stunthanger.com
News:
CLICK HERE---->    <----CLICK HERE
 
 
*
Welcome, Guest. Please login or register. October 22, 2014, 03:36:36 AM


Login with username, password and session length



Pages: 1 [2]  All   Go Down
  Send this topic  |  Print  
Author Topic: Line tension controlled ESC timer  (Read 11394 times)
Dave Adamisin
Guest
Trade Count: (0)
« Reply #50 on: December 02, 2009, 10:03:06 AM »

Forgive an IC guy a couple questions. It would seem that thrust/weight would have a damping effect on the loads and Igor, could the question about how to know when you're climbing or diving be sensed by the motor load in governor mode?
Logged

Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1405


« Reply #51 on: December 02, 2009, 10:27:52 AM »

Quote from: Dave Adamisin on December 02, 2009, 10:03:06 AM
Igor, could the question about how to know when you're climbing or diving be sensed by the motor load in governor mode?

strange that you are asking ... just today we spoke about it with Keith R.  Devil

no you cannot, but you can sense the prop slippage, what is exactly what we want to compensate

Just imagine you have very very large prop - it will give you very stable speed to air. But if you use smaller prop (because we must save the power somehow) the we have some slippage under load ( slowing uphill). IC engine solves it simple - loaded motor goes from 4cycling to 2 cycling because of better combustion and also loaded prop makes leaner mixture = better pull. It is a kind of acceleration sensor. So if you ask governor to keep little higher rpm (relative to measured slippage) then you can cancel it = keep constant speed independently on the load (drag)

Exactly the same we acn do on electric model ... slippage is linear to angle of attack on prop blades, it is linear to lift coefficient on prop blades and it makes induced drag on prop tips which mas square of power delivery (at constant rpm maintained by governor) ... so if you can measure power delivery (one 3 lin component) you can simulate 4-2-4 IC motor.


here you can see it reality:  Devil
Quote from: Dave Adamisin on December 02, 2009, 10:03:06 AM
Igor, could the question about how to know when you're climbing or diving be sensed by the motor load in governor mode?

strange that you are asking ... just today we spoke about it with Keith R.  Devil

no you cannot, but you can sense the prop slippage, what is exactly what we want to compensate

Just imagine you have very very large prop - it will give you very stable speed to air. But if you use smaller prop (because we must save the power somehow) the we have some slippage under load ( slowing uphill). IC engine solves it simple - loaded motor goes from 4cycling to 2 cycling because of better combustion and also loaded prop makes leaner mixture = better pull. It is a kind of acceleration sensor. So if you ask governor to keep little higher rpm (relative to measured slippage) then you can cancel it = keep constant speed independently on the load (drag)

Exactly the same we acn do on electric model ... slippage is linear to angle of attack on prop blades, it is linear to lift coefficient on prop blades and it makes induced drag on prop tips which mas square of power delivery (at constant rpm maintained by governor) ... so if you can measure power delivery (one 3 lin component) you can simulate 4-2-4 IC motor.


here you can see it reality:  Devil


and here is the device and its settings:
http://stunthanger.com/smf/index.php?topic=12673.0
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #52 on: December 02, 2009, 10:29:46 AM »

Forgive an IC guy a couple questions. It would seem that thrust/weight would have a damping effect on the loads and Igor, could the question about how to know when you're climbing or diving be sensed by the motor load in governor mode?

Dennis,

The main problem with the motor load is, like airspeed, when you sense it being higher (or slow), you have already slowed down.

That's what I see with my glow engine measurements too. When the engine finally breaks from 4 stroking into 2 stroking, the airspeed has already decayed away---that's what of course is providing the larger prop load.

Of course putting more power on, even if late, will speed things up, but I think it would have been better to apply the extra power when gravity is fighting you the most, when you pull the nose up initially.

