home *** CD-ROM | disk | FTP | other *** search
-
-
-
-
-
- Louis Ricker
- PO BOX 917
- ITALY TX 76651
- USA
-
-
-
-
-
-
-
-
-
-
-
- SELECTING GEAR TRAINS FROM A SET
-
- OF LATHE CHANGE GEARS
-
- This is an attempt at making gear train calculating more
- productive. Both a non-computer method and a computer method will be
- considered. For the benefit of both methods I have taken a look at
- fundamentals and have jettisoned some lumber that has encumbered
- method and mathematics.
-
- For this undertaking one should have at hand a calculator which
- has the logarithm, reciprocal, and change sign functions. Some
- calculators give only natural (base e) logarithms. Others give both
- natural and common (base 10) logs. Either can be used, but only one
- kind of log should be used throughout a process. The reciprocal of a
- number is one divided by that number. The 1/x key gives the reciprocal
- of the number in the calculator display. The change sign key, +/- ,
- changes the sign of the number in the display (+ to - or - to +). If a
- common log is in the display the 10^x (10 raised to the x power) key
- gives the number whose log is in the display. The e^x key gives the
- anti log of a natural log in the display.
-
- In addition to a calculator access to a computer is desirable. We
- can manage without a computer, but it is not possible to duplicate the
- work of the computer using only a calculator, because the amount of
- work involved is beyond human capacity.
-
- Although the methods discussed here can be applied to gear trains
- in general, we shall be thinking only of the change gear lathe and
- three applications of gear trains - screw cutting, spindle indexing,
- and linear indexing. It should be noted that our methods produce many
- approximate gear trains. The user must decide whether this is
- tolerable.
-
- Gear train solvers have always been plagued by multiple unknowns.
- In finding a three-pair compound train one must solve a problem having
- one known quantity (the ratio of the train) and six unknowns (the
-
-
-
-
-
-
-
-
-
- Louis Ricker, Gear Trains 2
-
-
-
- gears). There are two ways to do this. One way is to examine all the
- possible combinations of a set of gears. My set consists of sixteen
- gears having fourteen sizes. For one, two, or three-pair trains there
- are hundreds of thousands of combinations. This must be done by
- computer. The other way is to doodle. Doodling works, but it is not
- altogether satisfactory. It simply leaves too many solutions
- undiscovered. On the other hand, I have already received a complaint
- about "the vast amount of output" of correct solutions by the
- computer. I have responded by tightening the tolerances in the
- programs. This has reduced the number of trains printed. In some cases
- a tight tolerance will exclude all trains, and there will be no trains
- printed. Then the tolerance must be widened.
-
- In order to control the doodling some methods apply many rules.
- Others add known factors to the formulas. This tampering only serves
- to restrict the number of possible solutions. Often failure to
- properly label values has led to logical leaps. It is one thing to do
- arithmetic correctly, but quite another to give mathematics meaning.
- To do this we must be careful with language.
-
- Consider the following problem: "How many millimeters are
- contained in 0.125 inch ?" The answer to this problem is "how many",
- not "how many millimeters"; we already know it is millimeters. Now let
- us state the problem mathematically:
-
- Let x represent "how many".
-
- x mm = 0.125 in.
-
- If a quantity is divided by its equal the quotient is one:
-
- x mm
- --------- = 1 = the number of times 0.125 in. is contained in x mm .
- 0.125 in.
-
- By law: 1 in. = 25.4 mm .
- This is called a "conversion factor". It is not a factor.
- One divided by 25.4 does not equal one, but 1 in. divided by 25.4
- mm does.
-
- 1 in. 25.4 mm
- ------- = 1 = -------
- 25.4 mm 1 in.
-
- x mm 1 in. 25.4 mm
- --------- = 1 = ------- = -------
- 0.125 in. 25.4 mm 1 in.
-
- We want x or "how many"-1alone-0.
-
- 0.125 in. x mm 25.4 mm 0.125 in.
- --------- X --------- = ------- X ---------
- 1 0.125 in. 1 in. 1
-
-
-
-
-
-
-
-
-
- Louis Ricker, Gear Trains 3
-
-
-
-
- (If equals are multiplied by equals the results are equal.)
-
- 1 x mm 25.4 mm 0.125 1
- ---- X ---- = ------- X ----- X ----
- 1 mm 1 1 1 1 mm
-
- All labels cancel.
