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.
I agree with all of the above, I think it will work well. What will be your delay between dropping the first stage(with stabalization) and igniting the second stage?


Sent from my iPad using Rocketry Forum

My simulation of the boost was in Post 91. If I turn on stabilization a few seconds after burnout (say, at 5 seconds, to be determined yet if necessary), and then I want the motor to light at 14 seconds (600 ft/s), then allowing a second for the motor to come up to pressure, and assuming the stabilization unit is separated with the igniter, then stabilization would go from 5 seconds to 13 seconds. That should be plenty of time.

Jim
 
My simulation of the boost was in Post 91. If I turn on stabilization a few seconds after burnout (say, at 5 seconds, to be determined yet if necessary), and then I want the motor to light at 14 seconds (600 ft/s), then allowing a second for the motor to come up to pressure, and assuming the stabilization unit is separated with the igniter, then stabilization would go from 5 seconds to 13 seconds. That should be plenty of time.

Jim

Sounds good. Let me know when and where you plan to do some more testing.


Sent from my iPad using Rocketry Forum
 
Sounds good. Let me know when and where you plan to do some more testing.


Sent from my iPad using Rocketry Forum

Hopefully, the next test will be in a few weeks. For that test, I have fine-tuned the UDB5 mount and I plan to reduce the size of the canards a bit. I'll also add a small camera with a neat little 3D printed mount to look at the canards, and the flight will be on an L1000. After that, we'll install new firm ware to do the elevon & rudderon mixing (proportional control on tilt, rate control for roll on all four canards) with camera pointed down. After that, it will be time to move the stabilization system to the bottom of the rocket.

For the upcoming flights, I need to figure out how to organize the recovery. The top of the rocket is heavy relative to the fins and the rocket doesn't fly very well on drogue (top section straight down, which is not good). I need to do something different, but haven't decided what yet.

Jim
 
I'm a bit confused? If the rocket is vertical with some wind, is there not an AOA in that condition (along with a horizontal component to the flight path)? A vertical flight path would be a higher AOA?

Your question got me thinking about my three-stage strategy of putting stabilization under the second stage motor. I had thought that having the stabilization would allow a lower speed at second stage ignition. But, when stabilization stops and the motor lights, the rocket will weathercock into whatever wind there is, depending on the relative velocities, defeating the purpose of stabilization. I don't think I can let it slow down all that much.

Jim

Just because the rocket is pointed straight up does not mean that its velocity is in that direction.

In any case, vertical might not be the direction you want to go. Based on preflight simulations using the measured wind profile you would set a launcher azimuth and elevation to achieve the desired landing location. The guidance system would then maintain the angle set on the launcher.

Umm, hold on. If the guidance is pre-set on the launcher, at an angle (let's say 10 degrees from vertical, pointing East), then if the rocket rolls 180 degrees, it is going to fly to the WEST at 10 degrees from vertical.

I know Jim is talking abut adding roll control. I am concerned it will over-control in roll, there is a lot of forgiveness in the system for pitch/yaw guidance that is a bit too responsive. But if roll is too responsive, it will over-control, and can easily lead to rolling so far so fast that it locks up and can never stop the roll. I had that happen on my first test of roll control with a sunguidance rocket. It turned out that the vanes were too big, and there was a lot of deflection angle. But then that rocket did not boost very fast either, and aerodynamic forces increase with the square of the velocity.

So, until the behaviour of the roll control system is proven, one cannot assume the rocket would not roll 180 degrees.

Jim, for your roll control, you might consider using only two of the four canards to control roll. I think the rocket is not going to need much deflection angle for roll, compared to pitch/yaw at the velocity your bird will be flying. Let's say for example, one degree plus or minus for al four canards. How well can the servos and mechanical "slop" in the control system handle angular resolution that low? THere's going to be some amount of error or slop. But if you used two canards, and not four, then you could make those rotate twice as much (2 degrees in this theoretical example), so that the extra degree is free from the resolution and slop contained to some extent, in that first degree. If you get what I mean (and i'm sure you do).

However, I do not know if it is practical for the controller's software to control only two for roll and not four for roll. Probably is possible, but may be more complex to find out (though with the designer helping to progrsm it, that should not be as much of an issue.)

