I could use just a little guidance

The Rocketry Forum

Help Support The Rocketry Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Feel free to post additional pics and descriptions of what you're doing. Looks like 6 canards?

Jim

Thanks. The two upper pairs are for pitch and yaw. Each pair are connected via a shaft riding on ball bearings and operated by one servo. The lower two are for roll where each canard is operated by its own servo.
 
Unless I am missing something it seems to me that the pitch and yaw controls do need to be very fast in order to properly correct pitch or yaw when the rocket is rolling really fast. As a rocket rolls, what the system considered a pitch error will become a yaw error and then back to a pitch error again and so on. Consequently, if control fins are trying to pitch the rocket up then those control fins need to go back to a neutral position as the rocket rolls 90 degrees otherwise they will be yawing it right or left rather than pitching it up. As the rocket rolls another 90 degrees those control fins then need to swing in the opposite direction to pitch it up and so on. I have a mental image of the control fins swinging back and forth as the rocket rolls. The speed they need to move back and forth depends on the roll rate. (I think.)
Fast and slow are relative terms. In my world, the microsecond is normal speed......so a millisecond is slow......

Let's looks at the RC class timing and the roll and pitch rates of a rocket and see what it takes to control them.

In a RC electronics timing cycle, the basic time unit is 0.002 seconds which is the period of the servo control loop. It means you can command changes in servo positions 500 times per seconds. Servo rate response times vary from 60 degrees motion in 0.05 seconds (fast) to 0.2 seconds (very slow) with an nominal 0.1 second time so the nominal angular velocity correction rate is 600 degrees per second per second.

The nominal pitch rate of an aerodynamically stable rocket is slow, so lets use a maximum of angular pitch or yaw velocity of 10 degrees per second and a possible angular acceleration of 30 degrees per second per second which might be expected for a flight with an apogee time of 10 seconds and includes initial weather cocking and a gravity turn. Rockets roll faster and 2 to 3 rps is not uncommon, however it usually takes a few seconds to get that rate so for the purposes of discussion lets consider and angular velocity of 1000 degrees per second and an angular acceleration of 360 degrees per second per second.

So what do these numbers mean. For pitch control, a 600 degree per second per second servo acceleration rate far exceeds what is required for pitch control, and is more than adequate for typical roll control. It should be obvious that the ETG can easily control a rocket if the control mode is for pitch and yaw. With a maximum expected 30 degree per second per second pitch or yaw acceleration rate, the nominal servo should control this in 0.05 seconds or for disturbances to nominally 20 Hz.

While the required pitch corrections are due to wind and gravity which are time variant, the required roll control is due to fin misalignment and the dynamics of the required rocket dynamic for pitch correction. The fin misalignment is simply and offset correction and is quickly corrected and the if there is only pitch control and no yaw control, one has to rotate the rocket to correct yaw. It will talk not more than 90/600 = 0.15 seconds to roll the rocket 90 degrees. During that time the maximum change in pitch or yaw could be 30 * 0.15 = 4.5 degrees well within the capabilities of the system. It is a non-issue if you choose to use a pitch-yaw correction loop versus pitch-roll correction loop, however the former does not control roll whereas the latter does.

Both of the above control schemes require only 2 servos for either pitch and roll or pitch and yaw corrections, and Alyssa demonstrated this control scheme works for rockets in her research presentation. If you use 4 servos, the ETG can control roll, pitch and yaw simultaneously in an aircraft, however it is not clear to me from the ETG instruction sheet that it can be configured to do for a rocket without additional electronics. This would be the preferred route for rockets with video feeds.

Bob
 
Last edited:
A lot of sounding rockets have been launched to very high altitudes at White Sands Missile Range without active guidance. You might look into what they did there to minimize dispersion.
 
This is very interesting and I am looking forward to your further developments. Thanks for your contribution.
 
All unguided sounding rockets launched from any military or civilian test range are required to be spin stabilized and are launched at least 5 degrees off vertical to minimize dispersion and define the splash zone.

Bob
 
I just found this thread tonight.

