Staging Go-No-Go Redux Squared

[Art Upton]

Ok new year,

time to rehash old ideas and see if any new wisdom or technology has solved this yet or not.

Last we left our heros were talking about how to make sure our rockets where straight up and not off course before firing the sustainer.

Back in 2005 several mentioned tilt switches and many recanted how if you fall off the clift it still can't tell you are really going down or up.

2006 had many talk of the MAD and also others talk about MAD issues.

2007 had many point to the new 3-5, and 6 axis accelerometer

Bob mentioned that a 3-axis accelerometer and a compass could be used to tell your attitude. Robert mentioned you should be able to do it with 6-axis somehow.

Many pointed out the new tilt compensated compass, but again it was pointed out anything with tilt-compensataion will work for a robot or sub, but not a rocket in free space.

It's time to bring up our friends at SparkFun again with two items that have me thinking today:

3-Axis Magnetometer
https://www.sparkfun.com/commerce/product_info.php?products_id=244

IMU 6 Degrees of Freedom
https://www.sparkfun.com/commerce/product_info.php?products_id=8191

Pros?
Cons?

Lets Discuss

 

[jderimig]

3-Axis Magnetometer
https://www.sparkfun.com/commerce/product_info.php?products_id=244

IMU 6 Degrees of Freedom
https://www.sparkfun.com/commerce/product_info.php?products_id=8191

Pros?
Cons?

Lets Discuss

Do we have to limit ourselves to on-board inertial systems?

One idea I had was a low-power laser or very bright collimated light at the pad shining up. On the rocket is a pair of detectors looking down. The detectors would have a reasonable but not too narrow angle of view. If the detectors do not see the light then do not fire the second stage.

 

[uncle_vanya]

What about simply using a remote control and letting your human brain solve this? That's what our club prefect does. He can fire the separation charge and ignite the 2nd stage - or allow it to continue to arc over and watch the altimeter blow the laundry. With more channels of control you could make the decision to blow the separation charge and not light the 2nd stage.

 

[Art Upton]

John, what about rockets that travel so far up "rather" but not totaly straight that while still going nearly straight up, they are about 500 or more feet north of the launch pad at this point.

Would they still see the beam?
>>The detectors would have a reasonable but not too narrow angle of view

Does this help solve this?

Remote control idea sounds good Brad,

But look at this video here of this nearly straight rocket two stage flight:
https://BoosterVision.com/wmv/mtom.wmv

At 12,000 feet the human eye can't truly see if the rocket is straight or not until after the second stage is ignited and you can see the trail again.

How high has your prefect staged with the remote control, it does sound good for flights that can be seen.

In one or more cases what I am looking for might have to stage between 20-40,000 feet.

 

[Art Upton]

Do we have to limit ourselves to on-board inertial systems?
....

Interesting out of box thoughts happened. What if the rocket trailed two magnet wires behind it on really long spools?

Eh, your idea is much more practical that that one in reality

 

[als57]

Interesting out of box thoughts happened. What if the rocket trailed two magnet wires behind it on really long spools?

Well they basically do something like that with anti-tank missles and torpedos.

Al

 

[jderimig]

John, what about rockets that travel so far up "rather" but not totaly straight that while still going nearly straight up, they are about 500 or more feet north of the launch pad at this point.

Would they still see the beam?
>>The detectors would have a reasonable but not too narrow angle of view

Does this help solve this?

Got to do trig and all that to figure that out. Might involve tangents and stuff like that.

 

[Silverleaf]

Sun Guidance would be one suggestion.

I'm not familiar enough with the two items that you posted Art, but if you have more information/ideas/suggestions and are willing to post them here, I for one would be very interested in learning more.

Cheers,

 

[Art Upton]

Sun Guidance would be one suggestion.

I'm not familiar enough with the two items that you posted Art, but if you have more information/ideas/suggestions and are willing to post them here, I for one would be very interested in learning more.

Cheers,

Hi Silverleaf,

some vertical stability guidence has been done before, ER Media has a booklet on it. George Gassaway's sun guidence has been the lightweight guidence model so far.

