I could use just a little guidance

[JimJarvis50]

So, let's get back to the LDRS flight. This was the two stager with the stabilization section on top of the sustainer. The sustainer didn't light because the rocket didn't reach the required staging altitude. This was due to coning on the boost (seems to be the theme this year). I put together the video from the flight and it's attached.

The question on this flight is why did the rocket cone? One possibility is that it was too windy for the flight (I should have waited, as the weather got better). Another possibility is that roll control from the top of the rocket may just not work very well on a long rocket. Or, these two factors might have combined to cause the motion of the flight.

I'm planning to try this flight again in a few weeks (Distant Thunder). Motors would be a CTI M3100 (instead of the Gorilla M1665), followed by the Gorilla M745 that didn't light. I need to decide what to do with the control system. I can have it active at launch, as I did on this flight, or I can delay turning it on until after stage separation. The planned flight profile is also attached. Thoughts?

Jim

 

[OverTheTop]

Hi Jim. I have a G-switch in mine, so it does not attempt to control anything when the gyros and accelerometers are pegged. Most commercial gyro and accel chips are only specified to about 6G. Outside that and all bets are off. If the firmware has been written to ignore values in this part of the flight envelope then you will be ok. I don't know what gyros you have used, or what your code is like, but the hardware specifications will be key. Your sim shows acceleration of over 10G for both booster and sustainer burns...

FYI I also have a thin wire link as part of the enabling circuit. At deployment the tug on the harness cuts the link and puts the system into a fixed position. That stops the windmilling on the way down frying batteries or servos.

 

[jderimig]

Hi Jim. I have a G-switch in mine, so it does not attempt to control anything when the gyros and accelerometers are pegged. Most commercial gyro and accel chips are only specified to about 6G. Outside that and all bets are off. If the firmware has been written to ignore values in this part of the flight envelope then you will be ok. I don't know what gyros you have used, or what your code is like, but the hardware specifications will be key. Your sim shows acceleration of over 10G for both booster and sustainer burns...

FYI I also have a thin wire link as part of the enabling circuit. At deployment the tug on the harness cuts the link and puts the system into a fixed position. That stops the windmilling on the way down frying batteries or servos.

Once the gyros are pegged you are done. You have lost orientation for the rest of the flight. Even if they come back in range

 

[JimJarvis50]

Hi Jim. I have a G-switch in mine, so it does not attempt to control anything when the gyros and accelerometers are pegged. Most commercial gyro and accel chips are only specified to about 6G. Outside that and all bets are off. If the firmware has been written to ignore values in this part of the flight envelope then you will be ok. I don't know what gyros you have used, or what your code is like, but the hardware specifications will be key. Your sim shows acceleration of over 10G for both booster and sustainer burns...

FYI I also have a thin wire link as part of the enabling circuit. At deployment the tug on the harness cuts the link and puts the system into a fixed position. That stops the windmilling on the way down frying batteries or servos.

My system, similar to that of the Altus Metrum, disregards the accelerometers at launch detect. This is to avoid recalibrating the gyros to an incorrect direction for "down". I think systems that don't do this could have problems, but it would depend on the rate of recalibration. I don't really know what this rate is for RC autopilots for example. My system is designed such that the rate of recalibration, prior to launch, is slow. It takes nominally 10 minutes for the offsets to be reduced to zero (I have to make sure I wait that long after powerup before launching). The gyro offsets are corrected, slowly, while the rocket is on the pad, but the offsets are then not affected by wind motion on the pad or the initial movement of the rocket.

The gyros are currently configured for a range of 1000 degrees/sec. This was originally 500 degrees/sec, but that proved to be too easy to reach, so we changed to 1000. The only reason to have the system operating during the initial boost is to prevent the spin rate from reaching that value. It has been pointed out that most rockets won't get to that spin rate on their own, and it might be possible incur that rate if spin control is on but not working well, as might be the case for a long, fiberglass (not carbon) rocket. So, that's the decision I need to resolve.

Jim

 

[TonyL]

Jim,
it is worth testing your inertial sensors for g-sensitivity as well as looking at their specs, as mentioned if they rail it causes problems. Also control under boost is different from coast since you are steering the thrust vector. I would expect control during coast to be friendlier.

During the first burn the rocket only goes around 2 times, so I'm not sure 'coning' is the problem as much as the lateral component of the trajectory. Weather cocking and launch pad tip-off would be additional suspects. You also would not be the first one who's flight did not match the sim , you might look at the predicted drag vs. the measured drag to see if there are any big discrepancies.

I think you said you had roll control on the upper stage [and a 'spin-can' on the booster]? Had you turned the gain down on this one?

br/

Tony

 

[Chris_H]

Once the gyros are pegged you are done. You have lost orientation for the rest of the flight. Even if they come back in range

Can you please clarify this? Does this mean that once a gyro has undergone acceleration beyond a certain point (6G's?) that it no longer is able to give gyro control during that flght? Does that include vertical acceleration? Or is that a spinning measurent ?