Again, this is my opinion of course (where to apply the power, but not where the airspeed is lost!).
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1405


« Reply #53 on: December 02, 2009, 10:34:00 AM »

>>>when gravity is fighting you the most<<<

And how you know that gravity pulls you down? Model does not feel the gravity, it feels only aerodynamic forces  Devil

edit:  I mean in longitudinal direction
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #54 on: December 02, 2009, 10:49:44 AM »

>>>when gravity is fighting you the most<<<

And how you know that gravity pulls you down? Model does not feel the gravity, it feels only aerodynamic forces  Devil

edit:  I mean in longitudinal direction


Igor,

I think you  Evil are being sarcastic, but I am missing the fine point of the humor!

Clearly gravity is a force that affects flying--at least here on the surface of the earth! In level flight, the wing does your fighting. But when you are climbing vertical, its only the prop providing a force opposing gravity and drag.

But of course I know you know that! I think you are just baiting me!  n1  Wink
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1405


« Reply #55 on: December 02, 2009, 10:54:37 AM »

Unfortunately it is true ... your acceleration sensor in longitudinal direction canont see the gravity ... only and only aerodynamic forces (prop thrust and drag) Devil
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #56 on: December 02, 2009, 11:15:20 AM »

Unfortunately it is true ... your acceleration sensor in longitudinal direction canont see the gravity ... only and only aerodynamic forces (prop thrust and drag) Devil


Of course an accelerometer senses acceleration a, not forces, so if the plane is accelerating in the longitudinal direction, it should feel it.

Right after the corner, al=(Fthrust - Fdrag - Fg)/mplane, at least when climbing vertically. If thrust is basically just level lap thrust, then this quantity is negative.

From what I have seen from my sensors, and what I have calculated from simple kinematics, al is negative until roughly 50-70 degrees in the climb, then it goes positive. Primarily due to drag lowering (lower airspeed), plane moving more horizontally (less gravity in direction of flight), and a little more thrust due to prop loading (result of lower airspeed).

Just to be clear, I don't think we are disagreeing---are we??
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1405


« Reply #57 on: December 02, 2009, 11:26:56 AM »

>>>Right after the corner, al=(Fthrust - Fdrag - Fg)/mplane, at least when climbing vertically. If thrust is basically just level lap thrust, then this quantity is negative.<<<
Depends what is al ... it is is acceleratin acting on model, then yes, if it is acceleration measured by your sensor, I do not think so  Grin

take out the air around, you drag and trust is zero ... gravity acts on the model body, but what you feel inside the model? gravity or nothing? ... Einstein says nothing  Clown

you can feel gravity only in case that you can maintain constant speed ... but you do not ... you just want  Clown

by other words - if you will regulate you power train so that your sensor shows constant 0, you will slow down uphill ... if you want keep constant speed, you must keep acceleration on your sensor according to angle of longitudinal axle to vector of gravity ... but how kow that angle? that was my orriginal question :-)

Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #58 on: December 02, 2009, 12:17:19 PM »

Igor,
I just came back from my noon walk, and spent the time thinking about what you are saying.

You are right that the accelerometer does not measure directly the acceleration since it is also sensitive to the gravitational field (in the direction of that axis). I think the net result I would measure from the accelerometer along the longitudinal axis (vertical climb) would be the actual kinematic acceleration-9.8m/s2 (1g). So I think this is what you are saying and I have been missing  Head bang from your comments.

However I wasn't thinking about trying to adjust power to keep the longitudinal acceleration=0 (constant airspeed) , but actually instead use the inputs from all three axes to provide some type of a power "burst" (hopefully not explosive!) in the lower altitudes (below 45 degrees). So I am not trying to hit constant airspeed--I am not sure how important that is. Just improving the airspeed above 45 degrees may be enough. I don't really know.

But at first I'd just like to record what the accelerometer is seeing in a typical pattern. Also I will probably be experimenting with more propellers next year, just to try and see how I can improve things "passively". If I can really get my head around the setup I just ordered (logomatic and 3 axis + 16g accelerometer), maybe then I might consider actually feeding in a throttle input to the ESC. That issue is basically reprogramming the device.