Myself and some others how have posted on this before are looking for something for Min. or Near Min. Diameter staged rockets that will be close to or out of sight when staged. Many ideas were tossed up in the years past.

I believe vertical stability guidence can be done with these same IMU items that will give you this go-no-go ability. It is the next step that is possible if you have an IMU.

I know that Richard Hagensick is working also on vertical stability guidence on his Q motor project.

What I'm looking for is the guys that know this free-space-motion stuff to either say these types of IMUs can work, go do the rest of the work now Art; Or to shoot it down.

I'm also looking for someone like John who proposed an outside the box approach.

Funny thing is John knows his Trig and Tangents :surprised:

 

[JimJarvis50]

Ok new year,

time to rehash old ideas

Hey Art, as long as we can rehash ....

I use an approach that involves an altimeter and a staging timer. The altimeter has an output channel that can be programmed to fire "at a specified altitude on ascent". There are several commonly used altimters with this capability. The output from the altimeter is connected to an electronic relay which is connected to the timer/ematch circuit. The altimeter channel has to fire before the timer circuit can fire the ematch. The logic is best illustrated by example.

Say you decide to fire the sustainer 5 seconds into the flight. The latch time for the timer might be from 5 to 6 seconds, depending on the timer. You go to Rocsim and do a simulation, and calculate that the rocket should be at 6000 and 7000 feet at 5 and 6 seconds into the flight. So, you program the altimeter to fire at, say, 6500 feet (you could choose anything from 3000 feet to 6999 feet, depending on your comfort level). Then, if the rocket gets to 6500 feet in under 6 seconds, the "on ascent" channel will fire, and the electronic relay will allow the timer to light the sustainer. However, if the rocket takes longer than 6 seconds to get to 6500 feet, it won't light. So, in the event the rocket goes off at an angle, or has some other bad event, the sustainer won't light.

I've used this now on two flights, and in both cases, the flight was good at the staging point and the sustainer lit (I'm not really intersted in proving that it won't light if I can avoid it, but the capability to terminate the flight is there is something happens). Obviously, you need one of the higher-end altimeters to do this (one with the fire "on ascent" capability). Beyond that, however, the electronics budget is pretty cheap. I use a simple transistor switch for the relay, and the cost is about $2.00.

I recognize that this approach doesn't use a direct measurement of the rockets' attitude to control the sustainer ignition. However, I think that it is logical to assume that if a rocket is close to the expected altitude at a specified time, it is most likely pointed up.

 

[Art Upton]

Hey Art, as long as we can rehash ....

I use an approach that involves an altimeter and a staging timer. The altimeter has an output channel that can be programmed to fire "at a specified altitude on ascent". There are several commonly used altimters with this capability. The output from the altimeter is connected to an electronic relay which is connected to the timer/ematch circuit. The altimeter channel has to fire before the timer circuit can fire the ematch. The logic is best illustrated by example.

Say you decide to fire the sustainer 5 seconds into the flight. The latch time for the timer might be from 5 to 6 seconds, depending on the timer. You go to Rocsim and do a simulation, and calculate that the rocket should be at 6000 and 7000 feet at 5 and 6 seconds into the flight. So, you program the altimeter to fire at, say, 6500 feet (you could choose anything from 3000 feet to 6999 feet, depending on your comfort level). Then, if the rocket gets to 6500 feet in under 6 seconds, the "on ascent" channel will fire, and the electronic relay will allow the timer to light the sustainer. However, if the rocket takes longer than 6 seconds to get to 6500 feet, it won't light. So, in the event the rocket goes off at an angle, or has some other bad event, the sustainer won't light.

I've used this now on two flights, and in both cases, the flight was good at the staging point and the sustainer lit (I'm not really intersted in proving that it won't light if I can avoid it, but the capability to terminate the flight is there is something happens). Obviously, you need one of the higher-end altimeters to do this (one with the fire "on ascent" capability). Beyond that, however, the electronics budget is pretty cheap. I use a simple transistor switch for the relay, and the cost is about $2.00.