Trying to understand.

 

[mbeels]

Can you please clarify this? Does this mean that once a gyro has undergone acceleration beyond a certain point (6G's?) that it no longer is able to give gyro control during that flght? Does that include vertical acceleration? Or is that a spinning measurent ?

Trying to understand.

I believe he is saying that you lose your reference. You no longer know where "zero" is. You can obtain relative changes, but you no longer know with respect to what angle.

 

[jderimig]

Can you please clarify this? Does this mean that once a gyro has undergone acceleration beyond a certain point (6G's?) that it no longer is able to give gyro control during that flght? Does that include vertical acceleration? Or is that a spinning measurent ?

Trying to understand.

I believe he is saying that you lose your reference. You no longer know where "zero" is. You can obtain relative changes, but you no longer know with respect to what angle.

Yes. The rocket attitude is done by integrating small rotations on each axis then performing a coordinate system rotation. If the rotation rate in one axis exceeds the full scale rating of the gyro then the true position of that axis in unknowable. With rotations this is especially insidious because if one axis has an error then the other 2 axes will have an error also.

With accelerations the error is confined to the axis that has railed. The other 2 axes' integration is still valid.

 

[Charles_McG]

Yes. The rocket attitude is done by integrating small rotations on each axis then performing a coordinate system rotation. If the rotation rate in one axis exceeds the full scale rating of the gyro then the true position of that axis in unknowable. With rotations this is especially insidious because if one axis has an error then the other 2 axes will have an error also.

I figured that if you saturated a gyro, you'd loose a fixed orientation - like keeping fin 1 pointed at due north. But I would have guessed that once it desaturated, it would try to zero the rate. So it would stop rotation, but fin 1 would be pointed in a random direction.

 

[jderimig]

I figured that if you saturated a gyro, you'd loose a fixed orientation - like keeping fin 1 pointed at due north. But I would have guessed that once it desaturated, it would try to zero the rate. So it would stop rotation, but fin 1 would be pointed in a random direction.

Yes you can control rotations when the gyro comes back into its range but you won't know where you are pointing.

 

[Charles_McG]

Yes you can control rotations when the gyro comes back into its range but you won't know where you are pointing.

If you were only trying to control roll rate, and absolute roll-orientation didn't matter, then that would be ok.

But if the roll axis saturates, you also loose track of the roll induced error corrections in the pitch and yaw gyros (which exist because of how chip based gyros work?) - so when the whole system re-stabilizes, you might not only have fin 1 pointed in a random azimuth, but pitch and yaw will have uncontrolled (and uncontrollable) error, so the rocket may be pointed toward a new altitude and azimuth?

(Not doubting you - just trying to understand.)

 

[jderimig]

If you were only trying to control roll rate, and absolute roll-orientation didn't matter, then that would be ok.

Yes

But if the roll axis saturates, you also loose track of the roll induced error corrections in the pitch and yaw gyros (which exist because of how chip based gyros work?) -

If you lose track of the location of where the roll axis was rotated to (which happens when you saturate the gyro roll axis) AND you have a pitch, you would not know where the new z-axis (relative to the earth frame) is pointed.

Try it with your hand.

 

[JimJarvis50]

Jim,
it is worth testing your inertial sensors for g-sensitivity as well as looking at their specs, as mentioned if they rail it causes problems. Also control under boost is different from coast since you are steering the thrust vector. I would expect control during coast to be friendlier.

During the first burn the rocket only goes around 2 times, so I'm not sure 'coning' is the problem as much as the lateral component of the trajectory. Weather cocking and launch pad tip-off would be additional suspects. You also would not be the first one who's flight did not match the sim , you might look at the predicted drag vs. the measured drag to see if there are any big discrepancies.

I think you said you had roll control on the upper stage [and a 'spin-can' on the booster]? Had you turned the gain down on this one?

br/

Tony

The chip used on my controller, and on Altus Metrum I believe, is the Invensense MPU6000. Pretty common. Yes, they do have some g sensitivity. However, on flights where I can actually look at the tilt relative to the actual tilt (a horizon or gps trajectory), the measured tilt is pretty good. If it wasn't, I would have given up long ago.

Yes, coning is probably not the right term. There were a lot of flights at LDRS where there was a clear indication of shear a few hundred feet up. I can't say I saw it on the day I flew, but it was obvious on later days. In the video, it looks like this might have happened about 1.5 seconds into the flight.

I have checked the actual flights of this rocket against the simulations many times, and they are quite good. I don't think a mismatch of the simulation was the issue.