Now a wild card in this is how well the Flight Controller handles it. There is term called PID which involves three factors for the feedback and control loop:

https://en.wikipedia.org/wiki/PID_controller

I have barely touched on that myself. But if tuned well it may handle high speeds better without over-control. But there will be some speed beyond which it would over control. It needs far less control for roll than it does for pitch/yaw

Oh, the roll guidance probably should be active from liftoff. I forget if that is planned or not. If the roll guidance was not active until pitch/yaw was active, the rocket would probably already be rolling a LOT. It would take several precious seconds to stop the roll. Now, Pitch/Yaw guidance does work when the rocket is rolling, but there is a certain amount of roll (degrees per second) beyond which the system (mostly servo speed) can not keep up.

BTW - what UhClem said abut angling the rocket for launch, I hep he has misunderstood something. The procedure for a guided rocket like this should be something like:

Rocket on pad, everything prepared for flight except for guidance (and probably not arm ejection or staging system yet). Angle rocket vertically. Turn on guidance system so it can initialize and know where vertical and horizontal are. THEN angle the launcher however many degrees, say 10, in the desired direction for an impact zone if it goes ballistic. And do any final arming of anything else, get way, and hopefully the bird will not wait long time wasting batteries before it is launched (Generic safety issue, no rocket depending on electronics for a safe flight should be forced to wait a long time, but also the battery capacity onboard should be overkill and freshly charged or new).

As for angle of attack, AOA, a vertical rocket that is pointing vertically and is thrusting vertically, should have effectively zero angle of attack. Now if it took off in a 10 mph wind and was trying to steer vertical from liftoff, sure, at first it would weathercock some until the velocity was enough for the controls to be effective. It would begin to pitch up to vertical. By the time it would then become vertical, it is NOT flying a vertical LINE up at 90 degrees (unless there was no wind).

It is pointed straight up, flying vertical through the AIR...... so the vertical model and the air mass are moving together downwind at 10 mph (OK technically the winds tend to be higher at higher altitude, but this is for example). Think of people riding in a hot air balloon. The wind is 10 mph. But shortly after the balloon is in the air, drifting at 10 mph, the people in the gondola do not feel any wind at all. To them, there is no wind. This is ignoring wind gusts and the balloon moving up or down. Another classic example would be an aircraft carrier sailing downwind at 10 mph, to the sailors on deck.... it feels calm (again an ideal 10 mph wind, ignoring gusts and such)

So the path of motion is sort of a steep diagonal line, whose angle is relative to the windspeed versus rocket speed. I say sort of since the faster the rocket gets relative to the wind, the more towards vertical the line would become.

I have seen some vertical guidance flights in wind, models with long burn (8 sec) low thrust engines flying at probably 50-80 mph, and that is exactly what they look like, model vertical, with model and smoke under it vertical.... drifting downwind on a flight path that to a ground observer looks steeply diagonal, but to a passenger in a nearby (not too close) hot air balloon it would look vertical.

So, Jim, that approximately 8 degree off-vertical path. Was it tilted towards the wind drift direction? I am SO hoping it was. :)

- George Gassaway
 
Last edited:
Umm, hold on. If the guidance is pre-set on the launcher, at an angle (let's say 10 degrees from vertical, pointing East), then if the rocket rolls 180 degrees, it is going to fly to the WEST at 10 degrees from vertical.

Except that the gyro tells it that it has rolled 180 degrees. That control system for the Paiute-Tomahawk sounding rocket I posted a link to had no trouble. It didn't control roll but kept a fixed yaw and ran a pitch schedule trying to match the usual gravity turn.

Now a wild card in this is how well the Flight Controller handles it. There is term called PID which involves three factors for the feedback and control loop:

In practice the control loop should have gain that varies with velocity. The alternative is to set the gain low enough so that the control system is stable at the maximum velocity and just live with the reduced performance at lower velocities.
 
Except that the gyro tells it that it has rolled 180 degrees. That control system for the Paiute-Tomahawk sounding rocket I posted a link to had no trouble. It didn't control roll but kept a fixed yaw and ran a pitch schedule trying to match the usual gravity turn.
The control for the Paiute Tomahawk is not the same thing as a model aircraft/multicopter flight control board.