I recognize that this approach doesn't use a direct measurement of the rockets' attitude to control the sustainer ignition. However, I think that it is logical to assume that if a rocket is close to the expected altitude at a specified time, it is most likely pointed up.

Thanks Jim, this is a very interesting and I'm glad its been proven in your tests.

I've now filed it in my working method files.

I can understand not wanting to find out if the not fire situation works or not

>>However, I think that it is logical to assume that if a rocket is close to the expected altitude at a specified time, it is most likely pointed up.

I concur on this assumption

 

[Art Upton]

While looking at the 3-Axis Magnetometer, we see this is the data that can be obtained from it during flight.

I could see being able to "guess" at the rockets attitude with this, but I'm sure someone else has thought of a reason I haven't on why I can't determine the attitude of the rocket with this data.

 

[Adrian A]

If the MEMs gyros you can get at Sparkfun had a low drift rate, a large enough dynamic range to accommodate typical fast rolls, and weren't so affected by acceleration, then a straight 3-axis gyro solution would be the best. You wouldn't need any accelerometers at all, because you would directly measure how far the rocket rotated, and in which direction, since it was sitting on the pad. The data that I've seen from other's setups, though, shows that the drift rates are really big, and worse yet, it gets changed by the launch event. There are certainly gyros capable of this task, but they cost at least 10's of K$, weigh several pounds, and are about the size of a softball. Do do this job right with gyros (detect single-digit degrees of pitch/yaw), you have pretty much the same requirements as active guided missles.

Two accelerometers spaced a known distance apart could function as a coarse spin rate detector but I believe that the drift of the zero point would prevent sufficient accuracy in angular propagation. I haven't worked out any numbers on that, but from what I have seen of offset drift in the accels I use in my own altimeters, I don't see this as promising.

3-axis magnetic sensors could be promising, since they wouldn't be sensitive to propagation errors and shouldn't be sensitive to G-loading. Where I live, the Earth's magnetic field lines are about 30 degrees from vertical. (I found it quite surprising that they aren't closer to horizontal) By averaging the angle between the rocket axis and the field lines as the rocket spins, you could pretty accurately measure how much the rocket has pitched toward or away from North. But a 20 degree pitch toward the E or W might have the same measurement effect as 10 degrees toward the North. If a 50%+ error in the tilt angle is o.k., because you just want to make sure it's mostly vertical, then this might be o.k. The 3-axis unit didn't look particularly small, however, and I'm pretty sure nobody has actually developed this yet.

So my recommendation would be to first fly the identical stack but without igniting the sustainer, while carrying an accelerometer and baro sensor. Then use that data (along with some Rocksim analysis) to figure out what flight path angle the rocket actually took, to validate a Rocksim model you would use to calculate the lowest acceptable altitude as a function of time from liftoff, and use that as an in-flight ignition check.

I'm tempted to suggest using an accel with the baro sensor, without any other analysis or test flights, to do the same thing on board (make sure the accel altitude and the baro altitude agree) but I have seen accels drift out of calibration a few percent, even after compensation for temperature. And a few percent is the difference between straight up and a 15-20 degree flight path angle.

 

[bobkrech]

Ok new year,

time to rehash old ideas and see if any new wisdom or technology has solved this yet or not.

Last we left our heros were talking about how to make sure our rockets where straight up and not off course before firing the sustainer.

Back in 2005 several mentioned tilt switches and many recanted how if you fall off the clift it still can't tell you are really going down or up.

2006 had many talk of the MAD and also others talk about MAD issues.

2007 had many point to the new 3-5, and 6 axis accelerometer

Bob mentioned that a 3-axis accelerometer and a compass could be used to tell your attitude. Robert mentioned you should be able to do it with 6-axis somehow.

Many pointed out the new tilt compensated compass, but again it was pointed out anything with tilt-compensataion will work for a robot or sub, but not a rocket in free space.

It's time to bring up our friends at SparkFun again with two items that have me thinking today:

3-Axis Magnetometer
https://www.sparkfun.com/commerce/product_info.php?products_id=244

IMU 6 Degrees of Freedom
https://www.sparkfun.com/commerce/product_info.php?products_id=8191

Pros?
Cons?

