Canon SX260HS accurate geo-tagging

lanzo

28

Re: Canon SX260HS accurate geo-tagging

« Reply #10 on: 10 / January / 2017, 17:47:29 »

Advertisements

I made these 2 adaptor for the Canon S110, the simple one uses a Phototransistor, the other one uses a Photodiode and 2 BC547 input 5V output 3,3V
They work with Reach RTK but doesn't work with Pixhawk because of a bug or a defective hardware I think, I had no response in the APM forums

circuito-fotodiodo.jpg (128.9 kB, 1000x1260 - viewed 61 times.)

15992116_10212064164905185_1096750264_o.jpg (82.48 kB, 960x1280 - viewed 66 times.)

Logged

gio_kap

12

Re: Canon SX260HS accurate geo-tagging

« Reply #11 on: 13 / January / 2017, 02:54:41 »

Hi. Just an update on the latest test/configuration results.
Based also on Lanzo experience, I've modified the circuit adding a transistor amplifier (the phototransistor emitter goes to the transistor base via a 1 kOhm resistor; also a 1 kOhm pull-up resistor is used on the transistor emitter output; additionally, to rule out any 'noise' that could be introduced by the voltage regulator, I used a 3.3 Vdc battery directly).
After several tests using Waterwingz LED test script, now things seem to work consistently with a pulse width of 500 ms; with the standard KAP_UAV script it does not work well 'cause LED pulse width is not constant, and most of the times is too short.
So we're back to the question of how hard it is to implement this LED 'profiling' (constant pulse width of 500 ms) - even though my skills are limited in a lot of ares, I'm willing to support anyway I can.
Thanks
Regards

Logged

waterwingz

12138

Canon SX260HS accurate geo-tagging

« Reply #12 on: 13 / January / 2017, 09:53:26 »

Quote from: gio_kap on 13 / January / 2017, 02:54:41

After several tests using Waterwingz LED test script, now things seem to work consistently with a pulse width of 500 ms;

That's a little disconcerting. I struggle to understand how it could be so slow. It also makes me wonder how well the GPS signal will be synced with that much delay?

Quote

So we're back to the question of how hard it is to implement this LED 'profiling' (constant pulse width of 500 ms) - even though my skills are limited in a lot of ares,

I'll take a look. As a "quid quo pro", I've got an updated version of the script converted to take advantage of some new features in the latest CHDK stable release. I'll add the timing option to that and let you "beta" test for me. The risk here is very low but it does need some testing time.

« Last Edit: 13 / January / 2017, 14:15:03 by waterwingz »

Logged

Ported : A1200 SD940 G10 Powershot N G16

gio_kap

12

Re: Canon SX260HS accurate geo-tagging

« Reply #13 on: 14 / January / 2017, 11:00:30 »

Disconcerting, agree, and so frustrating: I can't believe the time spent on this so seemingly trivial phototransistor circuit.
Anyhow, with 500 ms pulse-width things appear to go as necessary: the phototransistor LED detection is essentially instantaneous, and the interrupt pin detects the change of state immediately (well below 1 V), so I do not foresee issues with the GPS time mark sync (the GPS time-mark accuracy is less than 50 ns, and a rough estimate of the time lapsing between LED turning ON and the logged time-stamp is < 1 ms - I might be wrong, and am planning on doing some test with a scope and an Arduino rig....hopefully I'll have the time to do this...).
If weather permits, I'll be soon doing some mapping tests to start checking how this system works and ortho-mosaics accuracy, so I'll be glad and honored to beta test the updated script.
Thanks a lot.

Logged

waterwingz

12138

Re: Canon SX260HS accurate geo-tagging

« Reply #14 on: 14 / January / 2017, 11:38:54 »

Quote from: gio_kap on 14 / January / 2017, 11:00:30

the phototransistor LED detection is essentially instantaneous, and the interrupt pin detects the change of state immediately

Okay - so here's where I'm confused. If the phototransistor & interrupt LED detection takes place essentially instantaneously, why does the LED need to stay on for 500 mSec? Once the phototransistor has changed state and the interrupt pin seen it, it should be safe to turn the LED off. I don't understand why it needs to be on for an additional 499 mSec?