They have to be initialized horizontally (or vertically depending how you look at it, and how the pitch sensor is orineted, or how the software deals with which axes are which). Anyway, if the model is tilted 10 degrees during initialization, it is going to have a 10 degree pitch (or yaw) error as the software assumes the model was NOT tilted. And it has no way to correct for that 10 degree error if it rolls 180, it is going to rotate the tilt direction by 180 degrees. Now of course a sufficiently advanced control system can be programmed to deal with that if it rolls, as in the Paiute-Tomahawk you describe. But the Multicopter type controllers I am familiar with cannot do that.

Man, the Paiute Tomahawk is ancient. Last flew in 1981. I would suspect that the method it used was literally a flywheel spinning gyro. So it is an Apples and titanium bolts comparison to say that just because a 40 year old rocket did it, that a modern hobby model helicopter system can do the same thing. Well, with the right person doing custom programming I'm sure it can. But thats not the same thing as what is commonly available, even among the fast-moving Open-Source software options (Like Cleanflght) that often are well ahead what most storebought multicopters can do.

And now for a message I had already typed and was going to post:

The NAZE 32 flight controller board and "Cleanflight" configuration has an option for "Black Box" storage of sensor flight data. With a suitable storage device (SDcard), it can record all sorts of data. Including what it senses for pitch and yaw angles, and how it responded in lateral acceleration loads. So you'd get a pretty good idea of how well the vertical rocket guidance system was working - not enough, just right, wobbling a lot, or crazy overcontrol.

Here is a link to RC Groups thread discussing it.

https://www.rcgroups.com/forums/showthread.php?t=2299805

The video in the first message has an impressive computer-generated graphics that displays that data in several interesting ways, overlaid on an onboard video of the flight. Now, Cleanflight does not create any of those graphics, it simply enables storage of the raw data.

Anyway, thought I'd mention it. Probably too late for this project, but an interesting capability to consider for future projects. Offhand, there is nothing I am planning on that would use the Black Box feature, but later down the road…. maybe.

Oh, I forgot to say something and to ask THE BIG QUESTION about the first flight.

First, congratulations!

Second….. where's the onboard video? :)

- George Gassaway
 
Last edited:
I don't understand why anyone would not have the guidance system turned on and activated before launch. Makes no sense to me but it's not my rocket. The baseline g and theta vectors will never be more stable then when the rocket is pointed vertically and attached to the rod....

Bob
 
The control for the Paiute Tomahawk is not the same thing as a model aircraft/multicopter flight control board.

They have to be initialized horizontally (or vertically depending how you look at it, and how the pitch sensor is orineted, or how the software deals with which axes are which). Anyway, if the model is tilted 10 degrees during initialization, it is going to have a 10 degree pitch (or yaw) error as the software assumes the model was NOT tilted. And it has no way to correct for that 10 degree error if it rolls 180, it is going to rotate the tilt direction by 180 degrees. Now of course a sufficiently advanced control system can be programmed to deal with that if it rolls, as in the Paiute-Tomahawk you describe. But the Multicopter type controllers I am familiar with cannot do that.

It is the current RC aircraft code that is complicated. The autopilot code listing (in assembly!) is only 16 pages of that report.

Most of what the RC aircraft code does isn't needed in a rocket and can be tossed. In fact it probably should be tossed. (Remember the first Ariane 5?) For example the code to compensate for gyro drift using the accelerometers just isn't needed. Before launch the gyro rates are known to be zero.
 
It is the current RC aircraft code that is complicated. The autopilot code listing (in assembly!) is only 16 pages of that report.

For example the code to compensate for gyro drift using the accelerometers just isn't needed. Before launch the gyro rates are known to be zero.

And after launch there is no gravity z-axis reference so the accelerometers are useless. All you need are simple quaterion or DCM rotation code from the gyros.
 
It is the current RC aircraft code that is complicated. The autopilot code listing (in assembly!) is only 16 pages of that report.

Most of what the RC aircraft code does isn't needed in a rocket and can be tossed. In fact it probably should be tossed. (Remember the first Ariane 5?) For example the code to compensate for gyro drift using the accelerometers just isn't needed. Before launch the gyro rates are known to be zero.

Agreed that most of the code is not needed for vertical rocket guidance. The hardware has incredible technology that is there to be exploited by whatever software can be created to use it

But, I am talking about how to use the existing software and adapt it for use.

You are talking about an expert doing a revamp of the code, and adding capabilities that are not there. The Apples and Titanium bolts comparison again.

