I am designing a functional replica of the Apollo FDAI (aka 8 ball). See pics Real FDAI Another view

From the Apollo 8 Flight Journal: The FD/AI (Flight Director/Attitude Indicator) or "8-ball" is one of the most important instruments in the spacecraft. Designers had originally intended to give the crew three separate displays to show their attitude; one each for roll, pitch and yaw. Being pilots, the crews quickly threw out the three displays for a development of the artificial horizon familiar from aircraft instrument panels.

Here's what I have so far (I know, those aren't steppers, it's only V1) V1_ball V1_view2

V2 will rely on three stepper motors to position the ball. Previously I put the motors more or less in line with their respective axes. Unfortunately, this introduced the challenges of dealing with rotating electrical connections, balance of the ball, motors turning the weight of other motors... Etc.

So, why not eliminate these issues at the cost of increased mechanical complexity! V2 sketch (please excuse the humanCAD)

My plan is to use gears (and belts potentially) to move the control axes to be colinear with the roll axis. All control axes will be concentric and will have its own shaft (two of which will be hollow). Basically like the three hands on a clock. This will allow all the motors to be outside the moving ball mechanism. The problem with this is it demands more performance of the motors.

For example: if I wish for the ball to pitch, I have to move the yaw motor in tandem with pitch to prevent the yaw bevel gear from moving relative to the pitching platform. Furthermore, if I want to move pitch and yaw at the same rate, the yaw motor must move twice as fast as the pitch motor. This issue is even worse if I want to roll, requiring compensation in both pitch and yaw axes.

I don't expect this to be a problem at low rates, but it severely limits the maximum rate of the ball. The pitch motor must be able to move at least twice as fast as the roll, even worse, the yaw must rotate 3x as fast as the roll motor. I'm not willing to sacrifice the accuracy of the instrument, so simply gearing the motors "faster" is not a solution.

I believe that an elegant mechanical solution exists, but I'm stumped. I think if I had a mechanism that behaves in a "I can turn you, but you can't turn me" manner might go a long way.


2 Answers 2


I’ve been working on and off on making a working navball since 2013 and have gotten the mechanics working.

Here’s the secret to getting it to work.

For rotating electrical connections you use something called a slip ring.

To resolve the angle the ball has turned to you use an angular resolver. There are two types. Incremental and absolute. Incremental will show 0 degrees every time you start it up and will need to be told what angle it’s actually at. The absolute version is tons better because when you start it up it can instantly return the correct angle.

I started out working with steppers but ultimately it ended up too heavy. It’s simpler to use an absolute angle sensor with a small geared motor and use a PID routine to align the ball to the correct angle. It turns them into a 360 degree servo motor... and the only way is to use slip rings like NASA did.

I’m on the big problem of making the actual hemispheres. I think I might be able to do it but it’s harder than I thought.


Have you considered using rollers attached to motors to rotate the ball? Look at old school computer mice with ball wheels. Instead of using sensors to track position, use rollers to drive your sphere. Add a 3rd motor perpendicular to the other 2 to get your 3rd axis of rotation. Nonius encoder with hall effect sensors can provide very accurate absolute angular position. No belts, no gears, no differentials.

Fewer components = fewer possibilities of error.

  • $\begingroup$ Drift is going to be an issue with this. $\endgroup$
    – ikrase
    May 9, 2020 at 3:52
  • $\begingroup$ That can be avoided with minimal preload. $\endgroup$
    – jko
    May 11, 2020 at 12:52
  • $\begingroup$ I don't find that entirely plausible especially with reverses and multiple revolutions. I think you'd need to close the loop. $\endgroup$
    – ikrase
    May 11, 2020 at 13:37
  • $\begingroup$ I don't understand what you mean by "close the loop" $\endgroup$
    – jko
    May 11, 2020 at 13:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.