PDA

View Full Version : Discussion Co-pilot signal help


Dimitris76
Jul 26, 2008, 04:50 PM
Hi all!

Here is my problem: I am using the thermopile sensor pack from the FMA's Co-pilot system for an OSD artificial horizon.

The firmware - as it is - supports only the + orientation of the sensors (sensor pairs along and perpendicular the fuselage) and NOT the X orientation (that has the sensors "looking" diagonally).

This sensor pack has a 4 pin connector: GND, Vref=3.3V, Pitch, Roll.
The thermopiles are coupled in pairs and their output is a voltage that variates from 0 to 3.3V, proportionally to the heatreading difference on each side. For example when the aircraft is flying level the voltage at the Roll output (V1) is 1.65V, when it's rolling left it's 0-1.65V and when it's rolling right 1.65-3.3V. Same for diving/climbing - the Pitch (V2) voltage variates according to the angle.
All these of course apply if the sensor is like I said mounted in the + mode and the thermopile pairs are looking front-aft, left-right...

After a lot of thinking and since modifying the firmware is not an option I came to the conclusion that an external circuit between the sensor pack and the OSD board needs to be constructed which will modify the voltages according to the following formulas:

V1'= (V1+V2)/2 and V2'=(Vref+V1-V2)/2

where V1' and V2' should be the voltage pairs on the diagonal mode.

And the circuit that I designed is this:

http://i12.photobucket.com/albums/a235/Dimitris76/MAHIopamps-1.jpg

All the resistors are 100k except the voltage divider ones that are 10k.

In the schematic I used the LM741 OpAmplifier ICs that I found in Eagle's library, and since these OpAmps have a maximum output voltage Vcc-1.5 volts I need +5V at the supply rail in order to get 3.3v at full angle deflection.
Mr-Cam suggested that if I use a rail-to-rail OpAmp I could omit this 5V supply and power the amplifier straight from the 3.3V Vref.

So, I searched and found this
http://www.elfa.se/pdf/73/734/07345945.pdf
dual CMOS rail-to-rail OpAmp that I "think" it might be up to the task... What do you guys think? Does the circuit look OK? Do you have another OpAmp or resistor values to suggest?

Dimitris

Dimitris76
Jul 27, 2008, 08:12 AM
Bump

Anyone?

dleroi
Jul 27, 2008, 10:27 AM
Dimitris,

I'm confused. Why wouldn't you simply mount the FMA sensor in the + direction?

- Don

Dimitris76
Jul 27, 2008, 02:45 PM
Don,

I wish I could, by unfortunately my middle wing, undercarriageless airframe does not allow clear horizon view for all 4 thermopiles in + configuration... neither does my heli.

Dimitris

dleroi
Jul 27, 2008, 03:49 PM
Dimitris,

I see. Well, your op amp summing circuit looks fine to me - except the outputs seem to be labeled backwards.

I see no reason why it wouldn't work the same with rail-to-rail op amps. BTW, I think "rail-to-rail" is dependent on load. You may want to check the specs to see how close you are with your voltage divider and whatever else the output is working into.

Another approach to this would be to read the levels with a PIC or such, do your conversion in software, then output to a serial D to A - like an LTC1257.

- Don

Dimitris76
Jul 27, 2008, 05:11 PM
Don,

thank you for the answer!

Since the output is grounded through two 10k resistors the current leak to the ground would be just 3.3/20000= 0.000165 Amps or 165uA.

Then the signals go directly to a PIC according to the schematic here:

http://www.rc-cam.com/imageahi/MAHI_Instr_RevB.pdf

do you have any idea what is the current draw there at the PIC? I am hoping negligible...

Unfortunately writing software for PICs is not my forte, but even if I could do it, the PIC and A/D converter solution would cost much more without any advantages.

Dimitris

Dimitris76
Jul 27, 2008, 06:43 PM
except the outputs seem to be labeled backwards.

Don

What do you mean?

Edit: I got it!
Dimitris

dleroi
Jul 27, 2008, 08:04 PM
do you have any idea what is the current draw there at the PIC? I am hoping negligible...


Dimitris

Dimitris,

Actually, I hadn't paid enough attention to the resistor values in your voltage divider. No worries. At 10K, you're going to get within 30 or 40 mV of the rails, according to the spec sheet in your link. And, you are correct. The PIC input impedance is not significant.

- Don

Dimitris76
Jul 28, 2008, 01:18 AM
I am building the circuit today and I'll let you know how it works.

Dimitris

Dimitris76
Aug 08, 2008, 03:37 PM
For the record, the circuit works flawlessly. Just one difference from the schematic above - 11.1volts (from the 3S Lipo) instead of 5 volts at VCC+.

Here is a short video showing the temporary setup

http://s12.photobucket.com/albums/a235/Dim...08082008002.flv

and here you can see it in action

http://s12.photobucket.com/albums/a235/Dim...08082008004.flv

Dimitris