Real Altimeter to flightsim conversion

[_alioth_]

As I said in other threads Im working in full panel to use with prepar3d A2A Comanche.

I have a simulator altimeter (from an old ast300) easy to interface.
I made some test in this video:
https://www.youtube.com/watch?v=QPLSP0p9l8o

The problem is it only goes to 10.000ft.
You dont usually fly higher than 10000 in a piper comanche, but you can do it.

I was looking for a servoed altimeter because It would be much easy to interface, but I haven´t found anyone that matches a Comanche cockpit.

I have purchased a real bendix encoder altimeter to trying to make a conversion.

http://www.mycockpit.org/forums/attachment.php?attachmentid=12275&stc=1


This bendix is a pressure altimeter. 0-35.000ft (enough for a comanche :mrgreen:)
It has encoder, so, it sends the altitude data to the transponder.
This is a good thing, because I can use this wires to connect it.
The baro knob rotate the whole inner mechanism and those wires exit through a spiral.

You can interface in some ways:

-Pressure pump. You need very precise pump and sensors. And a microcontroller with at least 12bits resolution.
35.000 / 4096 (12 bits) = 9 feet error in a perfect world. It is good enough. My arduino mega only has 10bits:
35.000 / 1024 = 34 feet. Mmm... not good enough.

-Stepper motor and hall sensors.


-Make the same mechanism the servoed altimeter usually have. Dc motor/ stepper motor, and two resolvers, one for coarse tune, and one for fine tunning.

I have choosen the third path, with a multipot (10 turns) to coarse tunning, and an absolute encoder to fine tunning.
So, with pot I move the needle near the actual altitude (+-200 feet), and then I use the absolute encoder to match my needle angle with the altitude needle angle. The precision in this way depends on the encoder. Im using 1024 positions per turn. So the theorical precision is 1000 feet / 1024 steps = 0,98 ft. Super precision!! This is theory...

lets go!

some 3d print to make the gears mechanism:


http://www.mycockpit.org/forums/attachment.php?attachmentid=12276&stc=1



10 turn pot, dc motor, and absolute encoder:

http://www.mycockpit.org/forums/attachment.php?attachmentid=12277&stc=1

http://www.mycockpit.org/forums/attachment.php?attachmentid=12278&stc=1



To be continued...

[_alioth_]

The gears connected to the original axis.
1 turn of the dc motor (gearboxed) -> 1 turn encoder -> 1 turn needle -> 0,29 turns pot
So.. 34 turns of the needle (34.000 feet in the range from -1.000 to 33.000) -> 10 turns potentiometer

http://www.mycockpit.org/forums/attachment.php?attachmentid=12279&stc=1


More 3d printing...

http://www.mycockpit.org/forums/attachment.php?attachmentid=12280&stc=1



Spiral wires. So It can be use the mechanical baro knob system, which turns the whole structure.

http://www.mycockpit.org/forums/attachment.php?attachmentid=12281&stc=1


The original aneroids to the right. I have taken the exactly same space, so I can use the original altimeter case.

http://www.mycockpit.org/forums/attachment.php?attachmentid=12282&stc=1


To be continued...

[FromUndaChz]

Excellent work! Looking forward to seeing more

[_alioth_]

FIRST TEST.
Programming ONLY the potentiometer without smooth readings.
You can see the needle jumping, because no smooth readings or smooth inputs to dc motor.
Each unit of the pot reading change 12 degrees the needle. And it is normal reading analogues potentiometer values to go +- 2 or 3 units. So it is normal to expect jumps from 30-40 dregrees around the target altitude.


https://www.youtube.com/watch?v=9Yo_5zAPbUE&feature=youtu.be

[_alioth_]

SECOND TEST.
Potentiometer + encoder. No smoothing.
There is a total range of 34 turns x 360 = 12.240 degrees.
I use the pot to go near the target angle (less than 120 degrees).
When I am 120 degrees near the final needle position, I use the absolute encoder to drive the needle.
The encoder is 1024 unit por turn. So I can drive the needle with a precision of 1000 feet / 1024 steps = 0,98 feet.
There is no smothing in readings either pot or encoder. So, there are jumps again. But because of the high resolution of the encoder, jumps are much smaller than first test.



https://www.youtube.com/watch?v=uZomZDIAitw&feature=youtu.be

[_alioth_]

THIRD TEST
Pot and encoder. Smoothing results.
With a little programming:
-Smooth pot reading
-Smooth encoder reading
-Software PWM to drive the dc motor so needle move smooth to the target.



https://www.youtube.com/watch?v=DIf1hzhhvbQ&feature=youtu.be


Now it is working smooth as silk in the full range from -1.000 to 33.000 feet

[_alioth_]

Finally, the baro knob. It has a mechanical calculator inside
So, I only send the standard altitude pressure from the sim. And it is trimmed to display this altitude when 29.92 in baro window.
Later, the correction is done mechanically
Simple in the programming side, complex in the mechanical side.


https://www.youtube.com/watch?v=QMvqYQhcnC0&feature=youtu.be

And you don't need to send the kollsman value to the simulator. Kollsman windows are not synchronized. But both simulator and real altimeter show the correct altitude with their kollsman value. Perfect!
I could add a pot and send the kollsman value to the simulator, but.. for what? It is not necesary.


Thanks for reading!
Arturo.

[iwik]

Well done and thanks for showing.
Regards
les