Actually, the spin can was on the sustainer and not the booster. But from the video, the spin can wasn't spinning very much, perhaps because the controls were turned down on this flight. In addition to turning down the overall response, the roll response was turned down relative to the yaw/pitch response. That, plus the relatively low velocity of the boost, is why I suspect wind and not the control system as the cause of the problem.

Jim

 

[Charles_McG]

If you lose track of the location of where the roll axis was rotated to (which happens when you saturate the gyro roll axis) AND you have a pitch, you would not know where the new z-axis (relative to the earth frame) is pointed.

Try it with your hand.

I seem to be misunderstanding 'roll'. If I stick my fist out with my thumb up, and 'roll' in the direction of my fingers (like current and magnetic fields), my thumb stay vertical.

If I tip (pitch) over my fist so that my thumb points at the window frame and do the same, my thumb stays pointed at the window frame. It only moves if I consider 'roll' to be around the original 'z' axis. But I would normally interpret roll to be around my thumb.

 

[JimJarvis50]

I figured that if you saturated a gyro, you'd loose a fixed orientation - like keeping fin 1 pointed at due north. But I would have guessed that once it desaturated, it would try to zero the rate. So it would stop rotation, but fin 1 would be pointed in a random direction.

You could still control roll after saturation, but if the rocket rolls, then pitch/yaw control gets screwed up unless just by coincidence, there was 360 degrees (or a multiple of that) of unaccounted-for roll. Pretty unlikely to get that lucky.

Jim

 

[jderimig]

I seem to be misunderstanding 'roll'. If I stick my fist out with my thumb up, and 'roll' in the direction of my fingers (like current and magnetic fields), my thumb stay vertical.

If I tip (pitch) over my fist so that my thumb points at the window frame and do the same, my thumb stays pointed at the window frame. It only moves if I consider 'roll' to be around the original 'z' axis. But I would normally interpret roll to be around my thumb.

Extend your hand and your index finger. Roll is the rotation about your wrist. Now pitch your finger as you roll your wrist. The final direction you are pointing depends on your roll angle.

 

[Charles_McG]

If you lose track of the location of where the roll axis was rotated to (which happens when you saturate the gyro roll axis) AND you have a pitch, you would not know where the new z-axis (relative to the earth frame) is pointed.

Try it with your hand.

<epiphany\>
Wait. Wait. Wait. I think I've got it.

The gyro is reporting difference from it's original 'home' state, not its last-prior-to-current state. If it's stabilized on the pad, then roll -is- around the -original- z-axis, not the one at the moment. So roll shows up as (what sin(pitch)?) on the pitch/yaw gyros.
<\epiphany>

Maybe not the way I'd imagined at first, but you don't have to integrate rates moment to moment to keep track of where you're pointed.

 

[Charles_McG]

Extend your hand and your index finger. Roll is the rotation about your wrist. Now pitch your finger as you roll your wrist. The final direction you are pointing depends on your roll angle.

Because your axis of rotation (arm) didn't move. I was thinking of roll relative to the long axis of the rocket - but that's not what the gyro is reporting (as I now understand it). And your wrist isn't built to make the motion I was thinking of - you have to turn your whole arm to do it.

 

[jderimig]

Maybe not the way I'd imagined at first, but it would me you don't have to integrate rates moment to moment to keep track of where you're pointed.

No you have to integrate from the last calculated rotation because rotations are not additive. The gyro measures instantaneous rotation. If you want to determine the absolute rotation in the earth frame then have to measure small rotations from the last position to calculate the new orientation of the rocket coordinate system relative to the earth coordinate system. At every step you are doing a 3D tranformation of the rocket coordinate system.

 

[Charles_McG]

No you have to integrate from the last calculated rotation because rotations are not additive. The gyro measures instantaneous rotation. If you want to determine the absolute rotation in the earth frame then have to measure small rotations from the last position to calculate the new orientation of the rocket coordinate system relative to the earth coordinate system. At every step you are doing a 3D tranformation of the rocket coordinate system.

Which is what I originally thought.
So is the gyro reporting degrees/sec relative to its orientation on the pad? Or its current orientation?

 

[jderimig]

Which is what I originally thought.
So is the gyro reporting degrees/sec relative to its orientation on the pad? Or its current orientation?

From its current orientation . Then the new orientation is calculated by a 3D matrix rotation or a quaternion operation depending on the numerical method chosen.

 

[Charles_McG]

From its current orientation . Then the new orientation is calculated by a 3D matrix rotation or a quaternion operation depending on the numerical method chosen.

Sorry for being dense, and thank you for taking the time to educate a biochemist.

But I'm still not seeing how roll around the local long axis translates into change in pitch/yaw. I took some pictures, but am having upload bandwidth trouble at the moment.

Imagine I hold a pen pointed up above my desk. I can 'roll' it between thumb and fingers and the clip changes azimuth while the point stays in place.