Lets Discuss

There's a lot of fun stuff at SparkFun. This Compass Module with Tilt Compensation - OS5000-S navigation system is interesting because it has a high refresh rate. https://www.sparkfun.com/commerce/product_info.php?products_id=8507 The manual is very enlightening. https://www.ocean-server.com/download/OS5000_Compass_Manual.pdf

Sparkfun is getting closer but isn't there yet.

Do we have to limit ourselves to on-board inertial systems?

One idea I had was a low-power laser or very bright collimated light at the pad shining up. On the rocket is a pair of detectors looking down. The detectors would have a reasonable but not too narrow angle of view. If the detectors do not see the light then do not fire the second stage.

The optical method is a bit more complicated than you might think because of background light and signal discrimination issues.

You only need one optical sensor on the rocket but you need to control the FOV (field of view) of the optical sensor on the rocket to a reasonable acceptance cone. For example if you do not want the rocket to stage if it tips over by more than a nominal 15 degrees, you need a simple telescope system with a 30 degree FOV. Provided that the sensor is located close to the pad, if the rocket does not tilt by more than 15 degrees with the 30 degree FOV sensor, the beacon light signal should be detectable by the electronics package on the rocket.

The beacon light signal also needs to have a similar or slightly larger optical FOV projection system to maximize downrage signal strength.

To distinguish the beacon light from sun light glints, you need to modulate it at several KHz to 10's of KHz and employ a agc lock-in amplifier/filtered photodiode sensor on the rocket to lock on to the beacon. It would be analogous to a range safety tone system which eliminates all DC interferences and the use of an AGC permit the maximization of the system gain. As long as the modulated signal is being received, a "tone" is present and a logical high tone signal is generated which could be used to enable the staging electronics.

The beacon needs to be as bright as possible but remain eyesafe and this is a challenge. For a 1 watt laser to be eyesafe, you need to expand the beam to a minimum area of ~1000 cm2 or ~ 1 ft square to be eyesafe and maintain a ~30 degree conical FOV.

So how much range could you get from this type of optical system? It really depends on your source irradiance and your sensor detection limit. If you throw out 1 watt of laser light in a 30 degree cone, and you have a detection sensitivity of a nanowatt /cm2 you only get a 2000' range. Increase the sensitivity to 100 picowatt/cm2 and you get a bit more than a 1 mile range. If you get a sensivity of 10 picowatt/cm2 you extend the range to just under 4 miles and if you can push the limit to 1 picowatt/cm2 you get almost 12 miles range.

Bob

 

[jderimig]

So how much range could you get from this type of optical system? It really depends on your source irradiance and your sensor detection limit. If you throw out 1 watt of laser light in a 30 degree cone, and you have a detection sensitivity of a nanowatt /cm2 you only get a 2000' range. Increase the sensitivity to 100 picowatt/cm2 and you get a bit more than a 1 mile range. If you get a sensivity of 10 picowatt/cm2 you extend the range to just under 4 miles and if you can push the limit to 1 picowatt/cm2 you get almost 12 miles range.

Bob

Bob,

Could you pulse the beacon at a low duty cycle and increase the pulse intensity to make it eye safe?

 

[jderimig]

Art,

Another idea I think you would like.......

If the rocket is long enough how about a RF transmitter on the pad and a doppler RDF system in the rocket similar to the N0GSG doppler which is popular with some folks. An antenna at the base and the nose and use the doppler shift to deduce the rocket tilt? Eh?

 

[overby]

Last we left our heros were talking about how to make sure our rockets where straight up and not off course before firing the sustainer.

I don't believe the entire six degrees of freedom is required. Read this application note:

https://www.freescale.com/files/sensors/doc/app_note/AN3461.pdf

Yes, I've been told that this will not work, but I was only provided "proof by intimidation". The proof was not explained. My flight of the Logomatic with a "5DOF" sensor gave me enough information to convince me that a 2 axis accelerometer with horizontal orientation is adequate to show me if I'm far enough off vertical that I need to inhibit staging. True, it does not tell me which way the pointy-end is pointing (north-south-east-west) but I don't care what direction the nose is pointing, I just care how far off of vertical it is.