For those who may not have followed the Eagle Tree Guardian thread, go here:

https://www.rocketryforum.com/showt...automated-rocket-guidance-Eagle-Tree-Guardian

A key difference from the big sounding rockets is Moment of Inertia. For example, a sounding rocket the size and mass of a telephone pole. If you could simply stand a phone pole vertically, then let go, it would fall over. But…. it would slowly fall over. Or, think of a tree being cut down, how long that it takes. As opposed to standing a Baseball Bat up vertically and letting go, it would fall over quickly. And then a pencil, REAL fast.

So it is the moment of inertia. The mass concentrations and relative length. It's a LOT easier for a 10 pound bowling ball to rotate than it would be for a 2 pound 10 foot long pole with 4 pound weights at each end (Same mass, totally different moments of inertia). And a LOT easier to rotate a soccer ball than to rotate a 10 pound bowling ball (approximately same size, way different mass, totally different moments of inertia).

Even when sounding rockets take off and have fins canted slightly (maybe a degree or two) to cause a slow roll, it will take a bit of time to start to roll much. So it is the moment of inertia that keeps it pointed the right way till the roll begins to become enough to help even out anything about the rocket itself that might make it begin to pitch or yaw (such as thrust not 100% symmetric due to a nozzle flaw).

Also some of the smaller sounding rockets take off incredibly fast, like the old Loki rockets (And Lokis were designed to be launched from a spiral-railed tower to make them spin at liftoff, not unlike the rifling in a gun barrel).

Some sounding rockets do use a guidance system. Some Black Brant VC's and Black Brant XII's for example.

HK5ANN9.jpg


I googled this: NASA Sounding Rocket System Handbook.

Start with Page 32, which has some info on the guidance unit used on those Black Brants.

https://sites.wff.nasa.gov/code810/files/SRHB.pdf

Here is a pic from it:
sX8cUy0.jpg


For roll control, you will not need very big surfaces, and they ought not to move very much or they could easily over-control. I point this out because you will not need nearly the sort of aerodynamic force to correct roll as will be needed to control pitch or yaw.

Fast and slow are relative terms. In my world, the microsecond is normal speed......so a millisecond is slow......

Let's looks at the RC class timing and the roll and pitch rates of a rocket and see what it takes to control them.

In a RC electronics timing cycle, the basic time unit is 0.002 seconds which is the period of the servo control loop. It means you can command changes in servo positions 500 times per seconds. Servo rate response times vary from 60 degrees motion in 0.05 seconds (fast) to 0.2 seconds (very slow) with an nominal 0.1 second time so the nominal angular velocity correction rate is 600 degrees per second per second.

A key thing Bob points out there is the response rate of the servos. Need the right servos for the job, not only to have enough torque, but to be fast enough. Now, for R/C servos, those two are opposing…. they get more torque by gearing them down, but then they are slower. Or, to make them faster, they gear them up, producing less torque. So a servo that has a motor strong enough to be able to be geared to be FAST and still with a lot of torque tends to be a significantly bigger and heavier servo (and more $). So, have to really check the fine print of the data for al the servos being considered and go for whatever is both fast and powerful. But if a tradeoff has to be made… .better to err on the side of POWERFUL, better to be a little slow than to not move the control surface at all.

If the roll control has smaller surfaces and does not have a have a lot of motion, so it does not need a lot of torque, then faster would be more of a priority.

BTW - the time unit of .002 second (2.0 milliseconds) is not quite right in the context of what was described as a "cycle". For most servos, the control pulse width varies from 1.0 to 2.0 milliseconds, with 1.5 milliseconds usually as neutral/center. But the cycle time between the start of each pulse to a specific channel is 20 milliseconds (so, if the signal pulse is 1.50 ms for neutral, it will be 18.5 ms to the start of the next control pulse). One reason for waiting so long is that for multiple channels, the other channel pulses are occurring in rapid succession, then the Pulse Train starts over again 20 ms after the previous pulse train began. So, the time between pulse cycles per servo channel is 20 ms, or 50 times a second. But that is fast enough.

After typing the above, I quickly googled a drawing to demonstrate.

