Basejumper.com - archive

I have invented a pure fucking magic bone stimulating machine - PFM BSM - aka Z-Device or Z-Schlong.

Coming to Orthopedic surgeon and sex toy shops near you.

that looks like a AOA sensitive airspeed indicator to me.

It is, plus accelerometers.

Since you're banned from the wingsuit forum where I introduced this beast on the dawn of the wingsuit-revolutionizing 2008, I introduce it here for you, one of the most creative people on Earth advancing the Dream of Flight, in this very appropriate Bone Stimulators thread.

Z-Device version 0.0.0.0.0.0.1

Yo!

Introducing Z-Device - a wingsuit flight characteristics measurement and recording device. It is capable of measuring directly at 10 samples per second:

- total airspeed V
- altitude Y
- components Gv1 and Gv2 of apparent gravity (parallel and perpendicular to velocity, respectively)
- components Gp1 and Gp2 of apparent gravity (parallel and perpendicular to pitch axis, respectively)
- temperature
- (optional) GPS data (4 samples/s)

From the raw data, these flight parameters can be derived:

- horizontal and vertical speeds
- glide ratio
- L/D (which is not equal to G.R., in general)
- adjusted lift and drag coefficients Kl and Kd (and from them, sustained horizontal and vertical speeds for given body position and AoA, and thus - polar curve)
- angle of attack
- lift and drag as functions of AoA

At the core, Z-Device is based on this idea: if an accelerometer is mounted on a vane, the ratio of the components of apparent gravity perpendicular (Gv2) and parallel (Gv1) to the airspeed is equal to [acceleration corrected, true] L/D:

L/D = Gv2/Gv1

In addition, vane allows one to mount static and dynamic Pitot tubes accurately pointing into the relative wind and thus accurately measure airspeed and altitude.

The vane is mounted on a collapsible pod on the belly so before the jump you extend it about a foot away from your body into the clean airflow where it tracks the relative wind.

Another accelerometer is installed on the bellymount and the relative readings of the two accelerometers allow one to calculate the angle between the relative wind and body's pitch axes - that is, the effective angle of attack.

Temperature is measured by a temperature sensor to make corrections to altitude and speed data.

Optionally, GPS sensor can be installed that samples 4 times per second. However, with accurate and true air speed and accelerometer data, GPS can be useful only on BASE jumps.

Also optionally, speed, descent rate, and accelerometer readings can be viewed in real time on a wireless dashboard that can be mounted on belly or chest strap.

For electronics, I used the wonderful Eagle Tree Systems Flight Data Recorder. (The accelerometers I am using are not the ones advertised on their site (which have +/-38g range and are not very accurate, they're custom-made with +/-1.7g chips, providing about 0.005g accuracy)). The total cost of electronics is ~800 bucks.

I made 3 jumps with Z-Device in Z-Hills on New Years Eve, the results so far are promising. The vane is rock stable - no wobbling, it just sits there quietly pointing into smooth relative wind. However, 1ft from the body may not be enough to get the vane away from the boundary layer, as on one jump I could see that the vane got stuck pointing more horizontally than it should have. I will investigate this by attaching wool threads to the stick at different distances from the belly and see where the air becomes 100% clean.

I couldn't calibrate accelerometers due to software issues, so the accelerometer data I got was garbage. This needs more work.

The data will be processed using Kalman filter based on wingsuit equations.

We'll see some amazing stuff in 2008!!!

Yuri

I split this from the original thread. it seemed a looooong way off-topic.

rub it in that I am banned from the wingsuit forum.

that is very, very cool.
I have used a 'total energy indicator' in sailplanes, it is a VERY cool system.

this looks even more advanced without the burden of needing to be certified.

VERY COOL YURI!!!

Took a look real quick and it looks like an interesting device. Just for your interest or further development: have you played with an "integrated" IMU (inertial measurement unit) ? It's basically a little box that gives you triaxial accelerometer, magnetometer, and rate gyro readings, all nicely temperature compensated. You could get yaw/pitch/roll out of it, as well as the angular rate. If you could sample this data and later play it back, it may be useful for people who do aerials (aerial guys, input?).

The one sticking point that I can think of is: how well do accelerometers work in a falling environment? We've got one on our robot and it seems to work, but that bugger doesn't accelerate nearly as fast as a falling jumper would.

I know Memsense makes IMUs but they tend to be on the expensive side.. There are probably cheaper integrated solutions out there.

I can only think of 2008 (in terms of wingsuits) as a old lady napping in a rocking chair and somebody puts fireworks all around her and sets them off.

The 100-year old reign of "we'll put some fabric between our arms and legs to look like flying squirrels" is over. We are about to see some really cool shit! If we try.

So let's just do it, man.