-
- x = 3.175 = "how many".
-
- Another "conversion factor" is Pi or 3.1416 .
- Properly labeled it becomes:
-
- 1 circumference = 3.1416 diameters.
-
- 3.1416 diameters
- or ---------------- = 1
- 1 circumference
-
- Pi and 25.4 are often injected into gear train formulas in the
- form of gear pairs. The arithmetic is correct, but many other pairs
- are excluded by this practice. For our purposes conversions should be
- done up front once and for all, not hundreds of thousands of times.
-
- In this article the gear on the spindle is a driveR ; the gear on
- the lead screw is driveN.
-
- Let "DRIVERS" equal the product of the numbers of teeth of the
- driver gears. Let "DRIVEN" equal the product of the numbers of teeth
- of the driven gears.
-
- Formula 1 is for gear trains between spindle and lead screw:
-
- Lead of thread to be cut DRIVERS
- RATIO = ------------------------ = -------
- Lead of leadscrew DRIVEN
-
- Formula 2 is for spindle indexing gear trains.
-
- Let DIV = Number of divisions of one spindle rotation.
-
- Let INDEX = Number of teeth indexed.
-
- Formula 2 :
-
- DRIVERS DRIVERS
- RATIO = DIV = ------- = --------------
- DRIVEN INDEX X DRIVEN
-
- DRIVERS
- INDEX = ------------
- DIV X DRIVEN
-
-
-
-
-
-
-
-
-
- Louis Ricker, Gear Trains 4
-
-
-
-
- Notice that the number of teeth indexed is equivalent to a gear
- of that number of teeth making one revolution.
-
- Formula 3 is for linear indexing of the leadscrew.
-
- Required advance of carriage by leadscrew
- RATIO = ----------------------------------------
- Lead of leadscrew
-
- DRIVERS INDEX X DRIVERS
- RATIO = ------- = ---------------
- DRIVEN DRIVEN
-
- Formula 3 :
-
- RATIO X DRIVEN
- INDEX = --------------
- DRIVERS
-
- The ratio is the number of turns of the leadscrew per index. The
- number of turns can be fractional. Notice that INDEX is a driver in
- Formula 3 while INDEX was driven in Formula 2 .
-
- THE NON-COMPUTER METHOD
-
- The chart on my Myford 7 lathe calls for the following gear train
- for a 2.25 mm thread lead:
-
- 45 X 60 X 21 DRIVERS
- ------------ = ------- = ACTUAL RATIO = 0.70875
- 40 X 40 X 50 DRIVEN
-
- My lead screw has a lead of 0.125 in.
-
- LEAD OF THREAD TO BE CUT 2.25 mm
- DESIRED RATIO = ------------------------ = --------
- LEAD OF LEAD SCREW 3.175 mm
-
- DESIRED RATIO = 0.708661417
-
- COMMON LOG OF DESIRED RATIO = -0.149561211
-
- How many turns per inch on desired thread ?
-
- 25.4 mm / 1 in. 1 turn X 25.4 mm 25.4 turns
- --------------- = ---------------- = ----------
- 2.25 mm / turn 1 in. X 2.25 mm 2.25 in.
-
- TURNS PER INCH OF THREAD = 11.28888889
-
- The inverse (1/x) of this is 0.088582677 in./turn.
-
-
-
-
-
-
-
-
-
-
- Louis Ricker, Gear Trains 5
-
-
-
- With the gear train one turn of the spindle produces 0.70875 turn
- of the lead screw. This times 0.125 in. lead is 0.08859375 in./spindle
- turn. At 11.28888889 turns we get 1.000125 inches. Who could ask for
- better ? But I have only one 40-tooth gear.
-
- Let P = NUMBER OF PAIRS. Let N = NUMBER OF SIZES.
-
- N times (N-1)
- Formula for NUMBER OF PAIRS: P = ------------- .