- George Gassaway

servo-control.png
 
Last edited:
For servos you should try to find servos that will run up to 7.4 volts and run a 2 cell lipo, this will typically get you more power in a given package. Most servos are rated at 4.8v-6.0v and may or may not burn out with 8v from a fully charged 2 cell. Also you should make sure you have metal gear servos with ball bearings. I'm the last couple of years they've come out with servos for robotics with aluminum cases, brushless motors, and metal gears. They're small powerful and built to take abuse.
 
For servos you should try to find servos that will run up to 7.4 volts and run a 2 cell lipo, this will typically get you more power in a given package. Most servos are rated at 4.8v-6.0v and may or may not burn out with 8v from a fully charged 2 cell. Also you should make sure you have metal gear servos with ball bearings. I'm the last couple of years they've come out with servos for robotics with aluminum cases, brushless motors, and metal gears. They're small powerful and built to take abuse.

What I got (a while back now), was the Hitec 5245 servos. They are not quite full size because I wanted to be able to mount them in either 4" or 3" rockets (per the pics). They do have reasonable torque (76 oz-in) and are reasonably quick (0.12 sec). They are ball bearing with metal gears (except for one), and they don't cost a fortune ($40). They also have a form factor that fits my design (many don't). I also happen to have a NiCad pack which is good for the digital servos, so I just designed around that. The system doesn't have to operate for very long. I'm hoping that these are OK (with one servo for each canard). Thanks for the advice though. Servos have certainly changed a lot since the last time I messed with them.

Jim

Servos 4.png

Servo 3.png
 
Where do you plan on using the active stabilization, booster fins, fins on the upper stage like forward canards, some added forward canards, or in multiple areas. I was wondering what advantages/disadvantages each scenario would have.

My objectives for active stabilization are probably different from most folks. I'm looking to use this on multistage flights, and what I think I need to accomplish is to have the sustainer vertical at the point where it ignites. If the rocket is a couple miles out at that point, no problem, and if it's vertical at that point, the dispersion should be cut in half, roughly, from what I would otherwise get. I'm hoping that will be enough for flight approval from the powers-that-be.

I really wanted to use fin tabs. To implement that, I might have kept the stages together so that the rocket would go vertical during the coast. But then, leaving the stabilization behind after that would be fine. However, since I'm primarily interested in minimum diameter, the control linkages would have been long (with servos located well away from the fins). I had some designs, but I doubt they would have worked well enough.

With the canards, a sustainer could be separated from the booster at burnout. I think it will be easier to stabilize a sustainer only (versus a long stack) so that would be an advantage for the canards. Another advantage of them would be to apply stabilization through the whole flight, although that's not really my objective. I'd really like the canards to provide stabilization during the coast and then go away to avoid the drag from that point on. Random Flying Object has some retractable rail buttons ..... hmmm?

Jim
 
Here's a bit more detail on my design. Comments are welcome. The layout of the equipment is in the first pic. The inerds of the device, on the right, include a piece of 3/8" allthread and some bulkheads for mounting the electrical equipment and the servos. This "open" design should make it easier to assemble.

The allthread section is then inserted into a case that consists of a piece of airframe and a coupler with a bulkhead mounted at the bottom (this coupler then goes into the top of the upper airframe). The allthread section would attach at the bottom bulkhead and at the plywood bulkhead holding the servos. The little wheel on the tops of the servos would be attached after installing the allthread section, and then the canards would be screwed onto the wheel. I'd like a close match between the size of the wheel and the hole in the airframe to help provide support for the canard. The servo wires would be "Y'd", and one lead from the Y would exit the top of the case so that I can center and then define the end points and the fail safe point after attaching the canards. After that, the nosecone would just be screwed down onto the top of the allthread. Thus, the nosecone/case assembly would stay together through the flight.

The circuit design is in the second pic. As I have stated before, the Guardian doesn't need to radio receiver to operate. I though it would require a 1.5 ms pulse to recognize "hands off the sticks", but it does not need that. The output from the Guardian (signal wires only) goes to the solid state switches so that I can control when the stabilization system is on using the Raven (if I want to do that). I have three switches in case I want to add roll control. For 2D control (to start with), I need to pairs of servos, which is how I've shown the diagram. The servos are powered separately from the Guardian due to the higher current that the digital servos could need. I have two switches - one that powers up all of the equipment and a second that will let me test the servos without having to "trigger" the Raven.