However.... the capabilties may be there in existing code after all (but you didn't know that or you'd have corrected me before I realized it). I may have been wrong when I said the multicopter code could not keep a rocket that is tilted on the pad at start-up, pointed in the same tilt direction if it rolled 180 degrees. There is a relatively new (new to me anyway) flight mode option called "Headfree", which for a multicopter means you move the yaw stick (vertical rocket yaw reference, otherwise roll stick for copter) left... then the copter moves left, regardless of which direction (East, North, West, South., etc.) it is facing horizontally. It's is a really new way for newbies to fly without knowing anything about pitch or yaw or whatever, move elevator stick forward and the copter flies AWAY from them. Pull elevator stick back, the model comes back. Move stick right, the model moves right. So the pilot is not controlling pitch or yaw, the pilot is controlling - move away, move back, move left, move right.

A lot of R/C multicopter pilots HATE it :).

Converted into rocket terms, that would imply it may indeed be able to keep a rocket that is pointed 10 degrees from vertical, angled towards the east, to still fly that 10 degree eastward path even if it rolled 180 degrees to "face" west. I'm not sure it would work, but on the surface it sounds like if it is possible to do in existing software, that may be it.

But I do not think Jim should do that. Anyone else who does a guided rocket, they are free to do it if they want to, but they have to be careful and certain it won't have unintended side effects. But most people who do this will want to have their rockets boost vertically, and it is easier to test and safely fly one vertically than to screw around with something that is forced to be in error from the instant it was turned on (tilted) and has to compensate properly the rest of the time.
 
Last edited:
Hopefully, the next test will be in a few weeks. For that test, I have fine-tuned the UDB5 mount and I plan to reduce the size of the canards a bit. I'll also add a small camera with a neat little 3D printed mount to look at the canards, and the flight will be on an L1000. After that, we'll install new firm ware to do the elevon & rudderon mixing (proportional control on tilt, rate control for roll on all four canards) with camera pointed down. After that, it will be time to move the stabilization system to the bottom of the rocket.

For the upcoming flights, I need to figure out how to organize the recovery. The top of the rocket is heavy relative to the fins and the rocket doesn't fly very well on drogue (top section straight down, which is not good). I need to do something different, but haven't decided what yet.

Jim

I see some discussion and a few questions for me. I'll get there, but limited time just now.

Just finished the second test flight. This was the Aerotech L1000 disposable motor, which is how I shall forever remember the motor - disposable. It looked like the nozzle was glued on by a third grader and that's about how it performed. Nozzle came off about a second into the flight. Might be some useful data on response under acceleration and some low velocity steering - we'll see. A little damage to the booster, but should be fixable.

Jim
 
I see some discussion and a few questions for me. I'll get there, but limited time just now.

Just finished the second test flight. This was the Aerotech L1000 disposable motor, which is how I shall forever remember the motor - disposable. It looked like the nozzle was glued on by a third grader and that's about how it performed. Nozzle came off about a second into the flight. Might be some useful data on response under acceleration and some low velocity steering - we'll see. A little damage to the booster, but should be fixable.

Jim

Really? I'm surprised. Of the L1000's I've seen I didn't think they looked all that bad. Dang shame. Hopefully that's the worse that happens in this project!

-Dave
 
Jim, for your roll control, you might consider using only two of the four canards to control roll.

However, I do not know if it is practical for the controller's software to control only two for roll and not four for roll.

Oh, the roll guidance probably should be active from liftoff.

So, Jim, that approximately 8 degree off-vertical path. Was it tilted towards the wind drift direction? I am SO hoping it was. :)

- George Gassaway

You have a good point on using 2 canards for roll control and not 4. The current program uses 4, but this can be un-done if necessary.

I have had concerns about activating guidance from the start due to the effects of acceleration. The current flight data, to be presented shortly, doesn't seem to have a problem with this acceleration. We speculate that the current gyros are improved technology relative to the gyros where I noted this problem.

There was not much wind on the first test flight. If anything, the 8-degree path was into the wind and not due to drift.

Jim
 
The NAZE 32 flight controller board and "Cleanflight" configuration has an option for "Black Box" storage of sensor flight data. With a suitable storage device (SDcard), it can record all sorts of data. Including what it senses for pitch and yaw angles, and how it responded in lateral acceleration loads. So you'd get a pretty good idea of how well the vertical rocket guidance system was working - not enough, just right, wobbling a lot, or crazy overcontrol.