And moving off topic now and onto the setup Kim D. was talking about, I wonder if decoupling the line tension from the control effort (Netzeband wall) might be just as effective in improving control near the top as trying to get the airspeed up---as long as the tension is >0 on the lines and the bellcrank/pot moves.
Logged
Howard Rush
2012 Supporter
Trade Count: (0)
Admiral
*
Offline Offline

Location: Bellevue, WA
Posts: 4702



« Reply #59 on: December 02, 2009, 12:24:13 PM »

"And moving off topic now and onto the setup Kim D. was talking about, I wonder if decoupling the line tension from the control effort (Netzeband wall) might be just as effective in improving control near the top as trying to get the airspeed up---as long as the tension is >0 on the lines and the bellcrank/pot moves."

You can do this mechanically with aerodynamic balance. The 707 elevator and aileron have no power boost. 

Logged

Impacts zijn oranje,
Noblers zijn blauw.
Het Jive Combat Team
Vliegt beter als jou.
Alan Hahn
Guest
Trade Count: (0)
« Reply #60 on: December 02, 2009, 12:30:01 PM »

"And moving off topic now and onto the setup Kim D. was talking about, I wonder if decoupling the line tension from the control effort (Netzeband wall) might be just as effective in improving control near the top as trying to get the airspeed up---as long as the tension is >0 on the lines and the bellcrank/pot moves."

You can do this mechanically with aerodynamic balance. The 707 elevator and aileron have no power boost. 



I really hate hijacking Erik's thread, but.....

Someone must have already done this. Since I don't see a lot of stunters with elevator ears (is that what they are called?) or the weights which I think were suppose to do the same thing, should I think it wasn't worth the effort??
Logged
Dave Adamisin
Guest
Trade Count: (0)
« Reply #61 on: December 02, 2009, 12:46:10 PM »

I really hate hijacking Erik's thread, but.....

Someone must have already done this. Since I don't see a lot of stunters with elevator ears (is that what they are called?) or the weights which I think were suppose to do the same thing, should I think it wasn't worth the effort??
You haven't looked back far enough. They used to be popular. Could've been we needed the aero assist to move the flippers due to the low line tension from the unresponsive ic's Layingdown. By the way it's not a big deal but I'm Dave (the lost Adamisin) I only wish I were Dennis...
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1405


« Reply #62 on: December 02, 2009, 02:24:32 PM »

I think the net result I would measure from the accelerometer along the longitudinal axis (vertical climb) would be the actual kinematic acceleration-9.8m/s2 (1g).

It can be true only at constant speed but we know (and you measured it) that model slows down after turn up.

I think we are still speaking about time immediatelly after turn up - means speed is equal to the speed in horizontal flight and thus drag of the model equal prop thrust. You expect to see 1g on sensor, but you will see 0 on sensor, because gravity will act, it is true, but the model will slow down, because trust = drag and gravity will cause slowing down exactly in value -1g. You will see 1g after a time, when speed will be so low, that drag will be so low, and thrust so high that diffenece between them will be equal to weight of the model in gravity field (and model will not slow down anymore - it will be constant - and that the condition when you can read 1g on sensor) ... means if model goes vertically, you will see 0g in level flight, then still 0g in vertical climb and slowly growing to 1g.

If this is clear, then we COULD regulate speed on base of that sensor, but only in vertical axle, not on sphere, because model is angled, so if you want to do it well, you must somehow derive its angular position, and I affraid, you cannot do it without angular sensor (gyroscope) at least I have no idea how ... and if yes, I start speaking about wind an turbullence  Clown
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1405


« Reply #63 on: December 02, 2009, 02:30:46 PM »

And moving off topic now and onto the setup Kim D. was talking about, I wonder if decoupling the line tension from the control effort (Netzeband wall) might be just as effective in improving control near the top as trying to get the airspeed up---as long as the tension is >0 on the lines and the bellcrank/pot moves.

There are many ways how to move Netzeband wall up and down, one way is trick mentioned by Howard, and there are many other ways, like proper hinging, using of large bellcranks (I have 4" in my indoor, the same like I have in my stunter for outside), but worst is hingemoment from flaps ... I use logarithmic device on flaps, and that model is my best chice in turbullent air, when you never know line tension in last picosecond before corner Evil
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #64 on: December 02, 2009, 02:57:03 PM »

You haven't looked back far enough. They used to be popular. Could've been we needed the aero assist to move the flippers due to the low line tension from the unresponsive ic's Layingdown. By the way it's not a big deal but I'm Dave (the lost Adamisin) I only wish I were Dennis...