The above app note describes why the resolution drops off at 45 degrees. IMO, if I inhibit staging at 20 or 30 degrees, that won't be a problem.

Glen OVerby

 

[Tolppisouth]

I'm a little out of my league with the bright folks commenting here but I'll venture another possible solution.

Why not use IR or NIR imaging sensors on let's say the fin tips and filter for a particular bandpass and compare the voltages for a go or no go. If, let us say, a lack of a certain color due to absorbtion by the ground is keyed upon then a sudden rise in voltage on one or more sensors would be indicative of the rocket tipping. I'm thinking of situations where a rocket is launched from a field of reasonably uniform coloration like a field of grass or desert.

Just a thought. The photodiodes and filters should be relatively cheap. Just hitting on a usable combo and algorithm would be a project.


John

 

[Art Upton]

I don't believe the entire six degrees of freedom is required. Read this application note:

https://www.freescale.com/files/sensors/doc/app_note/AN3461.pdf

Yes, I've been told that this will not work, but I was only provided "proof by intimidation". The proof was not explained. My flight of the Logomatic with a "5DOF" sensor gave me enough information to convince me that a 2 axis accelerometer with horizontal orientation is adequate to show me if I'm far enough off vertical that I need to inhibit staging. True, it does not tell me which way the pointy-end is pointing (north-south-east-west) but I don't care what direction the nose is pointing, I just care how far off of vertical it is.

The above app note describes why the resolution drops off at 45 degrees. IMO, if I inhibit staging at 20 or 30 degrees, that won't be a problem.

Glen OVerby

That and interesting test Glen.

Have you done a few more flights?

Can you share the data?

 

[Art Upton]

Art,

Another idea I think you would like.......

If the rocket is long enough how about a RF transmitter on the pad and a doppler RDF system in the rocket similar to the N0GSG doppler which is popular with some folks. An antenna at the base and the nose and use the doppler shift to deduce the rocket tilt? Eh?

Nice teaser John;

Scratches chin, drinks a beer....

A base transmitter and transponder....Humm.......

 

[Art Upton]

.... Where I live, the Earth's magnetic field lines are about 30 degrees from vertical. (I found it quite surprising that they aren't closer to horizontal) By averaging the angle between the rocket axis and the field lines as the rocket spins, you could pretty accurately measure how much the rocket has pitched toward or away from North. But a 20 degree pitch toward the E or W might have the same measurement effect as 10 degrees toward the North. If a 50%+ error in the tilt angle is o.k., ....

Hi Adrain, I think that is the reason a single MAD would not work everywhere.
The averaging concept to remove spin is an idea. but the killer again is the 20 degree vs 10 degree issue.

10 degrees OK, 20 degrees marginal.

 

[Art Upton]

There's a lot of fun stuff at SparkFun...

Hi Bob,

any hope the IMUs and MEMs can make a good vertical stability guidence platform?

 

[Art Upton]

I'm a little out of my league with the bright folks commenting here but I'll venture another possible solution.

Why not use IR or NIR imaging sensors on let's say the fin tips and filter for a particular bandpass and compare the voltages for a go or no go. If, let us say, a lack of a certain color due to absorbtion by the ground is keyed upon then a sudden rise in voltage on one or more sensors would be indicative of the rocket tipping. I'm thinking of situations where a rocket is launched from a field of reasonably uniform coloration like a field of grass or desert.

Just a thought. The photodiodes and filters should be relatively cheap. Just hitting on a usable combo and algorithm would be a project.


John

Hi John,

That idea does work with units like the FMS co-pilot RC airplane leveler. But the issue with the rocket is the spin. As the rocket spins the system needs to know what and which way is level with referance to the IR LEDS.

Or, we build a non-spin rocket with a computer/gyro/fin device that eliminates spin, then the self leveler can work.

After we stage, we can induce spin again to help stabilise the rocket in a thin air environment.

 

[jderimig]

Nice teaser John;

Scratches chin, drinks a beer....

