Pale Ale for a while then.
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
1501 wrote:

So was I, drink enough and you don't care about the mistakes:)
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
On Mon, 10 Nov 2008 14:28:56 -0000, "Amateur Machinist"

Well, I would have to agree of course - your experience with messing up marking the numbers was exactly the kind of reason that took me in that direction.
Of course, there are many options for providing electronic control - older generation PCs are available dead cheap these days, and there are a variety of cheap/free software options for doing this stuff. There is a pieve of software available that was aimed primarily at clockmakers - written in Oz I believe - that would fit the bill very well. It was shown at the Harrogate show last year (or possibly the one before) - I have the details somewhere but it may take a while to chase it down.
Regards, Tony
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>

Definitely the last version, now with the enclosing circles having read through the tutorial. Haven't had such an enjoyable day hacking code since the days of the "Oric" home computer!
The centrepop code was removed as being redundant because when I drew the circular arcs, I got an initial radius drawn in and couldn't get rid of it, so used 4 of such radii to be the centre marker.
----------
%! 300 600 translate
/Times-Roman findfont 15 scalefont setfont
/linelength bind def /strokelength bind def /bigstrokelength bind def /bigstrokeaddendum bind def /divisions bind def /multi bind def /str 3 string def
/rotateanddraw { linelength strokelength sub 0 moveto linelength 0 lineto stroke % Is this a multiple of multi dup dup multi div floor multi mul eq { linelength bigstrokelength sub 0 moveto linelength bigstrokeaddendum add 0
lineto stroke linelength 10 add 0 moveto str cvs show } ifelse 360 divisions div rotate } bind def
% Inner circle 0 0 moveto 0 0 linelength strokelength sub 0 90 arc 0 0 moveto 0 0 linelength strokelength sub 90 180 arc 0 0 moveto 0 0 linelength strokelength sub 180 270 arc 0 0 moveto 0 0 linelength strokelength sub 270 360 arc stroke
% Outer circle 0 0 moveto 0 0 linelength 0 90 arc 0 0 moveto 0 0 linelength 90 180 arc 0 0 moveto 0 0 linelength 180 270 arc 0 0 moveto 0 0 linelength 270 360 arc stroke
0 1 divisions 1 sub { rotateanddraw } for
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>

Thanks again for your input. I reverted to that approach with the minor additional edit of having the numbers go round the other way. I have attempted to stabilize the printed paper by covering the whole thing with the sticky-backed-plastic sold for covering books.
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Amateur Machinist wrote:

A note for next time: You can get several types of print-on-able transparent film - overhead projection film is one, but ask your stationer, they may have the type of film used for hand-drawn animation, or even one for pcb photography (not pcb transfer film) - the thicker ones will have better dimensional stability than paper too.
Don't cost much either, unless you run into minimum quantity.
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Amateur Machinist wrote:

A note for next time: You can get several types of print-on-able transparent film - overhead projection film is one, but ask your stationer, they may have the type of film used for hand-drawn animation, or even one for pcb photography (not pcb transfer film) - the thicker ones will have better dimensional stability than paper too.
Don't cost much either, unless you run into minimum quantity.
Another tip is to get it laminated - my local stationer charges 30p - or even laminated twice.
-- Peter Fairbrother