.. where's the onboard video? :)

- George Gassaway

Starting with the onboard video, for my next test flight, I will have two cameras. No onboard video for the first two though.

I'm now getting the acquired data from the UDB5. A few graphs of the second test flight are attached.

The flight consisted of an L1000, giving about 12G's, but losing the nozzle after a 1-second burn. The rocket arched to the north and then recovered.

The attached graphs include the accelerometers, the tilt vector and the gyros. The coordinate system, relative to the UDB (mounted vertically) is X axis positive to the left, Y axis positive up and Z axis positive into the board.

The acceleration graph, in G's, shows launch at 28:31, burnout at 28:32, deceleration after burnout and the apogee charge at 28:42.5.

The tilt vector graph is more complicated. The tilt vectors are the three direction cosines of earth-vertical relative to the UDB5 orientation. The numeric value is 16,384 times the cosine of the angle of earth-down to the respective axis. The Y axis is the tilt from vertical. My calculations give an angle of 1.8° max in the wiggle during acceleration, 1° at 28:35, 2.1° at 28:38 and 8.8° at 28:41. The X and Z tilt vectors are difficult to interpret without looking at the other data.

The gyro data is in degrees per second. The initial turn is about 150° clockwise (negative roll on the Y vertical axis). It went pretty straight after that.

I don't see much evidence of guidance during this flight. However, guidance wasn't activated until 5 seconds into the flight, and at that point, the velocity had decreased to below 250 ft/s.

The tilt vector data show a change from vertical during the motor burn. However, there is x axis acceleration which suggests that the rocket actually did turn a little. So, we think the data during acceleration was good.

One of these days, I'll get a flight with a full motor burn and video.

Jim

Accel graph.png

Tilt graph.png

Gyro graph.png
 
I don't understand why anyone would not have the guidance system turned on and activated before launch. Makes no sense to me but it's not my rocket. The baseline g and theta vectors will never be more stable then when the rocket is pointed vertically and attached to the rod....

Bob

The data suggest that stabilization can be active from the start.

Jim
 
And after launch there is no gravity z-axis reference so the accelerometers are useless. All you need are simple quaterion or DCM rotation code from the gyros.

The accelerometers are not used to provide the vertical reference after launch.

Jim
 
Oops, I calculated the tilt angle incorrectly (still learning about directional cosines and how to extract the rocket information we want). A better graph of this is attached. These data are much more consistent with the gps track (can post that later).

Jim

Tilt Angles.png
 
Oops, I calculated the tilt angle incorrectly (still learning about directional cosines and how to extract the rocket information we want). A better graph of this is attached. These data are much more consistent with the gps track (can post that later).

Jim

Is the inflection point at 28:40 the time at which the stabilization system turned off?


Sent from my iPhone using Rocketry Forum
 
Is the inflection point at 28:40 the time at which the stabilization system turned off?


Sent from my iPhone using Rocketry Forum

No. The stabilization system was off at launch and then turned on at 28:36 (5 seconds after launch) for the duration. There's a small inflection at that time, which could indicate some stabilization, but the velocity at that point was less than 250 ft/s. I've heard from others that the canards aren't very effective at that velocity and below.

Jim
 
No. The stabilization system was off at launch and then turned on at 28:36 (5 seconds after launch) for the duration. There's a small inflection at that time, which could indicate some stabilization, but the velocity at that point was less than 250 ft/s. I've heard from others that the canards aren't very effective at that velocity and below.

Jim
I would assume in general the canards should be as aerodynamically effective as fins are at a given velocity. The only issue is the torque that the canards develop versus the fins.

In very general terms, the effectiveness of the fins is proportional to the area of the fins multiplied by the lever arm distance (which is the difference between the CP location of the fins - the CG if the rocket). Similarly the effectiveness of the canards is proportional to the area of the fins multiplied by the lever arm distance (which is the difference between the CP location of the canards - the CG if the rocket).

To minimize drag you want the drag area of the canards to be as small as possible but for control purposes you want the lever arm to be a long as possibly which is why ideally you want the canards as far forward as possible.....as it minimizes drag while maximized corrective torque......