A base transmitter and transponder....Humm.......

No, alot simpler than that.

You have two antennas about a 1/4 wavelength apart, easy to do on rocket at 430 Mhz right?

Both antennas are fed into a mixer. You electrically rotate (switch) the antennas at some FM band frequency, lets say 108 Mhz. If one antenna is farther away from the transmitter than the other (like when the rocket is vertical) you will get a phase modulated signal (because of the arrival time difference between the antennas) that an ordinary fm receiver will pick up. If the antennas are the same distance from the transmitter (like when the rocket is perpendicular to the ground there will be no phase difference and you will get a null.

In between you will get a signal level between 0 and the vertical position. This signal will tell you what the orientation is of the rocket.

 

[Art Upton]

.... You electrically rotate (switch) the antennas at some FM band frequency, lets say 108 Mhz. If one antenna is farther away from the transmitter than the other (like when the rocket is vertical) you will get a phase modulated signal (because of the arrival time difference between the antennas) that an ordinary fm receiver will pick up. ..

Yes easy and clearly understood, WOW.

Lets keep going on all ideas, this brain storm list is filling up a nice mind map

 

[FROB]

No, alot simpler than that.
You have two antennas about a 1/4 wavelength apart, easy to do on rocket at 430 Mhz right?...
...In between you will get a signal level between 0 and the vertical position. This signal will tell you what the orientation is of the rocket.

Thats brilliant! why didnt i think of that? its similar to the old aircraft directional radio beacons. should be fairly straightforward to try at any rate.

But as long as we're competing for simplicity, i suspect the easiest is the method often used in low end RC helicopter-like toys that i've seen recently- using 4 cheap passive IR cells, (same as those used in security PIR motion sensors ) arranged around the periphery and pointed horizontally. You subtract the signal from opposite pairs (a simple op-amp circuit) to get an signal proportional to the angle relative to horizon for both X and Y.
It seems at deep IR the Earth is pretty uniformly "hot" and the sky is generally "cold", at least within the range of interest near the horizon. (you can thank the abundance of CO2 and the greenhouse effect for that so i'm told) so when the one side is tlited up, the other down, the difference... you get the idea.
I suspect this would have issues with the sun, especially near dusk or dawn, so its not perfect, but good enough for those toys to depend on it.
It's a known and proven technique, with a little refinement and testing it could be accurate enough for this purpose, under most common circumstances anyway.

Personally i'll be toying with these angular rate sensors this winter:
https://www.melexis.com/Sensor_ICs_Inertia/Sensor_ICs_Inertia/Angular_Rate_Sensor_582.aspx
for both spin-nulling use and staging safety. Maybe active stability later. If that proves dissapointing or if time premits, i may try this PIR method. I think I still have a box of leftover bits and pieces of a commercial PIR unit i develloped many years ago in my basement somewhere, if i didnt toss it last time i moved that is. I'll report my findings here of course.

 

[overby]

That and interesting test Glen.

Have you done a few more flights?

Can you share the data?

Thanks, it was fun! I haven't had the opportunity to make more flights. That rocket flew at the last club launch of the year. I've been flying it in a BT-70 rocket that probably won't come in under 1lb with a motor that will lift it a decent height, and I haven't tried it in a BT-60 since I added devices to it (I don't think it will fit). I don't have a good winter launch site yet - I've been waiting for the ice on the lakes to get to a safe thickness. At least I assume I won't get arrested for flying rockets on a lake in MN?

I bought a faster SD card, a SanDisk "Ultra II", that allows me to raise the data rate to 400 samples per second for 6 channels, or 1500 samples per second for one channel.

Glen

 

[Peartree]

Maybe this wouldn't work but since there seems to be some free associationgoing on...

Could you "hang" an IR emitter on a free (aka frickionless) gimbal inside of a section of rocket (perhaps in a smaller internal tube or electronics bay) pointed "down" toward a reciever? If the rocket went too far off vertical the emitter would point toward the ground and away from the reciever. Would the g-forces at staging be sufficient to point the emitter on the gimbal toward the bottom of the rocket or would it point toward the ground?

