TeleMega: Is it Reliable?

reddrock said:

Does anyone have information on how the TeleMega deals with rocket spin or what amount of spin it can sustain and still have good tilt measurement? For example, the RockeTiltometer manual says it can handle 520 degrees per second of spin (~93 RPM) and not affect the tilt calculations. Is there any similar information on the TeleMega?

Click to expand...

The limit of the sensor is 2000 degrees[2000/360=5.56] or 5.56 rps......look at graph of high thrust motor in spin-stabilized rocket [ my 1/3 scale Sparrow-Arcus.......

At .4 sec into flight rotation reaches 5-6 revolutions and sensor pegs............




This is checked by overlaying magnetometer over graph. You can count peaks between 1 second & 1.4 second...it's right at 5.5! Rocket was spinning at 15-16 RPS.
How I intend to keep things flying straight with my 2-stage.



So any spin over 4-5 rps will negate use of tilt. Found this out during first flight.
Rocket was nutz straight, but sustainer did not fire. Limit of sensor is right in specs at 2000.

You lose about .5 or 1/2 degree of accuracy with the tilt control for every 1 second of coast.
12 second coast could easily be off by 6 degrees.
These facts were confirmed by Keith after discussion of my flights. I needed to determine roll rate, so I began using the magnetic field sensors and counting peaks between time.
All must be considered when setting limits.

The problem is that is if you set the gyro max range to 2000 deg/s then there is fairly poor resolution to accurately integrate the smallish rotation rates in the yaw and pitch axes.

blackjack2564 said:

You lose about .5 or 1/2 degree of accuracy with the tilt control for every 1 second of coast.

Click to expand...

The gyros are also sensitive to acceleration, with an additional error of 0.1 degree/sec/g. For most applications, you can pretty much ignore this, but 50g acceleration and you're adding another 5 deg/sec of error in the measurement.

Tilt measurement relies on being able to accurately integrate rotation along all three axes; errors in any axis end up in the final result. We added the mag sensor so I could eventually construct a Kalman filter with that and the gyros to reduce the integration error from just using the gyros. Getting that right means having a lot of data to test with, along with finding time to develop the code. For those of you with long flight times, I would appreciate it if you would send me the .eeprom files so I can add them to the test set.

I'm also starting to explore using tilt measurement for apogee detection; I've got a few high flights and they're all showing a strong 'here's apogee' signal with the airframe rotating through 90 degrees and beyond. Data confirming or contradicting this hypothesis would be greatly appreciated.

keithp said:

Tilt measurement relies on being able to accurately integrate rotation along all three axes; errors in any axis end up in the final result. We added the mag sensor so I could eventually construct a Kalman filter with that and the gyros to reduce the integration error from just using the gyros.

Click to expand...

Including the magnetic data can cause trouble if the launcher includes ferromagnetic materials. I attached a three axis magnetometer to a RDAS many years ago and you can clearly see the effects of the launch rail. That rail was built using the BlackSky model rails (aluminum so no problem) backed up with iron pipe.

So long as you are certain that the launcher includes nothing that will mess up the local magnetic field it will work.

keithp said:

I'm also starting to explore using tilt measurement for apogee detection; I've got a few high flights .

Click to expand...

If the flights are REALLY high the rocket doesn't have to tilt over at apogee.

jderimig said:

If the flights are REALLY high the rocket doesn't have to tilt over at apogee.

Click to expand...

Right, rockets can backslide, but rarely do. What I'm thinking is that some combination of tilt angle, integrated acceleration + delay and baro + accel when in range might work; effectively using the integrated accel + delay as a back-up for the tilt sensor when the flight is above the baro range (about 30km). The trick is to not be too clever and end up with unexpected corner cases...

UhClem said:

Including the magnetic data can cause trouble if the launcher includes ferromagnetic materials. I attached a three axis magnetometer to a RDAS many years ago and you can clearly see the effects of the launch rail. That rail was built using the BlackSky model rails (aluminum so no problem) backed up with iron pipe.

So long as you are certain that the launcher includes nothing that will mess up the local magnetic field it will work.

Click to expand...

Thanks for the reminder. It might be that we actually perform mag sensor calibration *after* the rocket leaves the rail, while the accelerometer-based attitude values are still close enough, but after we're away from external effects on the magnetic field.

keithp said:

Right, rockets can backslide, but rarely do. What I'm thinking is that some combination of tilt angle, integrated acceleration + delay and baro + accel when in range might work; effectively using the integrated accel + delay as a back-up for the tilt sensor when the flight is above the baro range (about 30km). The trick is to not be too clever and end up with unexpected corner cases...

Click to expand...

When I meant by really high was out of the atmosphere like > 150Kft, especially if the rocket has some spin, there is not a lot of air on the fins to tilt over the rocket.

Why not simply use the magnetometer and the vertical component of the magnetic field, ala Robert Galejs?

jderimig said:

When I meant by really high was out of the atmosphere like > 150Kft, especially if the rocket has some spin, there is not a lot of air on the fins to tilt over the rocket.