A properly airfoiled fin will have minimal drag at zero angle of attack, and will have the best lift to drag ratio at any angle of attack up to the stall point. For supersonic flight, a hexagonal or double wedge airfoil is simple to fabricate and is low drag at low altitude and high velocity which is where you need the drag to be minimized. You spend no time during boost at low velocity where a classical airfoil shape wins, and by the time you slow down to a velocity where it might matter, the air density is so low that it won't make any practical difference. IMO the altitude gain by staying vertical at high speed far outweighs any drag considerations at low speed near apogee simply because you don't have much air to create drag.

Bob
 
No. The stabilization system was off at launch and then turned on at 28:36 (5 seconds after launch) for the duration. There's a small inflection at that time, which could indicate some stabilization, but the velocity at that point was less than 250 ft/s. I've heard from others that the canards aren't very effective at that velocity and below.

Jim

Roger that, I thought you had it turn back off after a duration of flight time.


Sent from my iPhone using Rocketry Forum
 
I would assume in general the canards should be as aerodynamically effective as fins are at a given velocity. The only issue is the torque that the canards develop versus the fins.

In very general terms, the effectiveness of the fins is proportional to the area of the fins multiplied by the lever arm distance (which is the difference between the CP location of the fins - the CG if the rocket). Similarly the effectiveness of the canards is proportional to the area of the fins multiplied by the lever arm distance (which is the difference between the CP location of the canards - the CG if the rocket).

To minimize drag you want the drag area of the canards to be as small as possible but for control purposes you want the lever arm to be a long as possibly which is why ideally you want the canards as far forward as possible.....as it minimizes drag while maximized corrective torque......

A properly airfoiled fin will have minimal drag at zero angle of attack, and will have the best lift to drag ratio at any angle of attack up to the stall point. For supersonic flight, a hexagonal or double wedge airfoil is simple to fabricate and is low drag at low altitude and high velocity which is where you need the drag to be minimized. You spend no time during boost at low velocity where a classical airfoil shape wins, and by the time you slow down to a velocity where it might matter, the air density is so low that it won't make any practical difference. IMO the altitude gain by staying vertical at high speed far outweighs any drag considerations at low speed near apogee simply because you don't have much air to create drag.

Bob

Good points Bob. For this rocket, the lever arms are about the same for each set of fins. Not much control over that for the test rocket. Also, I continued to reduce the size of the canards (the strategy being to allow more turn for my direct-mounted canards). A pic of the shape of the canards for the two flights so far is attached. Perhaps I can increase the traverse a little? I was using a gain that gave 1° on the canard for each 6° of tilt.

Jim

IMG_0053.JPG
 
Really? I'm surprised. Of the L1000's I've seen I didn't think they looked all that bad. Dang shame. Hopefully that's the worse that happens in this project!

-Dave

As it turns out, an issue with the L1000's was discovered and is already fixed - Sweet! Something to do with the fiberglass and epoxy used in the assembly. Seems like I'm always the last to know - LOL.

The timing is good, though. I should get a new and improved replacement about the time I have the rocket fixed (fortunately, the grains snuffed when the case failed, so my rocket didn't get burned to a crisp). Just lucky I guess.

Jim
 
Good points Bob. For this rocket, the lever arms are about the same for each set of fins. Not much control over that for the test rocket. Also, I continued to reduce the size of the canards (the strategy being to allow more turn for my direct-mounted canards). A pic of the shape of the canards for the two flights so far is attached. Perhaps I can increase the traverse a little? I was using a gain that gave 1° on the canard for each 6° of tilt.

Jim
I may be ignorant or missing something, but I would think you want 1 degree of canard angle for 1 degree of tilt. That minimizes drag by streamlining the canards to the airflow, and maximized the pivot length for the aft fins to provide the directional correction.

I also wonder if you would gain by increasing the aspect ratio of the canards for better lift to drag. The smaller canard don't project very far into the airstream. I wonder if you maintained the area on the left photo of the canard with the width of the right canard to obtain a 2:1 aspect ratio. You should get more lift but not much more drag as the wetted area is maintained.

Bob
 
I may be ignorant or missing something, but I would think you want 1 degree of canard angle for 1 degree of tilt. That minimizes drag by streamlining the canards to the airflow, and maximized the pivot length for the aft fins to provide the directional correction.

I also wonder if you would gain by increasing the aspect ratio of the canards for better lift to drag. The smaller canard don't project very far into the airstream. I wonder if you maintained the area on the left photo of the canard with the width of the right canard to obtain a 2:1 aspect ratio. You should get more lift but not much more drag as the wetted area is maintained.

