Magnetic Apogee Detector

First up, my apologies, I'm a bit stuck for time at the minute, this is going to have to be a quick post that I come back to.

A few years ago.. OK, before some of you were born. A clever guy by the name of
Bob Galejs worked out that you can use the sensors out of electronic compass' to work out that your rocket is tilting over at apogee. I can't find his original page but this looks to bee a good copy of it http://www.rocketreviews.com/reviews/al ... nsor.shtml

There have been quite a few commercial versions of this circuit produced. By and large they work well though some people have reported problems with them firing when powered on, and suffering interference from large metal objects (eg launch towers).




Whilst Bobs design is elegant. There are three basic problems with it:

1.The circuit simply monitors the output of the magnetic field sensor and fires the ejection charge when a specific fixed voltage is reached. This causes a problem as you move north or south on the planet and the angle of the Earths field changes. if you launch one of these things at the equator, the MAD thinks the rocket is at about 45 degrees whilst it's sitting on the pad.
If you've set your MAD to fire at 45 degrees then you have a problem!

2.Like all things ferromagnetic, the sensor itself can become magnetised. This can happen in the presence of even small EM fields and usually manifests itself as an error in the output of the sensor. To deal with this, the sensor has an additional coil built in. When you pulse some current through this coil it resets or zero's the sensor. Bobs circuit and all of the other variants either have a separate button you push to do this, or pulse the coil when the circuit powers up. Here's the problem: You turn on your MAD and reset the sensor, then you put the rocket on the launch rod (the METAL launch rod that's about 1/2" from your sensor) and the sensor now has an erroneous output, if your lucky and you've set your MAD up to fire when the rocket is horizontal then you'll probably be fine. If you're not or you haven't then you probably won't be fine.

3.Safety. IMHO no recovery electronics should be able to fire the squibs from the moment you turn it on. Time and again we hear stories of people having charges go off in their faces within a second of powering up their electronics. Now there are a few reasons this can happen, but adding a few minutes delay between power up and arm costs you about 20bytes of ram and gets rid of a couple of them. All the common MAD circuits suffer from this.....

OK. If you've got this far then you're clearly committed (or there's nothing on the TV).
My circuit uses a little micro controller to process the amplified output of the sensor and decide what to do when. This allows :

A 2 minute delay between power up and arm.

Just prior to arming (when the rocket is on the pad), the reset coil is pulsed, nulling out any error.

After arming a set of readings is taken, the average of these is stored as the value for pointing straight up, the squib only fires when the sensor voltage swings a fixed amount from the "straight up" value...

Here's the schematic (sorry, I sketched this from memory on a plane coming back from Adelaide)




I like to run these straight from a single 110mAH LiPo cell, it packages up nice:




The board is double sided but there are only a couple of tracks on the back so you could easily do it as a single sided board and just run some wires...


The source code and binary file for programming the PIC are here

Apart from the resistors and caps which are detailed on the schematic, you'll need

A KMZ51 sensor Rockby have them http://www.rockby.com.au/searchres.cfm?searchkey=kmz51
A pic12F675 (everyone carries them)
An INA122U I get'm from Digi-Key, but farnell have them too.
A Buzzer, something like http://www.rockby.com.au/searchres.cfm? ... k_no=12466
Some SOT23 2A FETS (or another package if you are cutting your own board), again Farnell are probably your best bet for small qty's

All up parts cost for one of these is about $15 at QTY1 prices.
If there's sufficient interest, I can cut a small batch of boards and program some micro's.
Nb if you want one of these and are on the list of people who bought a GPS kit, then be ready to prove you've built and flown the GPS when you ask for one of these!

PK

Looks interesting, I think I'll be up for one (maybe make a little microhybrid vehicle for doonside). You still have my GPS so you know it's flown

Thanks for posting that schematic PK, I had been meaning to ask you for it. Andrew, I have boards etched and most of the parts if you want some bits.

Nah!.......I'm just too damned lazy to build such a thing, and my eyes and soldering skills just ain't what they used to be.......................................besides I know someone who can knock one up for me!

Yeah, I'll take one.

Hi PK,

Please put me down for one - I could hear the "scratching" on the plane

After careful calculations on motor selection and delay timing (7 seconds), I suffered a zipper today at a deployment of 10 seconds plus well after apogee - even Rocksim said I was on the money.


I am looking forward to more electronic deployment devices in the fleet.....

Regards,

Understand that I'm offering to make some boards and program a micro or two. It'll be up to y'all to source the rest of the bits.
I'll throw in the INA122U cos I've got thousands of them...
Let's make the cutoff next Friday.....
PK

That's very nice work PK. I'm very interested, though ideally I would need to reprogram the PIC to give me a PWM signal out to drive a servo for my parachute deployment, if you are willing to share the source code.

Please put me down for one as well. I can still use it as is, along with the flight computer we are using, and just hook up the MAD output to a trigger on the comp.

One question:

You mention that:


And also


Will the sensor have an error when it leaves the rail? Or will it not be an issue?

- George

The C source is in the zip file linked in the first post.
George I'm happy to do a mod that generates either a 1 or 2ms pulse every 20ms. very easy to do..
You could always just wrap some nichrome wire around some fishing line, or (heaven forbid) actually use pyro ejection....... I won't tell your mamma!

In my experience (and I should qualify this as about 15 flights on this circuit, with two crashes and the charge firing every time. In those two cases it was clear that the charge produced insufficient pressure to eject the 'chute) it's not an issue. The fact that you zero on the pad and then look for the difference from that value is enough.
Your point is valid though. A better approach would be to detect launch via some other means, wait one second, then reset the sensor. I've done it that way on 3 more significant payloads...

PK

PK
I'm interested in one thanks.
I have used Bobs early ones and found they worked upside down in the southern hemisphere to how he had them set.
Crowie

yeah, I got bitten by that in an early prototype too!

PK

I tried downloading the .zip file (It says it's about 4Kb) but when I try to open it, it's saying it's empty. And I get an error if I try to extract the data from it. I'll try another zip tool in the morning on it in case it is a problem at my end.

I'm happy to do the servo changes to the code. I also need to add a bit of code at the beginning that will set the servo position back to the default closed position after power up and prior to arming so I can actually close the parachute door. ... ejection charges are so 'yesterday'

Tried to download source code but got "Page cannot be found"

also would these power up a Glow Plug as is or need to do a bit of re-programming

Yeah, probably should have tested that link... Fixed now...

Can't see why not. Output power is just a function of the transistor you use. With the existing circuit, you'd have to either find a glow plug that would run on 3.5V, or use some kind of current limiting circuit... Or use a separate supply for the plug.

PK

Better put my hand up for one too thanks PK