One off Galloping Ghost system, for Pulstar

Waggly tails
Martin
Posts: 744
Joined: 16 Feb 2018, 14:11
Location: Warwickshire

Re: One off Galloping Ghost system, for Pulstar

Post by Martin »

belli wrote: 09 Jun 2018, 23:39 Hi Martin,

I have just connected mine up and have a few questions:
I am not sure if my Arduino Pro Mini is 8MHz or 16MHz, does it make a difference? I'm assuming mine is 16MHZ as it is a 5V unit.
I tried at first with what looks like a Pro Mini but has a USB connection, it identifies as a Leonardo according to the IDE, the sketch won't compile, it this a pin conflict issue? A nano compiled and seemed to run fine.
Is there a way to 'hard code' the frequency? I have an Adams actuator which oscillates very nicely but seems happier at the higher frequency offered by 'full up elevator'. It would be nice to have the option to set the frequency in the software for a 'single channel' only option. I guess the PWM could be hard coded too as per your suggestion above.

Thank you for a great project! I am busy with an airframe, probably a Sniffer as I have just always like the cute stubby nose... I'll probably post the laser files here if anybody else is interested.

Cheers,
The board that looks like a Pro Mini but with USB is the Micro, or Pro Micro. It uses a slightly different chip, the 32U4 instead of the 328. I think the correct board to choose for it in the IDE is "Arduino/Genuino Micro" but the Leonardo is very similar. The pulser sketch won't compile for this chip without modifications as the 32U4 handles pin change interrupts in a different way: you will get an error message saying that PCMSK2 is not declared in this scope.

The sketch will work on either the 8MHz or 16 MHz versions of any 328 board (Uno, Nano, Pro Mini, ...) If you're not sure which board you have you can find out by means of the Blink sketch. The standard Blink puts the LED on for a second than off for a second so you should see 30 blink cycles each minute. If you've chosen the wrong 16 MHz / 8 MHz choice for your board and upload using a programmer, then the sketch will still run but will give you either 15 or 60 blink cycles in a minute instead of the correct 30.

You don't have to worry too much about the 5V / 3.3V thing when working with Pro Micros unless you have connected peripherals to the Arduino that won't tolerate 5V. The chips on both types of board are exactly the same and both types will work fine at 5V. The only differences between a 5V board and a 3.3V one are the voltage regulator fitted that regulates the incoming 'Raw' or 'V in' voltage down to 5V or 3.3V and the crystal - either an 8 MHz or 16 MHz one. The reason for the speed change is that the makers of the chip say that running it at the higher speed requires the higher voltage for reliable operation - so if you fit a 16 MHz crystal but only work at 3.3V then you are operating outside the specification and the chip will be less reliable. Despite this, I've got some Chinese clone boards with 3.3V regulators but 16 MHz crystals and they actually work fine - I've never had a problem with them. Maybe they would crash when pushing some of the other specified limits of the chip such as operating temperature or electrical load on the pins.

You can 'hard code' the pulsing frequency of the rudder instead of using the elevator input by changing the line in the main pulser.ino file:

Alter the line:

uint16_t cycle = receiver::channel(1); // elevator input on D6: nominally 1000 - 2000 = 0 - 100%

so it's just:

uint16_t cycle = 1000;

That will give very fast pulsing (about 12Hz), 2000 will be quite slow (about 2 Hz). You can experiment by choosing values anywhere in the range 1000 to 2000 to find the one that suits your actuator best.

Similarly you can alter the line:

uint16_t power = receiver::channel(2);

to:

uint16_t power = 1400;

To set the operating power - low numbers below about 1125 will stop the outputs working completely. 1500 means the outputs are powered about 50% of the time and values of about 1800 up to 2000 will make the output be powered almost 100% of the time. Again, experiment and choose a low-ish number that gives reliable operation - the lower the number you choose the less power drain from your battery (although the power drain with the motor I used is never very high).
belli
Posts: 39
Joined: 19 Feb 2018, 19:19

Re: One off Galloping Ghost system, for Pulstar

Post by belli »

Thanks Martin, I am sure that will get me sorted.
Martin
Posts: 744
Joined: 16 Feb 2018, 14:11
Location: Warwickshire

Re: One off Galloping Ghost system, for Pulstar

Post by Martin »

Sorry about the delay for those who have shown an interest in getting ready-built systems from me. Most of the bits arrived ages ago, but I'm still waiting for two shipments to arrive from China. I was beginning to think that both would never arrive and I'd have to reorder.

I checked again today and the most important shipment, ten off 3.3V 8 MHz Pro Mini Arduinos, have shown some movement on their tracking information in the last week, so I'm hopeful they could arrive any day soon.

The other item, just some heatshrink tubing, hasn't shown any movement in over five weeks - I don't think the seller ever posted it as the last (and only) tracking info from June 6th just says 'Tracking number provided'. I can apply for a refund next Friday if it's still not arrived. Anyway, I can easily get some other tubing quickly enough.