Jim

Layout.jpg

Circuit.png
 
I have made a bit more progress on the servo/canard section. Up next is the airframe/coupler pieces.

Jim

DSCF1004.jpg

DSCF1005.jpg

DSCF1006.jpg

DSCF1008.jpg

DSCF1009.jpg
 
I had some time to work on the vertical stabilization system this weekend (launches canceled due to rain). Here's a "progress video".

Jim

[video=youtube_share;i7OqOj7GHuc]https://youtu.be/i7OqOj7GHuc[/video]
 
The system looks really nice.

Two comments. The AA battery holder concerns me. Very easy for acceleration loads (or vibration) to cause a momentary power loss, which would cause the Guardian to reset far too late into the flight (and no longer calibrated to know where vertical is) to be useful. Worst yet, the servos might have moved to a non-centered position when that happened, and spend seconds "locked" non-neutral to make it veer off course (I have been there, done that. Made a Sunguidance flight where the battery went dead before liftoff, and with the control surfaces locked in the position to correct it towards the sun before the battery went dead, it looped into the ground).

I hope that the AA pack is for convenience of bench testing, but will not assume that, and that you'll use some secure soldered-up battery packs like Nicads or LiPolys.

If you use LiPos, take note that two LiPos produce 7.4V, which will fry most servos, which usually are rated for 4.8-6.0 V (I've accidentally fried at least two). What I do for my R/C models with that problem (when not using servos rated for 8 volts) is simply use a 7805 voltage regulator, as i'm OK with running the servos at 5V. A 7806 might be better for the servo speed/torque

Now that I see how you have the servos set to go to center when the sustainer takes over, this occurred to me. In case there was a small unmeasured error of a fraction of a degree, in two opposing servos, which would cause a little bit of a pitch or a yaw. You could consider making two opposing servos set to be about 1/4 degree, or 1/32" , or whatever small measurement you can consistently make, to cause a slow roll for the unguided sustainer part of the flight. So, that would even out any unmeasureable error.

- George Gassaway
 
Last edited:
The system looks really nice.

Two comments. The AA battery holder concerns me.

You could consider making two opposing servos set to be about 1/4 degree, or 1/32" , or whatever small measurement you can consistently make, to cause a slow roll for the unguided sustainer part of the flight. So, that would even out any unmeasureable error.

- George Gassaway

Thanks George. Yes, the AA pack is just for convenience. I have a 1700 mah NiCad pack that I'm planning to use. However, I only have a 50 ma charger at the moment, and I've read that using what is essentially a trickle charger isn't recommended. I just need to drop by the hobby shop and get something a little better.

The purpose of the NiCad's is to provide the current that could be required for the servos. Hitec indicates 2 amps is a reasonable assumption for a servo under high torque. So, 8 amps total. Since the Guardian can't pass more than 5 amps, I am providing power directly to the servos.

Interesting idea about arranging for a bit of roll just to eliminate the effects of misalignment. That would probably work, although I have a concern about going too far here or using the canards to intentionally control roll. My concern is that the canards will be located well away from the CG. I would think that trying to induce or cancel roll would lead to coning?

You comment, though, is a good segway into the topic of my test rocket. I need to build something that I can use for testing, and I'm starting to think about what I want it to be able to do. Just as examples, would I want to be able to video the action of the canards? Do I want to be able to vary the CG/CP relationship? Is it possible to record the servo position or the PWM signal from the Guardian? I do plan to fly an EasyMega during the test flights which will provide acceleration and roll data. But, other ideas of what can or should be tested would be appreciated.

Jim
 
If you use LiPos, take note that two LiPos produce 7.4V, which will fry most servos, which usually are rated for 4.8-6.0 V (I've accidentally fried at least two). What I do for my R/C models with that problem (when not using servos rated for 8 volts) is simply use a 7805 voltage regulator, as i'm OK with running the servos at 5V. A 7806 might be better for the servo speed/torque

The 7805 and 7806 are a sub-optimal choice here because they have a 2V drop out voltage. The input must be at least 2V greater than the output for them to work.