The problem with IMU's is that they drift. You'll need to set it up on the ground and by the time you get to altitude in airplane they'll show garbage. In BASE, they might work, if you turn them on and calibrate just before the jump. But with the idea of accelerometer on a vane, we get direct measurement of L/D, so IMU is not necessary.

The prototype I built needs much improvement before it'll output really useful data. Currently, the biggest problems are vibrations from the turbulence [behind the vane], misalignments, and boundary layer. But these problems are solvable, the version 0.0.0.0.0.0.0.2 is in the works.

In terms of the cost, Eagle Tree's stuff is the most affordable. The recorder with the wireless dashboard is about $300. The accelerometers came out expensive - $200 each, since they are custom made 1.7G versions of the standard 38G accelerometers.

I considered Memsense but rejected it because of the costs and they are not really suitable for what I want.

Great work, Yuri! I look forward to seeing the results when it's fine tuned.

As I've mentioned before, you have way too much free time!

Have you looked at SparkFun? I haven't actually bought anything from them, but it's one of these sites I visit frequently just to see what new toys they have.

Michael

I do look at SparkFun once in a while, but they're oriented more towards electronics whiz's, which I'm not (when I grab a soldering iron, I often painfully find out that I'm holding it by the wrong end). Eagle Tree is 100% plug-n-play and has all the sensors I need.

Yuri,

I assume you're doing all filtering/sensor fusion post-collection, right? In what package? (PM if it's secret! )

Any intent to fuse the IMU and GPS data together? It's a viable option to correct for the inevitable IMU drift (at least for some DOF).

Does the 10Hz sampling rate apply to all instrumentation? How well does the Kalman filter track 'truth' at that acquisition rate? Any possibility of higher collection rates?


It'll happen from time to time.

.jim

Jim,

The data filtering will be done post-collection, using Kalman filter based on wingsuit equations. I will write the code myself.

Kalman filter by its very idea fuses data from various sensors in order do derive unknown parameters and smooth out the measured values.

10Hz applies to all sensors (actually, accelerometers can be polled at 100Hz, there's an option for that in configuration software). GPS module (which I do not have as I see not much need for it, and it's $150) samples at 5Hz.

Since Kalman filtering will be done post-processing, there is no need to worry how well it will track fast changes at 10Hz, we are only interested in relatively slow changes. (Z-Device is not a system to provide instant L/D feedback, although the flight parameters can be visually monitored on a chest-mounted dashboard.)

Currently, I'm mostly interested in accelerometer data (since it gives me L/D directly) and airspeed data. IMU and GPS may enter the picture later, however, I hardly see any need for them - the current set of variables is self-sufficient for Kalman filtering.

Thanks for your questions, I want more!

Yuri

Any plans for a head-mounted display of real-time data?

Yuri,

With regard to an intermediate step from the suggestion of the poster above:

Any intention of turning this into a variometer-like device except with L/D (i.e a changing tone indicative of glide ratio)? I know you mentioned that this isn't meant to quickly track, but it seems like that might be an awesome way to figure out how to really max out a suit (besides 'feel' alone).

.jim

Jim,

It's not there yet. The design requires that the vane is in clean airflow. This already makes it awkward - you need to mount it on the belly or helmet. It's definitely not a protrack for the masses, it's more of a research tool.

It is possible to make it into real time L/D indicator, that will require some embedded programming. Anybody who understands what I'm talking about is welcome to join the revolution!

Yuri

Did you say "embedded programming"? I could try to help. I've got some experience with Microchip 16F, 18F, dsPIC devices. They are small microcontrollers with build-in flash, ADCs, GPIO, and other peripherals. And the best thing is: you can get free samples of them!

What kind of embedded device would be needed? What would the computational load be? What exactly would you want the program to do?

For example, take input from two-axis accelerometer (accelerometers at SparkFun) and calculate ratio of the X and Y components. Smooth it out (Kalman is plus, but not absolutely necessary) and display on a small chest-mounted or HUD display. Make the whole thing as miniature as possible so it can be mounted on a small vane on helmet.

Real time L/D meter..... music to my ears!!!

That does not sound too bad. You could probably pull it off using a PIC16F876A.

It's got a few ADC inputs that can be used to grab data from the accelerometer. I expect it outputs a signal in the range of 0-Vcc, depending on the measurement.

After the data comes back from the ADC inputs, it would have to be smoothed out somehow... I don't know anything about Kalman filtering, but maybe a simple moving average filter would work?

Then it's a matter of dividing the numbers and displaying the result. I'm thinking of a generic parallel LCD, something like this, although I have seen a smaller (8x2 instead of 16x2) version. Most of these character displays take the standard 4 or 8 bit protocol, so you could run it using 6 I/O pins.

Fortunately, there is a C compiler out there for the 16F chips although it has its bugs.

Accelerometers can be polled 100 times a second, so simple averaging of 100 points every second should give you nice smooth data.