So I've not forgotten you, and I hope to P.M. you with news of progress soon.
jmendoza
Posts: 167
Joined: 18 Feb 2018, 23:07

Re: One off Galloping Ghost system, for Pulstar

Post by jmendoza »

Hi Martin,
I share your pain with the items from China.
It takes anywhere from 4 to 8 weeks, depending, in my experience.
The modules from HK take even longer.
:D
Jay
User avatar
tiptipflyer
Posts: 393
Joined: 16 Feb 2018, 22:49
Location: Germany

Re: One off Galloping Ghost system, for Pulstar

Post by tiptipflyer »

Just finished my GG pulser according to Martin´s project. I omitted the aux channel and used
uint16_t power = 1250
for good results with a simple transmitter on rudder and elevator only.
Also I had to omit the part
|| !receiver::channel(2)
from the line
// wait for valid receiver signal on the rudder, elevator and auxiliary channels
from the sketch to make it pulse without the aux input.
It works great now and is ready to be installed in a model and flown.
Thanks Martin for another great project.
SAM_3485.JPG
Frank
Martin
Posts: 744
Joined: 16 Feb 2018, 14:11
Location: Warwickshire

Re: One off Galloping Ghost system, for Pulstar

Post by Martin »

Well done Frank! Thanks for explaining how to fix the code to work without the 'aux' channel for those who only have two channels available or don't want the extra complication.

Looking forward to seeing how you get on with flying the system!
User avatar
tiptipflyer
Posts: 393
Joined: 16 Feb 2018, 22:49
Location: Germany

Re: One off Galloping Ghost system, for Pulstar

Post by tiptipflyer »

This morning I tested my new GG pulser. It is installed in the Blunderbird, just to test it out. I had two great 10 minute flights without any problem. Good control and it does not even get warm although we had 32° in the shadow and the bird had to fly in the sun. It was well worth to build and test it.
Meanwhile I have finished three sets, one for rudder only, one for rudder-elevator and one with full setup. All are working great.
SAM_3490.JPG
SAM_3492.JPG
Frank
Martin
Posts: 744
Joined: 16 Feb 2018, 14:11
Location: Warwickshire

Re: One off Galloping Ghost system, for Pulstar

Post by Martin »

Brilliant Frank! :D Well done.

I updated the Arduino sketch attached to the post on the first page (pulser.zip). You probably don't want to alter yours Frank if it's already working, but I modified it a bit to suit the latest batch of pulsers I've built. These have the fast motor/gearbox that is happy to work from 5V, so I dispensed with the 'power' wiring and they just operate from the normal receiver (usually BEC) 5V supply to keep the wiring simple. I also altered the sketch so that it auto-detects whether or not the Elevator and Auxiliary channel signals are present and runs without either/both of them if they're not there. So in the simplest installation you only have to plug one 'servo connector' into your receiver rudder channel to fly.

I tweaked the default timing signals to suit 5V operation and the new motors, so again if you already have a working system don't update it or you'll have to retune your rudder/elevator/auxiliary end points on your transmitter.
Twitcher
Posts: 53
Joined: 16 Feb 2018, 12:17

Re: One off Galloping Ghost system, for Pulstar

Post by Twitcher »

Being an inveterate fiddler and happening to have a couple of the same motor/gearboxes doing nothing, I starting wondering if there might be another way to take an output to a rotating torque rod driving the birdcage at the model's back end. There is. This is how I did it. Took all of half an hour!

To begin at the beginning, a while ago Tobe (hiya Tobe!) kindly sent me a bunch of spares so I could repair one of his excellent GG actuators that had gone sick. Being the generous man he is, he also actually included enough parts to build an Adams actuators from scratch! The point is, this appears to use the same gearbox as Martin's set-up. If the motor is the same for all the available ratio variants, it must be.

On Tobe's Adams actuator the rudder drive plate sits directly on the gearbox's output shaft, secured by a) a very snug fit and b) the D-shaped moulded hole that matches the "flat" ground onto the output shaft.

It took just minutes to carve a spare rudder plate into a "T" with a short vertical extension, with a self-tapping screw providing the anchor for a centering spring (I could have just used it "as is"). One scrap of 1.5mm ply, a few M1.6 machine screws and a medium-strength spring later the job was done.

Fired up with Martin's Arduino/H-Bridge driver, it seems to work a treat. The idea is that the torque rod would have a pitchfork-type wire extension to engage in the output adaptor's holes.

Apologies for the quality of the attached photos, but they show how this simple alternative looks.
Attachments
Act 2.jpg
Act 1.jpg
Barry
Posts: 7
Joined: 21 May 2018, 22:11

Re: One off Galloping Ghost system, for Pulstar

Post by Barry »

This weekend I successfully flew my Pulstar, fitted with one of Martin's ready-built Galloping Ghost systems.

Congratulations to Martin on an excellent bit of kit, it works really well.

This was my first venture into Galloping Ghost, and prior to this I had absolutely no idea how the system worked. I had just seen 'Twitcher' flying his Mangled Wot at Middle Wallop earlier this year and thought it would be interesting to find out how this apparent random oscillation of bits of bent wire could possibly control a plane.

So, whilst I was waiting for Martin's kit to arrive I played around with some calculations in Excel to see how the system might work. When I eventually found out that Elevator controlled the pulse rate and Rudder controlled the Mark-space ratio, things started to become a bit clearer. :)

Having now understood the basics, I modelled the mechanics of the birdcage system in 3D CAD to see where the rudder and elevator hinge lines should be in relation to the torque rod to give sufficient control surface movements.

I have posted a video on YouTube which shows the animation of the 3D CAD model, plus the actual system in operation, followed by the first test flights.



Since the video was taken, I have added a bit more downthrust and tweaked the controls to get a bit more down elevator. Flights are now much more controlled and predictable.

Altogether a great project which has taught me a great deal. Next, I want to learn more about Arduino sketches so that I can tweak the control parameters, etc.

Barry
Post Reply