Sorry Dave, I must have put the "D" in and the fingers typed the "ennis" without input from the brain!
Logged
Dave Adamisin
Guest
Trade Count: (0)
« Reply #65 on: December 02, 2009, 03:01:06 PM »

No prob Alan. I always read your posts with interest and have really enjoyed the ic speed tracking.
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1405


« Reply #66 on: December 06, 2009, 04:55:23 AM »

Allan, look one small video from yesterday:



regulation is based on prop slippage as a positove feedback to thrust (rpm=function(slippage)) ... means something what longitudinal acceleration sensor can do, IF we could subtract gravity, because slippage=function(speed) and speed changes can be derived from accelerations sensor LESS gravity contribution (which we do not know)

It is visible, that strongest pull in loops is between 9 and 12 o'c, while pull past 12o'c  is already descending, because of positive feedback. That feedback implements little advance in regulation (in first derivation), because it enough that model accelerates, even still slow, it "asks" for lower rpm and thus make slippage smaller. It can lead to oscillations, if gain is more that slippage. (but the regulator has short low pass filter anyway - shorter than time for corner)

I think that regulation ONLY on base of longitudinal acceleration sensor (rpm = function(acceleration)) will add most power at top and past 12o'c because of gravity. Gravity will not allow to "see" slowing in climbing. Pulling on top of loop will be too late.
Logged
Dean Pappas
Trade Count: (0)
Admiral
******
Offline Offline

Age: 57
Location: Central New Jersey
Posts: 1159

Welcome to the Stunt Hanger.


« Reply #67 on: December 06, 2009, 02:25:57 PM »

Yes, Igor,
What I suspect is needed is a longitudinal "jerk" sensor. This is English Engineering slang: jerk is defined as the 1st derivative of acceleration. It will provide the time-lead you see the need for, and I think I know how to build one. It is mechanical, and after i finish my present writing, I will draw this up and publish here.
Dean
Logged

Dean Pappas
Dean Pappas
Trade Count: (0)
Admiral
******
Offline Offline

Age: 57
Location: Central New Jersey
Posts: 1159

Welcome to the Stunt Hanger.


« Reply #68 on: December 08, 2009, 08:26:02 PM »

Hello again,
Let's see if this PDF comes out readable. I think this is a differential jerk-meter.
The idea is that in a breeze-free box in the outboard wing panel, there is a balance beam.
The pivot is a vertical wire (probably with rubber motor style  bead bearings) and the inner arm is half as long as the outer arm. The beam is "balanced" so the mass on the outer end is half that of the one inboard. As the longitudinal acceleration changes, the beam will momentarily unbalance because the "center of percussion" is outboard of the pivot.

Constant airspeed either straight up or straight down does not move the beam from center, but changes in airspeed will. Conceptually a weak centering spring and position sensing potentiometer will create a signal that can be used to help the governor maintain airspeed. If I do get a chance to build this, it the spring would be replaced with a magnet(s) glued to the center of the beam and an electro-magnet with control loop (using some form of non-contact sensing) to maintain the beam centered. The current in the control magnet is the desired signal. This would make the beam "virtually stationary" and should improve the frequency response.

any thoughts?
Dean P.

* jerk_meter.pdf (11.07 KB - downloaded 94 times.)
Logged

Dean Pappas
Howard Rush
2012 Supporter
Trade Count: (0)
Admiral
*
Offline Offline

Location: Bellevue, WA
Posts: 4702



« Reply #69 on: December 08, 2009, 11:29:46 PM »

Looks like you will have a yaw rate sensor, which is more useful than a jerk sensor anyhow.  Use a servo to keep it centered, and measure the servo current (or maybe voltage).  Or just get one of these: http://www.analog.com/static/imported-files/data_sheets/ADXRS150.pdf and try to solder it.
Logged

Impacts zijn oranje,
Noblers zijn blauw.
Het Jive Combat Team
Vliegt beter als jou.
Dean Pappas
Trade Count: (0)
Admiral
******
Offline Offline

Age: 57
Location: Central New Jersey
Posts: 1159

Welcome to the Stunt Hanger.


