[MOC] 4-Digit Mechanical Calculator "TECHNIVAC II"

[2GodBDGlory]

After building my previous mechanical calculator, which featured two sets of buttons and an adder mechanism to add two single-digit numbers, I wanted to go beyond that and create a multi-digit calculator. I eventually settled on a four-digit setup, and came up with this model here.

Before I actually show it, though, I should mention that this model was functionally a failure, because it would not add or subtract with tolerable accuracy (The accuracy was about +/- 2 in each decimal place, so if the real answer was 5426, the calculator might tell you 3204, which is totally unacceptable!, and that's assuming that all the carrying was correct!) Despite that, I think it is at the least a good theoretical framework of how such a calculator can be constructed, and a good first attempt. I'm already starting to think of how accuracy could be increased, so maybe I'll have something to show in several months.

Anyways, I'll now endeavor to show this machine, though my YouTube video at the bottom may end up being clearer.

Overview images:

Spoiler

 

So, the basic working principle is that when a number button is pushed, first a linkage moves to engage a transmission driving ring on one of the four digit axles

which causes one of these dials to rotate until it strikes the axle pushed down by the particular button, allowing a button press to roughly choose 36 degree divisions of rotation.

Then, the axle transmits drive through this freewheel

which allows the main dial to be reset without affecting the output dials.

Next, the drive moves through this carrying mechanism, which is very similar to a design Nico71 used in his mechanical calculator

This is known as a sautoir mechanism, and allows for gravitational, or in my case, rubber band, energy to be gradually stored as an axle rotates, and then released all of a sudden. This is used so that as, say, the units dial rotates from zero to nine energy is gradually stored so that when the dial moves from the nine position onward to the zero position again, energy can be released to move the tens digit forward 1/10th of a rotation, effectively doing a carry from 09 to 10. This can work all the way down the line, such as in 0999+0111=1000.

Each axle continues past the carrying mechanism to the forward/reverse gearboxes, which are simple, low backlash sliding gear designs, though transmission driving ring parts are used simply to slide axles forward and backwards.

After this, drive is taken to these white dials.

Now, in order to add numbers, it is not merely sufficient to press, say, the one button and then the two button in a single decimal place and expect it to result in a three. Rather, the axle would just continue rotating from the one to the two position without any adding. To counter this, there is a resetting mechanism that must be employed between inputting numbers. This is engaged by reversing the main drive motors, which can be done in a few ways, as I will discuss later. This causes a worm gear to slide along an axle in a direction-sensitive gearbox, activating a long axle (Sometimes the 24T gear needed a bump to engage it)

When the axle rotated, it sequentially pressed down the levers activating the transmission driving rings for each number, which caused the axles to rotate backwards, one at a time, until this stop

was activated and the friction clutch slipped. This had to be engaged separately to prevent the reset mechanism from stopping once the first dial reset and slipped the shared clutch. Because of the freewheel mechanism discussed earlier, resetting the input mechanism did not affect the output mechanism.

The control mechanism was based on a PF switch (Actually a knock-off one, because all my Lego ones range from finicky to broken, ironically) and had several buttons, which were ON, OFF, AC, +, and -. ON just turned on the drive motor, which then stayed running until another button was pressed. OFF pressed the switch back to neutral, and held it there with friction. AC reversed the drive motor, and while + and - also reversed it, these buttons also flipped the forward/reverse gearbox in the calculator. A typical control sequence might look like ON, 5678, -, 1234, AC, OFF, with the minus being necessary for both resetting and ensuring that the gearbox is in the right position.

Spoiler

OFF

ON

AC

-

+

 

Another mechanism was designed to sense when a negative answer was reached. Originally I had planned to make this automatic, but in the end it requires close observation, followed by a flip of a lever.

The theory was that, provided that the 1000s, 100s, and 10s digits were zero, any time the 1s digit went from 0 to 9 backwards the number would be negative, while if the 1000s and 100s were zero, any time the 10s went backwards from 0 to 9 the number would be negative, while the 1s would make no difference, and so on. In the end, there were four levers with orange parts on the end that could be made visible through slits. Each digit above the ones had cams on a shaft to bump the levers to orange when the digit was at zero, and all four had ratchet-type mechanisms to bump the levers extra high, but only when travelling backwards from 0 to 9. Additionally, each lever was set up so that lifting it would lift all of the levers below it. If it were ever observed that all four slits showed orange at once, this would show that the output had become negative, and a red lever would need to be flipped, revealing a different number on each dial, which was the negative complement. For example, if the number started as 0003, and 0004 was subtracted from it, the dial would originally read 0009, which is obviously incorrect. Once the dial was flipped, though, the negative complement of 0009, namely -0001 would be revealed, which turns out to be the correct answer.

Spoiler

 

 

Finally, I should discuss the procedure to reset the calculator after use. First, the AC button had to be pressed to reset the input half, but then the tan 20T gears visible here

would have to be manually turned to reset the output dials. The big issue was that whenever the negative button was pressed the dials would get out of sync, forcing the user to flip the calculator over and disengage some 20T gears by sliding them along their axles, prior to recalibrating the zeros on the dials with the zero positions on the carrying mechanism. This is something that was quite disappointing to me, that I hope I can fix on any future iterations.

Here is the video, which may be more easy to understand!

My images can be found here: https://bricksafe.com/pages/2GodBDGlory/whirlwind-rescue (Yes, I put them in the wrong folder by accident. Ignore the half-built helicopters and you should be fine)

Overall, this was a very interesting mechanical exercise, but was far from a working calculator, to my regret.

[Scoar Sonander]

(speechless)

[aeh5040]

Brilliant!  Do keep working on it.  I'm certain you will overcome the quirks and hiccups...

[whitepen]