- 2
-
- I made a list of all the pairs (91) of my gears and divided the
- larger gear by the smaller. All the ratios were greater than one. The
- logs of these ratios are all positive. This avoids confusion when
- subtracting logs. I set all the logs in a table in ascending order,
- along with their corresponding ratios and pairs. Then I began to
- doodle, adding and subtracting logs, trying to get them to equal the
- log of the desired ratio. This was like looking for a needle in a
- haystack. However, there were many "needles", and after a few tries I
- found this train :
-
- DRIVERS 25 X 70
- ------- = ------- = 0.708502024 RATIO
- DRIVEN 65 X 38
-
- This gives 0.999775078 in./11.28888889 turns. But I did not find it
- directly. First I subtracted the log of 70/38 from the log of 65/25.
- (0.41497 - 0.26531 = 0.14966). Since I subtracted a log its pair had
- to be inverted.
-
- 65 X 38
- ------- = 1.41143.
- 25 X 70
-
- But this is the wrong ratio. The inverse of this ratio is 0.7085. If
- the ratio is inverted the entire train is inverted :
-
- 25 X 70
- ------- = 0.7085
- 65 X 38
-
- How did I know I had the right gears if I had the wrong ratio ? This
- is the beauty of logarithms. The log of 1.4114 is 0.14965. The log of
- 0.7085 is -0.14965. The calculator keys, 1/x and +/- , make these
- calculations relatively pleasant.
-
- Gear trains may be rearranged without changing the ratio provided
- the drivers remain drivers and the driven remain driven. Two-pair
- trains can be arranged four ways, and three-pair trains can be
- arranged thirty six ways without changing ratio.
-
- The computer revealed dozens of trains which had ratios closer
- than the ratio of the train I found by doodling. If you need to
-
-
-
-
-
-
-
-
-
- Louis Ricker, Gear Trains 6
-
-
-
- compensate for an error in the lead screw you can use the computer to
- sift out trains having that error simply by varying the tolerance in
- the program. To sift out such a train by doodling requires more luck
- than I have ever had.
-
- Doodling for indexing gears is essentially the same except that
- not all the logs are in the table. We have to take each gear size and
- pair it with DIV to get Log (DRIVER divided by DIV) or with RATIO to
- get LOG (RATIO X DRIVEN). Then we search the tables for another pair
- or two. We will get an INDEX for every train, but obviously "INDEX
- 57.5 teeth on the 30-tooth wheel" is useless. After trying this
- doodling I was convinced that I needed a computer. I did not know how
- to program it, so I had to learn as I went along with this article.
- The programs included here are somewhat better than the first ones,
- but I know they can be improved.
-
- THE COMPUTER METHOD
-
- The program was written for the Commodore 64 computer. In order
- to translate the program for your computer you need to:
-
- 1. understand what is going on in my program overall.
-
- 2. know what each instruction to the computer is in my program
- and what it does.
-
- 3. know or be able to find the corresponding instructions for
- your computer.
-
- The computer takes logs from pigeon holes and assembles them to
- represent prospective gear trains and tests them. If a train is good
- enough it is printed on the screen or - if you use a printer - on
- paper. For some trains the computer will tell you to invert the last
- pair in relation to the other pairs - and/or to invert the entire
- train.
-
- Earlier in my article I described two methods of solving gear
- train problems. Here is another method. First arrange the pairs of
- gears and the corresponding logarithms of their ratios in order. I
- chose ascending order just as with the doodling method. Incidentally,
- I believe L. C. Mason was the first to call it doodling. Second select
- in turn two pairs. Third conduct a systematic search for a third pair.
- And fourth and most important, let the computer do it all.
- # # #
-
- Postscript
-
- This essay and the program were written many years ago. Since
- then my brain has become addled as a result of my addiction to chess.
- (Lads and lassies, don't go near chess; at least don't inhale). I
- stand by my masterpiece essay, but the program... Obviously 'fine
- tuning' is needed. A routine to sift out the optimum train of gears
- would help. If you have only one gear of a particular size and the
-
-
-
-
-
-
-
-
-
- Louis Ricker, Gear Trains 7
-
-
-
- computer spits out a train having two gears of that size, that train
- should be eliminated. Likewise trains that are not mechanically viable
- and duplicate trains. If my memory serves I did not do all the
- possible inversions of gear pairs in "gear trains", but I believe the
- inversions were done in "my gears", which was created by "gear
- trains". (See lines 1080 and 1085.) I'm too old to figure this stuff,
- and the search algorithm - the moment I finished programming it I
- could no longer understand it. THAT should make the case for extensive
- REMs. Have fun.
- LR
-