burning question - HSI?

[AndyT]

We need to know the RPM and the smallest step the servo is capable of to determine the correct gear ratios.

[ruprecht]

Given that all 4 major axes (compass card, heading bug, RMI needle, CDI assembly) need to rotate continuously, I don't think servos are viable. Servos have only a limited range of movement, we really need to be looking at stepping motors. Servos are ideal for the CDI needle and the glideslope indicator, with only limited (~ 90 degree) range of movement required.

Even the highest performance aircraft can only turn at about 20 deg/sec. I'm not sure how fast you can get common 48 steps/rev steppers to turn, but at 1/4 degree steps, 20 deg/sec equates to 80 steps/sec or 80Hz. Is that possible? Given that Mike Powell's book (henceforth BSAI for simplicity!) talks about accelerating through 250Hz to 1KHz (pp 192), 80Hz seems like it should work easily.

[AndyT]

Most servos have a small tab that keeps them from 360 movement. Break it off and there you go. But I agree that steppers are likely a better choice.

And as small as this unit is likely to be, I think 1 deg steps will be more than enough resolution.

[cscotthendry]

Whether we use modified servos or steppers we still need to have a home sensor. However, steppers allow you to predict where the driven gear is at any given time. Modified servos work on the priciple that 1.5ms PWM signal centers or stops the servo rotation, A narrower pulse causes the servo to turn one way and a wider pulse the other way. The problem is that you don't know how far the servo has turned unless you couple an additional feedback mechanism such as a rotary encoder or some other type of encoder. Now you're getting really complicated and then steppers look a lot simpler.

If you have a 48 steps/rev stepper and you want .5 degree res, then you need to get 720 positions on the driven element. You just need a combination of gears that gives you 720/48. It works out that a ratio of 15:1 will give you that. This can be accomplished a few ways. You could have a large gear with 180 teeth and a pinion on the stepper that has 12. Or you could have a divide by three followed by a divide by five. This could be done with a compound gear.

The problem with a 180 tooth gear and a 12 tooth gear is the overall case size and the size of the stepper motor. If you position even a small stepper at the radius that a 180 tooth gear will demand, You will find it probably won't fit within the required profile of the instrument. However, we can use multiple gear shafts to offset the stepper. This makes the construction more complex, but only a little.

While you're sketching, have a scope around at steppers on the net. We'll need a source of cheap steppers all the same size. I bought a batch from a company called Futurlec, but they're not terribly prompt suppliers to do a project from. But they are CHEAP. I've seen some criticism of them in some forums, but my experience has been generally that they are slow, but basically honest. BTW, my steppers are 40 steps/rev. and they're an excess stock.

Also, if at all possible, look for units that already have a 48TPI pinion mounted (around 12-14 teeth).

Hope this helps. You guys are doing a great job tossing this around. I'm confident that we can come up with a workable design.

[ruprecht]

While you're sketching, have a scope around at steppers on the net. We'll need a source of cheap steppers all the same size.

Jaycar carry this one: YM-2751 48 steps/rev though I'm sure our american cousins can get steppers cheaper than AUD $19 each!

[ruprecht]

PS I am about to do an order to Serv-o-Link for some gears for the Altimeter project in Mike Powell's book - if any aussies want some gears and want to split the shipping, let me know before thurs night.

[ruprecht]

Just been sketching the gear trains in my "pit book" I think I have it sorted:

For the main ring gears (x 3):
- components: 48 steps/rev stepper, 1 x 12 tooth gear, 1 x 36 tooth gear, 1 x 24 tooth gear, 1 x 120 tooth gear, idler shaft.
- assembly: the idler shaft has a 24 at one end and a 36 at the other. The 12 is mounted on the stepper. the 12 mates to the 36 for a 1:3 ratio, and the 24 (at the other end of the idler shaft) mates to the 120 ring for 1:5 ratio.
- result: 48 steps gives 1/15th rotation, or 1/2 deg/step.

for the CDI:
- components: 48 steps/rev stepper, 1 x 60 tooth gear, 2 x 12 tooth gear, 1 x 36 tooth gear, CDI shaft, idler shaft.
- assembly: a 12 on the stepper, a 60 on the end of the CDI shaft (with the CDI assembly at the other end), idler shaft has a 36 and a 12. Mate the 12 on the stepper to the idler 36 for 1:3, and the idler 12 to the CDI shaft 60 for 1:5.
- result: 48 steps gives 1/15th rotation, or 1/2 deg/step.

Thanks for the help on the gearing, now to see if the gears will fit inside. I feel like I'm on an episode of MythBusters - "Hobbyist HSI - it can't be done!". Hoping to bust that myth!

[AndyT]

I just had a great reason for making layered gear rings and I got interupted and now I lost it. It had something to do with embedding position sensor triggers.

So before I forget, we can use flexible drive shafts for some of the elements. Ever see a Dremel with the Flex cable hooked up? Same principle.

[AndyT]

After spending two days on this, I have a much better understanding why BK charges $15,000 for an HSI. I don't even want to think about what Simkits would charge.

And by the way,
By posting the development process in the forum, we are making the designs that result public property and not avialable for any person to patent or copyright.

[Michael Carter]

Collins are even higher. But they have a couple more functions as well.