Unless the interrupt pin is not really a hardware interrupt? If it's just polled in software and if the controller gets busy it might need the LED on longer to assure it catches it every time. But if that's the case, accuracy will go way down when that happens?

In any case, the script changes are complete. Once I've done a little regression testing I'll PM you with a private download link.

Logged

Ported : A1200 SD940 G10 Powershot N G16

lanzo

28

Re: Canon SX260HS accurate geo-tagging

« Reply #15 on: 14 / January / 2017, 13:13:01 »

What's the hardware you use attached to the led and how it works?
I have a problem with pixhawk, there is no documentation and maybe you can help me, thank you

Logged

waterwingz

12138

Re: Canon SX260HS accurate geo-tagging

« Reply #16 on: 14 / January / 2017, 13:34:59 »

Quote from: lanzo on 14 / January / 2017, 13:13:01

What's the hardware you use attached to the led and how it works?
I have a problem with pixhawk, there is no documentation and maybe you can help me, thank you

Tell us more about your problem?

Logged

Ported : A1200 SD940 G10 Powershot N G16

lanzo

28

Re: Canon SX260HS accurate geo-tagging

« Reply #17 on: 15 / January / 2017, 17:32:06 »

This is my post on ardupilot forum:
I have a plane with Pixhawk where I use the shutter throught a circuit with 2 transistor, it works perfectly, now I want to use the CAM_FEEDBACK_PIN feature with a cable I made for the Canon S110 with a photoresistor that input 5V from PIN 102 and output 0V normally and 5V when the camera get picture hooked to an AUX channel set in the software by CAM_FEEDBACK_PIN.
The cable works very well if I test with external 5V supply but when hooked to the pixhawk it doesn't, so I have attached a voltmeter to the cable and discovered that when i set CAM_FEEDBACK_PIN 5x (tested 51 to 54) and test if it works clicking on "Do Action" "trigger camera now", the pixhawk immediately outputs 3.3V to the PIN I set with CAM_FEEDBACK_PIN, the 3.3V stay there until I remove battery, connect battery again and it goues to 0 Volts, click again on "Do Action" "trigger camera now" and goues to 3.3V.
I hope I have explained well myself.

CAM_DURATION,10
CAM_FEEDBACK_PIN,53
CAM_FEEDBACK_POL,1
CAM_MAX_ROLL,0
CAM_MIN_INTERVAL,0
CAM_RELAY_ON,1
CAM_SERVO_OFF,1100
CAM_SERVO_ON,1300
CAM_TRIGG_DIST,0
CAM_TRIGG_TYPE,1

RELAY_DEFAULT,0
RELAY_PIN,50
RELAY_PIN2,-1
RELAY_PIN3,-1
RELAY_PIN4,-1

I have tried to set RELAY_PIN2,53 and CAM_TRIGG_TYPE, 0 but same problem

Logged

waterwingz

12138

Re: Canon SX260HS accurate geo-tagging

« Reply #18 on: 15 / January / 2017, 17:37:01 »

Quote from: lanzo on 15 / January / 2017, 17:32:06

"Do Action" "trigger camera now", the pixhawk immediately outputs 3.3V to the PIN I set with CAM_FEEDBACK_PIN, the 3.3V stay there until I remove battery, connect battery again and it goues to 0 Volts, click again on "Do Action" "trigger camera now" and goes to 3.3V.

So is the problem that the pixhawk is sourcing 3.3V but you were expecting 5V?

Logged

Ported : A1200 SD940 G10 Powershot N G16

lanzo

28

Re: Canon SX260HS accurate geo-tagging

« Reply #19 on: 15 / January / 2017, 18:37:29 »

Quote from: waterwingz on 15 / January / 2017, 17:37:01

Quote from: lanzo on 15 / January / 2017, 17:32:06
"Do Action" "trigger camera now", the pixhawk immediately outputs 3.3V to the PIN I set with CAM_FEEDBACK_PIN, the 3.3V stay there until I remove battery, connect battery again and it goues to 0 Volts, click again on "Do Action" "trigger camera now" and goes to 3.3V.
So is the problem that the pixhawk is sourcing 3.3V but you were expecting 5V?

I have 3 pins Ground +5V AUX, 5V goes to circuit and return 5V or 0V to the AUX pin, now the AUX pin begin to outpur 3.3V when (I think) it must stay at zero V

Logged