« Reply #70 on: December 09, 2009, 07:14:45 AM »

Hi Howard,
I don't think it measures yaw rate: there is a yaw acceleration term as well as a longitudinal jerk term.
The big thing is that it will ignore gravity in both climbs and dives.
Dean
 
Logged

Dean Pappas
Howard Rush
2012 Supporter
Trade Count: (0)
Admiral
*
Offline Offline

Location: Bellevue, WA
Posts: 4702



« Reply #71 on: December 09, 2009, 09:11:30 AM »

Yep, yaw acceleration, not rate.
Logged

Impacts zijn oranje,
Noblers zijn blauw.
Het Jive Combat Team
Vliegt beter als jou.
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #72 on: December 10, 2009, 05:10:31 AM »

This gyroscope board is a basic carrier/breakout board for the ST LISY300AL single-axis gyro, which measures rotational motion about the yaw (z) axis with a 300/s range and outputs an analog voltage.

http://www.pololu.com/catalog/product/765

It would be interesting to hook this one up to a data logger and see what it comes up with.



Logged
Howard Rush
2012 Supporter
Trade Count: (0)
Admiral
*
Offline Offline

Location: Bellevue, WA
Posts: 4702



« Reply #73 on: December 10, 2009, 12:32:33 PM »

It would be interesting to hook this one up to a data logger and see what it comes up with.

Yes, it would.  As we speak, there's a rate gyro evaluation board in my laboratory--left there by Dr. Briggs--and I have a Logomatic, which is ironic, because I live less than a kilometer from Eagle Tree.
Logged

Impacts zijn oranje,
Noblers zijn blauw.
Het Jive Combat Team
Vliegt beter als jou.
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1405


« Reply #74 on: December 10, 2009, 02:00:40 PM »

 Grin ... so why did not you use eagletree and usual heli gyro? you not need to do anything, just plug few connectors
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #75 on: December 10, 2009, 03:19:38 PM »

Grin ... so why did not you use eagletree and usual heli gyro? you not need to do anything, just plug few connectors

The issue there is that the small e-logger doesn't support the gyro's or the accelerometers, so you need the bigger Flight Data recorder+ the sensors. I was going to say it is more costly too, but that assumes the time you take to learn and actually program something like the logomatic is zero! Layingdown

It would have been nice if the Eagletree elogger system would support "foreign" sensors, but up to now it doesn't. But at least they allow you to read out their altimeter and airspeed (I2C) sensors.
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1405


« Reply #76 on: December 10, 2009, 03:26:36 PM »

hmmm ... looks like I remember wrong ... I thought it has usual servo input :-(
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1405


« Reply #77 on: December 10, 2009, 03:39:11 PM »

haaaaaaah ... yes, it has throttle input, so you can logg output of the gyro

http://www.eagletreesystems.com/MicroPower/3.htm
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #78 on: December 10, 2009, 06:26:25 PM »

That's a good point as long as the gyro puts out a PWM signal (1-2ms), which of course a heli gyro would. I am not sure what the gyro puts out without an input (typically a 50 Hz input coming from a rudder output on a Heli), but I haven't looked either.

I'll also comment that my data logger is the V2 version. I have been resisting buying the newer V3 version, but I am guessing that "resistance is futile".
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1405


« Reply #79 on: December 11, 2009, 01:32:49 AM »

Usual rate gyro puts input signal to output (repeats it), but adds small difference regrading its angular velocity * gain. So have to feed him by something - for example signal from timer.

On picture - you can see my attempt to control rudder by a gyro feeded from timer and powered 4nicd  Layingdown ... that was age before lipos ... looks lile I am getting old  Grin



* DSC01306.JPG (38.08 KB, 614x460 - viewed 182 times.)
Logged

Tags: F2B E 
Pages: 1 [2]  All   Go Up
  Send this topic  |  Print  
 
Jump to:  


Powered by MySQL Powered by PHP Powered by SMF 1.1.20 | SMF © 2013, Simple Machines

Valid XHTML 1.0! Valid CSS! Dilber MC Theme by HarzeM