I tip the pen in one axis so it points toward the computer screen. I can still roll the pen around, and it stays pointed at the same point on the screen. If I turn it around a line normal to the desk (parallel to the pen standing vertically), then the tip points around in a circle in azimuth, tracing out a cone. That's not the motion I would call 'roll' - but that's how the gyro sees it?

 

[jderimig]

You will need to pitch and roll simutaneously to see that the final position is not the same as the sequence of uniaxial rotations. Google Euler angles to see all the gory math in detail.

Sorry for being dense, and thank you for taking the time to educate a biochemist.

But I'm still not seeing how roll around the local long axis translates into change in pitch/yaw. I took some pictures, but am having upload bandwidth trouble at the moment.

Imagine I hold a pen pointed up above my desk. I can 'roll' it between thumb and fingers and the clip changes azimuth while the point stays in place.

I tip the pen in one axis so it points toward the computer screen. I can still roll the pen around, and it stays pointed at the same point on the screen. If I turn it around a line normal to the desk (parallel to the pen standing vertically), then the tip points around in a circle in azimuth, tracing out a cone. That's not the motion I would call 'roll' - but that's how the gyro sees it?

 

[JimJarvis50]

Also control under boost is different from coast since you are steering the thrust vector. I would expect control during coast to be friendlier.

On the system I have, I can have the canards fixed in a neutral position, and then roll and yaw/pitch control can be turned on separately during the flight whenever I want. One option might be to have roll control on from launch but not activate yaw/pitch until after separation. On the long rocket, yaw/pitch control would be pretty slow anyway.

Jim

 

[OverTheTop]

Once the gyros are pegged you are done. You have lost orientation for the rest of the flight. Even if they come back in range

That has not been my experience so far. I use an autopilot for an RC aircraft in my system (no customisation) and the outcome of that is that the reference vector drifts from vertical slowly once powered up. The flight still happily tracks to a straight vector once the VTS system enables the control surfaces.

My next step is to get into the firmware of a control system so I don't get the drift before the flight.

 

[OverTheTop]

One option might be to have roll control on from launch but not activate yaw/pitch until after separation.

If the gyros are outside their operating envelope (acceleration) the results can be unpredictable. Depends on the type of gyro and their specifications.

 

[Charles_McG]

Jim et al, I'm sure this is well known - but the Malemute (Sounding rocket) Development PDF posted over in :
https://www.rocketryforum.com/threads/malamute-terrier-malamute-scale-data.155496/
has a section on exoatmospheric coning. To avoid a spin-motor issue, they fly a profile with no spin during the burns, then spin up during coast in air, then despin above the atmosphere. With graphs about what they are trying to avoid - both in burn and in tumble.

 

[JimJarvis50]

Jim et al, I'm sure this is well known - but the Malemute (Sounding rocket) Development PDF posted over in :
https://www.rocketryforum.com/threads/malamute-terrier-malamute-scale-data.155496/
has a section on exoatmospheric coning. To avoid a spin-motor issue, they fly a profile with no spin during the burns, then spin up during coast in air, then despin above the atmosphere. With graphs about what they are trying to avoid - both in burn and in tumble.

I'll take a look. Thanks.

Jim

 

[BABAR]

How do you pick the minimum altitude for lockout? If the you can trust the onboard sensors that the rocket is

On a vertical trajectory
At adequate velocity that the sustainer is stable

What is the lowest safe altitude to let the sustainer fire?

 

[JimJarvis50]

How do you pick the minimum altitude for lockout? If the you can trust the onboard sensors that the rocket is

On a vertical trajectory
At adequate velocity that the sustainer is stable

What is the lowest safe altitude to let the sustainer fire?

On the LDRS flight, I used a Raven, which uses the logic statement "light at Altitude = A if Time < T". Based on the simulation, I found the value of T at the point where the altitude was expected to be A, and then I set Time < T to be about 1.5 seconds later than what was predicted in the simulation. This is a pretty tight approach, but if you are using altitude and not tilt, then it needs to be that way if you expect any real protection from an off-axis flight. As it turned out, the rocket didn't get to A soon enough and the sustainer didn't light. Indeed, the rocket never even reached A. Given the angle of attack issues during the boost, the approach did exactly what it was supposed to do.

In the repeat of the flight, I plan to use an EasyMega, which is more suited to this flight. I will still use an altitude check, but it won't be as tight as what I used for the Raven (since tilt is the primary criteria for inhibiting the sustainer). On the EasyMega programming, I will also use a one second delay in the programming. It turns out that tilt, along with all of the other permissives, will have to be within the set limits during that one second delay period, of the sustainer won't light. This approach ensures that the rocket is not tumbling. The EasyMega program is attached.

I would have used the EasyMega at LDRS except that I had to reserve it for the Balls flight, just in case there was an issue with Kate. Oh well, just like a failed certification, I have to fly it again. Dang...

Jim