 

[jderimig]

Maybe this wouldn't work but since there seems to be some free associationgoing on...

Could you "hang" an IR emitter on a free (aka frickionless) gimbal inside of a section of rocket (perhaps in a smaller internal tube or electronics bay) pointed "down" toward a reciever? If the rocket went too far off vertical the emitter would point toward the ground and away from the reciever. Would the g-forces at staging be sufficient to point the emitter on the gimbal toward the bottom of the rocket or would it point toward the ground?

During ballistic trajectory there is no down. Ie..the body is "weightless", the gimbal can point anywhere.

 

[bobkrech]

Bob,

Could you pulse the beacon at a low duty cycle and increase the pulse intensity to make it eye safe?

Eye safe is ~2 mw/cm2. It's not that difficult to make a 1 watt laser eye safe by increasing the area of the beam.

You really want a 50% duty cycle to maximize detection sensivity.

I don't believe the entire six degrees of freedom is required. Read this application note:

https://www.freescale.com/files/sensors/doc/app_note/AN3461.pdf

Yes, I've been told that this will not work, but I was only provided "proof by intimidation". The proof was not explained. My flight of the Logomatic with a "5DOF" sensor gave me enough information to convince me that a 2 axis accelerometer with horizontal orientation is adequate to show me if I'm far enough off vertical that I need to inhibit staging. True, it does not tell me which way the pointy-end is pointing (north-south-east-west) but I don't care what direction the nose is pointing, I just care how far off of vertical it is.

The above app note describes why the resolution drops off at 45 degrees. IMO, if I inhibit staging at 20 or 30 degrees, that won't be a problem.

Glen OVerby

The solution works if the object is not moving and the only acceleration is gravity. On page 3 of the app note, it states that the sensivitity to tilt perpendicular to gravity is 17.45 mv/degree. If you want to trigger an event if the tilt exceeds 20 degree, all you have to do set +/- 349 mv trigger level and you are fine if your reference is gravity = 1g.

The problem occurs when you launch the rocket. Say the rocket accelerates at 30 G. If you rocket tilts a mere 1 degree, the signal generated is ~17.45 mv/degree * 30 = 523.5 mv which exceeds the 1 G threshold of 349 mv for a 20 degree tilt!

I'm a little out of my league with the bright folks commenting here but I'll venture another possible solution.

Why not use IR or NIR imaging sensors on let's say the fin tips and filter for a particular bandpass and compare the voltages for a go or no go. If, let us say, a lack of a certain color due to absorbtion by the ground is keyed upon then a sudden rise in voltage on one or more sensors would be indicative of the rocket tipping. I'm thinking of situations where a rocket is launched from a field of reasonably uniform coloration like a field of grass or desert.

Just a thought. The photodiodes and filters should be relatively cheap. Just hitting on a usable combo and algorithm would be a project.

John

You don't mean true IR imaging sensors. They are not affordable for amateur efforts and the image processing requirements are too computationally intensive, but IR is the way to go.

Futaba Pilot Assist PA-1 and PA-2 modules have been used to keep rockets vertical by gimballing the motor. https://www.futaba-rc.com/radioaccys/futm0999.html https://www3.towerhobbies.com/cgi-bin/wti0001p?&I=LXNA35&P=7

You might want to see how it's done even better with inexpensive IR sensors. Check out https://www.eganfamily.id.au/archive30nov2007/monash/research/papers/TaylorBristol2003.pdf for an IR based horizon detector.

The most promising off the shelf IR sensor package is the FMA Co-Pilot CPD4 Infrared Flight Stabilization System. https://fmadirect.com/products.htm?cat=20&nid=6 https://fmadirect.com/support_docs/item_1039.pdf and https://www.fmadirect.com/detail.htm?item=1489&section=20 It should work quite well as IR FMA Co-Pilot is not as susceptible to solar glint interference as the Futaba PA-2.

Also of interest https://www.isr.umd.edu/Labs/CSSL/horiuchilab/projects/horizon/horizonchip.html and https://www.mil.ufl.edu/publications/reu2000/SADTU.doc

Bob