The current rating is bit anaemic for all but the smallest servos. Something with lower dropout and higher current would be better and there are lots of choices. The LM1084 for example has a maximum dropout voltage of 1.5V at 5A. But it is better than that because the dropout voltage gets lower with increasing temperature.
 
My concern is that the canards will be located well away from the CG. I would think that trying to induce or cancel roll would lead to coning?

I don't believe that there is any connection. The ATACMS has its control fins at the back end and it flies roll stabilized with a spin up just prior to dispense. During that spin up the CG is somewhere in the payload section a long way from the fins.

The ATACMS Block II does sometimes do controlled coning to dissipate energy so that can be done if desired.

The editor insists on changing the URL I wanted to include to an embedded video so I will move it to the bottom:

[video=youtube;Ipr_hPAcR_Q]https://www.youtube.com/watch?v=Ipr_hPAcR_Q[/video]
 
Although this only addresses the stabilization of roll, I thought it was pretty cool. The rollerons on the AIM-9 sidewinder are air driven gyroscope ailerons that are fairly simple in design ( simple meaning you are not me and know mathematics ) that were tested in an amature rocket here:

https://psas.pdx.edu/news/1999-09-13/
 
Although this only addresses the stabilization of roll, I thought it was pretty cool. The rollerons on the AIM-9 sidewinder are air driven gyroscope ailerons that are fairly simple in design ( simple meaning you are not me and know mathematics ) that were tested in an amature rocket here:

https://psas.pdx.edu/news/1999-09-13/

I saw that concept too. Pretty interesting.

Jim
 
I don't believe that there is any connection. The ATACMS has its control fins at the back end and it flies roll stabilized with a spin up just prior to dispense. During that spin up the CG is somewhere in the payload section a long way from the fins.

Now, THAT is a cool video. Thanks David. So, perhaps roll control could work away from the CG. Might have to give that a try.

Jim
 
You comment, though, is a good segway into the topic of my test rocket. I need to build something that I can use for testing, and I'm starting to think about what I want it to be able to do. Just as examples, would I want to be able to video the action of the canards? Do I want to be able to vary the CG/CP relationship? Is it possible to record the servo position or the PWM signal from the Guardian? I do plan to fly an EasyMega during the test flights which will provide acceleration and roll data. But, other ideas of what can or should be tested would be appreciated.

I think that onboard video is the most useful way to document how well the system works. In my original 1988 Sunguidance R&D, there were 5 flights that used a Cineroc to show what the system did. Two of those were roll-only flights, which showed problems that were not visible at all from ground observation. And three pitch/yaw flights which showed the system working well to control it.

BTW - onboard video of a rocket with guidance is just plain COOL!

In the last 10 years or so, I have done some that used an onboard video camera, a Gearcam transmitting video. Also a few Keychain cameras, which are more practical (lighter & simpler than the gearcam and no ground equipment to set up and record with).

See the video linked below. (You will probably need to download it), which has several Sunguidance flights on it, most of it old, but a few newer. Some are onboard such as the Cineroc and Gearcams.

https://georgesrockets.com/video/VidFiles/Sunguidance_Web.mov

Here is a a web page documenting a sunguidance flight that was overcontrolled. The general flight path was fine, but the rocket quickly wiggled back and forth:

https://georgesrockets.com/video/Airvid/Video_SEARS_Mar05.htm

Frames from that video are in my avatar.

SunTesTAnimation3.gif


I had used light sensors that were pretty much parallel to the body tube. That maximized the sensitivity, but it was too much. Later I changed to an angled nose cone which provided a "softer" sensitivity near neutral, so the rocket flew much smoother but still had plenty of control authority.

sWsjb3F.jpg


So, anyway, you will be able to see how well it works out. One possible wrinkle your method may have, with one servo per canard rather than one servo moving two canards in the same axis (common drive shaft), would be if one servo does not produce the exact same angular response as the other servo, which would introduce a rolling moment. I ran into that with electronically mixed elevons, long ago. But those were sort of crappy servos, and digital should not be as prone to it… but that does not mean impossible. So, with onboard video, you could look to see if there is some rolling that seems to be in response to a lot of pitch control, or a lot of yaw control. If you really wanted to go slightly nuts with the documentation, use two cameras 180 degrees apart so you could document all four canards and if you are semi-proficient with decent video editing software you could edit them to both be side by side and in sync. That would not be necessary though.