Why not simply use the magnetometer and the vertical component of the magnetic field, ala Robert Galejs?

Click to expand...

All of the flight data I have from above 100k' (30km) shows the rockets tilting over at apogee. It's not a big sample set, that's for sure, and we know that rockets *can* backslide, so we'd need a backup plan, but for a first-order apogee detection above the range of the barometer, it seems promising.

The problem with the magnetometer is that the measurement it provides is essentially a vector pointing along the magnetic force lines; you can't tell when you rotate *around* those force lines, so you effectively get only two of the three orientation values.

keithp said:

The problem with the magnetometer is that the measurement it provides is essentially a vector pointing along the magnetic force lines; you can't tell when you rotate *around* those force lines, so you effectively get only two of the three orientation values.

Click to expand...

Then how does Galejs' magnetic apogee detector work? Or doesn't it?
https://serge77.rocketworkshop.net/magsens/Magnet3.pdf

jderimig said:

Then how does Galejs' magnetic apogee detector work? Or doesn't it?

Click to expand...

Quoting from that article:

"The ejection angle does depend on E/W/N/S flight path so that it’s behavior will vary from launch to launch"

In particular, a 'perfect' flight where the airframe rotates only perpendicularly to the magnetic field lines will not sense any rotation at all. That's really unlikely, especially as rockets generally spin enough to eventually trigger this mechanism, but I kinda like to have systems which don't have known failure modes, even if unlikely...

keithp said:

In particular, a 'perfect' flight where the airframe rotates only perpendicularly to the magnetic field lines will not sense any rotation at all.

Click to expand...

I think maybe you're not getting how it works. A one-axis magnetometer senses the dot product between its axis and the magnetic field lines. The Galejs system triggers when this goes to zero, with the sensing axis along the body of the rocket. If magnetic field lines were perfectly vertical this would be perfect for tilt detection. They're not vertical, but they're within 35 degrees of vertical everywhere in CONUS according to the map in the paper. If you're confident that the rocket is going to turn all the way over at apogee, this is good enough -- it may be a little late getting to zero but it will still get there, regardless of the azimuth of the tilt.

I'm not convinced that for a high flight relying on a rocket turning over at all is a good idea, but I would want to review all the available data, especially video, from such flights.

Can GPS data be used for some of these measurements or is it to slow?

Tinker

keithp said:

Quoting from that article:

"The ejection angle does depend on E/W/N/S flight path so that it’s behavior will vary from launch to launch"

In particular, a 'perfect' flight where the airframe rotates only perpendicularly to the magnetic field lines will not sense any rotation at all. That's really unlikely, especially as rockets generally spin enough to eventually trigger this mechanism, but I kinda like to have systems which don't have known failure modes, even if unlikely...

Click to expand...

All I know is when I used one back in 2004-2005 it NEVER failed to fire charges at apogee. (and it always worked in my hands...)

jderimig said:

All I know is when I used one back in 2004-2005 it NEVER failed to fire charges at apogee. (and it always worked in my hands...)

Click to expand...

Ditto. I fly with Wooshtronics uMAD magnetic apogee detectors regularly. In fact on Saturday I flew twice successfully with magnetic apogee detection. It's the best "it just works" apogee device I've ever flown. I've only had one on the pad deployment mishap out of my entire usage history. If I didn't have 4 of em I'd consider picking up one of these. It may be worth looking into for you John?

https://zeptobit.com/zeptomag-with-screw-terminals.html

In particular, a 'perfect' flight where the airframe rotates only perpendicularly to the magnetic field lines will not sense any rotation at all.

Click to expand...

This statement is correct, as is the one about the dot product. Magnetic field lines are not parallel to the earth, or perpendicular, so there will almost always be an output from the magnetometer that is varying as the rocket rotates (except in that unlikely event it is actually spinning and parallel to the field lines). The Galegis' analog output fluctuating up and down at the rotation rate of the bird does not matter. What matters is when that output passes the threshold of the comparator that follows it. This generates the signal to fire the deployment. This trigger angle does vary from location to location, because of the inclination of the magnetic field lines. Using a "vanilla" MAD in different geographic locations will have it trigger deployment at different angles. Similarly it varies for the rotation of the rocket. Think of it like a non-spinning rocket having different deployment angles depending on which way it was facing. Likewise, for us Southerners in Oz, we need to mount the MAD the other way up as the inclination is in approximately the other direction.

The MAD is very simplistic in the single-axis mode, hence the variability of the trip angle. If a three-axis one is used, and the maths done for a vector sum of the orthogonal readings, it is relatively straight-forward (in a slightly complicated way) to reliably trigger at a particular tilt angle.

Tinker said:

Can GPS data be used for some of these measurements or is it to slow?

Tinker

Click to expand...

GPS appears to track accurately in about 2/3 of the flights up this high. And in the other 1/3, it fails to re-acquire lock during boost and misses apogee entirely, sometimes by a long time.