Bob

I'm pretty sure that using a gain of 1° per degree of tilt would lead to oscillation, and I think it maximizes the drag, doesn't it?

For my specific need, I can afford a slower response to vertical. I need to find the relationship between gain and canard size that gives a response that's "fast enough". I haven't made much progress on that yet.

There are a few people doing similar projects, and their conclusions so far have been to reduce the gain and canard size. I'm not convinced that their conclusions are correct, but that's a story for another day. I'd just like to get some actual data, and hopefully, look at this a little more analytically. I think you're right on the canard shape.

Jim
 
I may be ignorant or missing something, but I would think you want 1 degree of canard angle for 1 degree of tilt. That minimizes drag by streamlining the canards to the airflow, and maximized the pivot length for the aft fins to provide the directional correction.

I also wonder if you would gain by increasing the aspect ratio of the canards for better lift to drag. The smaller canard don't project very far into the airstream. I wonder if you maintained the area on the left photo of the canard with the width of the right canard to obtain a 2:1 aspect ratio. You should get more lift but not much more drag as the wetted area is maintained.

Bob

I tend to agree in general. I have a gut feeling that the canard area is too small for such limited throw.

But, the size and throw might be right for the maximum velocity ultimately planned for the record attempt. But if it is right for the maximum velocity then it will be way too little for even say half the maximum velocity (25% as much control effectiveness).

Next flight, you may want to try it with 1 degree deflection for 1 degree of tilt. And have that danged camera onboard, with one or two canards visible! (see my avatar above left) Then you will be able to see not just from the data but visually, how well (or poorly) it responded. If it overshoots, good, you have proven it works and then consider how much it seemed to overcontrol and dial it back some.

If it only corrects some, but not nearly as much as you wanted, then you know you need to gain more control authority. And that would then probably mean making the canards bigger, as 1 degree control for 1 degree tilt seems better than going for say 2 or 3 degree control motion for 1 degree tilt to try to force more control from the same canard size.

- George Gassaway
 
Last edited:
I'm pretty sure that using a gain of 1° per degree of tilt would lead to oscillation, and I think it maximizes the drag, doesn't it?
You can reduce oscillation by including an angular velocity term in your control loop. If the rocket is moving away from vertical, the angular velocity term would increase the angle of the canard, and if the rocket is moving toward vertical it would decrease the angle. This introduces a damping factor. So the canard angle would be given by "Angle[canard] = A*Angle[gyro] + B*Velocity[gyro]".
 
I tend to agree in general. I have a gut feeling that the canard area is too small for such limited throw.

But, the size and throw might be right for the maximum velocity ultimately planned for the record attempt. But if it is right for the maximum velocity then it will be way too little for even say half the maximum velocity (25% as much control effectiveness).

Next flight, you may want to try it with 1 degree deflection for 1 degree of tilt. And have that danged camera onboard, with one or two canards visible! (see my avatar above left) Then you will be able to see not just from the data but visually, how well (or poorly) it responded. If it overshoots, good, you have proven it works and then consider how much it seemed to overcontrol and dial it back some.

If it only corrects some, but not nearly as much as you wanted, then you know you need to gain more control authority. And that would then probably mean making the canards bigger, as 1 degree control for 1 degree tilt seems better than going for say 2 or 3 degree control motion for 1 degree tilt to try to force more control from the same canard size.

- George Gassaway

OK, I'm convinced - more gain on the next flight. And two cameras, one pointing up to see the canards and one pointing down to see spin more easily.

For my use on the planned three-stage flight, the velocity range where control would be employed is something like 0 to 1300 ft/s on the boost dropping down to perhaps 600 ft/s at staging. I'm wondering if the system should be turned off within the transonic range? By the time I do this flight, we (er ... Bill, but maybe we) will have programmed the UDB5 to vary the system gain as a function of velocity.

Their are some new features on the current version of the UDB5. The UDB5 now provides the centering pulses when the system is off (instead of the servo failsafes). The Raven directly connects to the UDB5 to switch between centering or control. All four canards are used for roll control (although we can switch to two if necessary). 2/3 of the servo travel is reserved for tilt and 1/3 for roll, with maximum roll control deflection at 1 Hz. These are our best guesses - kibitzing is welcome. There is also much more recorded data.

Jim
 
Back
Top