You can make up a simple and easy-enough test model using 2.6" tubing. To me, I think your method for neutralizing the servos after stating looks good and that part probably is not necessary to be tested in smaller scale. Small and reasonably cheap (but not crappy) servos, with the Eagle Tree Guardian. Make the test model to be a downsized model of the big one. Especially with a "lower stage", because that is the configuration it will be in for the actual controlled part of the flight. You would not even need to stage it, the "sustainer" could be considered more as a very special long payload section that has fins at the back and canards up front.

The CG, based that on what the final big rocket CG is likely to be. For the test model, you may need to design it to be able to put ballast in the back of the first stage (Recessed aft centering ring?). CP, you'll want to be sure the fins on the first stage are big enough that the canard fins won't be able to make it go aerodynamically unstable (yes, I know guidance system, but aerodynamically unstable and high speeds is not a good mix for a model plane system to try to keep under control. And "fail safe" means the rocket is stable without this system. not relying on it)

I think the biggest thing to test for with a smaller model is how well does a given size of canard work, at a given deflection angle range, for that configuration? I must say that they look a bit… small. But then the big rocket will be flying a lot faster than the ones I have flown with "Sidewinder" size canards. So the aerodynamic forces will be much greater, so somewhat smaller canards would be more suited. But you don't want them to be too small to be able to be effective.

Moment of inertia is also somewhat of an issue. But that can be a compacted thing to get into, accounting for the difference, as a smaller rocket is easier to control than a larger one. But on the flip side, a smaller one can get into trouble quicker than a big one. If you really wanted to take that into account, then you could work out mathematically. I have never used the feature but a quick Google indicates that Rocksim can calculate the moment of inertia. There is also one heck of a simple way to determine the moment of inertia for a built rocket, by suspension it from a torsion spring and timing the cycle time to pivot back and forth. I'll go into a little more detail on the mechanics setting up that test in case it is of interest.

As well, velocity is an issue.

- George Gassaway
 
Last edited:
This looks AWESOME!

I wish I had more time for side projects, this would be a PERFECT application for 3D Printing. The servo mount could pretty easily be made all in one piece.
 
A five stager George? Seriously?

Looks like I need a couple of keychain cameras. Probably something with a high frame rate?

The canards I made were based on what another fellow is using who is doing much the same project as me. He's using 1.5 square inch fins I believe, and mine are slightly larger. I suspect I will end up going smaller, but that's to be determined. In a typical flight, I would separate the booster shortly after burnout and then coast for perhaps 15 seconds. So, only the sustainer would be guided during that period and there is lots of time to get vertical. Even a minor correction back towards vertical will be an improvement.

Test rocket is going to have to be 4". That's the size of the unit I made. I like the idea of weight in the back of the rocket.

Jim
 
This looks AWESOME!

I wish I had more time for side projects, this would be a PERFECT application for 3D Printing. The servo mount could pretty easily be made all in one piece.

After looking at George's three and five-stage flights, I'm starting to get interested in trying to get something ready for my three-stager at Balls this year (try #2). The sustainer on that is 3", and the canards would need to be pretty small, as there is not much static margin to play with. Not sure I can pull off the testing of a 4" and then construction of a 3" version (the time required for side projects and all). And if there was a printed servo holder on board, I'd probably just lose it anyway. Neat idea though.

Jim

Picture1.jpg
 
I have a rocket that has a high spin rate and I never noticed any problems with recovery.

That's not that fast, and also by the time you hit apogee, it has slowed down a lot. When you simulate a really really high performance rocket like Jim's staged rockets or Bare Necessities (which could definitely have benefited from spin stabilization) with half a degree of fin cant, it ends up spinning extremely quickly, and since the air thins out a lot, it no longer has any friction slowing the spin.
 
Back
Top