stunthanger.com
News:
CLICK HERE---->    <----CLICK HERE
 
 
*
Welcome, Guest. Please login or register. July 25, 2016, 11:47:52 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 15149 times)
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« on: October 29, 2009, 09:13:28 AM »

I have built a device that controls the ESC output as a function of line tension. I modified my 1994 Shanghai WCh model to accommodate this sensor timer and electric power plant. The model was originally built for a Super Tigre .46 and flown on the Dutch Nationals in 2009 to the 4th place within 100 points per flight of the winner. This is quite good as the model is now 16 years old, 6 oz heavier than with the .46 and the pilot had not flown a competition since 1996. I used a Castle Creations Speed Controller

This will allow you to fly in extreme conditions, I have test flown the setup to prove the potential in heavy winds and posted a movie on youtube

This is what it does:
The timer consists of a processor board and sensor that are put into the airplane. (18 gram) The LCD controller is used to set some flight parameters and stays on the ground.

When do I want it to work:
In level flight I want NO reaction from the controller, constant airspeed makes the model groove, constant groundspeed makes the model hunt.
In the low maneuvers I want NO reaction from the controller, ANY sign of 4-2-4 stroke makes the loops wider than high, so the sensitivity of the device is set in such a way that it barely hits the 4-2-4 stroke on the top of the loops.
In the double maneuvers and the verticals I DO want the sensor to open up the throttle.

How do I control it:
The ESC value (FLIV) is increased by a proportional factor (PPOS) multiplied with the loss of line tension (SENSOR VALUE) when the line tension is below the minimum line tension (GMIN)
IF SENSOR < GMIN THEN ESC OUT = FLIV + (PPOS * (GMIN – SENSOR VALUE))
IF ESC OUT > MAXV then ESC OUT = MAXV (to limit the maximum output)

The following parameters can be set to adjust the sensitivity:
FLIV   ESC OUT for Flight Value
WTHT   Walk to Handle Time, motor off
FLIT   Flight Time
PPOS   Proportional factor, defines the steepness of throttle up
MAXV   The maximum output, in moderate wind you may not need full throttle
GMIN   The line tension where you want the sensor to increase the output

Most surprising is that a horizontal and vertical eight normally take 10 secs with an IC model, in wind the overhead eight is then flown in 12 secs. My device opens the throttle in the overhead eights and I can fly them in 10 secs too.

Tech details:
The system is promising but needs more time to be developed further. The reaction of the system is fast, the processor running at 20mhz can do 10.000 sensor comparisons per second, I reduced this to 20 per second. The CC45 reacts very fast I am totally happy with the setup.
So from the PID controller I am only using the Proportional factor. I tried to add a Differentiator but this does not improve anything in the air. I do not intend to add an Integrator as I want the model to lose some speed anyhow.

Now all I need is a better airplane and more practice, seems that in 16 years nothing has changed.


* Eagle .46E small.jpg (111.86 KB, 653x490 - viewed 294 times.)

* SENSOR CONTROLLER.jpg (149.55 KB, 775x519 - viewed 265 times.)
« Last Edit: October 30, 2009, 08:36:44 AM by Erik Janssen » Logged

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

Posts: 1645


« Reply #1 on: October 30, 2009, 05:02:08 AM »

Hi Eric, that number 26m/s is surprising, how did you measured it? By eye I think the wind was may be 8m/s. The landing looked like that, also I did not see any traversing in 8 overhead and hourglass which will be necessary in such wind. Was it a permanent wind 26m/s or just single gust?

Normaly people cannot stay on ground in such wind, I cannot imagine flying. :-)
Logged
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #2 on: October 30, 2009, 06:58:44 AM »

Thanks for the compliment, my flights always look as if there is not much wind, that seems to be the story of my life. I wanted to prove that the system could work in heavy wind conditions. My flying field is 5km away from an official website where the airport reports the weather. When returing home I make a copy of the weather conditions, see attachment. I grew up in Holland close to the sea where there is a lot of wind.

Normally in governor heavy wind blows you out of the sky and you cannot fly an hourglass. I proved that in Almere in Spring where the competition was cancelled and I flew my plane without a sensor and nearly lost it.

See how?

With the sensor the motor goes full throttle to battle the wind. I have logged in flight an ESC input of 57A on a 4S battery at the top of the hourglass.

At 45 degrees the line tension is reduced bij 0,7 and overhead the model hangs towards the pilot so the loss is 1g
I set my parameters in such a way that from 0,8g it starts adding throttle and at a loss of 1,2 it is on full throttle.

I compared the time needed per maneuvre of two very windy flights with the time Beriger needed in Spain in 2006. This shows that the sensor really helps to fly in wind, without the sensor I could not fly my hourglass.



* Wind schiphol.jpg (11.63 KB, 443x141 - viewed 207 times.)

* Maneuvre speed.jpg (53.56 KB, 591x474 - viewed 213 times.)
« Last Edit: October 30, 2009, 07:47:10 AM by Erik Janssen » Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #3 on: October 30, 2009, 07:40:29 AM »

This needs a cross posting in the Electric Forum! (maybe I should look first and see if it already is.)
Logged
PerttiMe
Trade Count: (0)
Admiral
******
Online Online

Location: Finland
Posts: 966



« Reply #4 on: October 30, 2009, 07:59:42 AM »

Doesn't the weather report say "windsnelheid (kts) 19" (19 knots)
??

Gusts 27 knots?

That is just under 10 m/s and 14 m/s. A good wind.
Or am I missing something?
Logged

I built a Blue Pants as a kid. Wish I still had it. Might even learn to fly it.
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #5 on: October 30, 2009, 08:37:16 AM »

That is correct, I corrected the title of the movie. Our flying speed is in the range of 26 m/s not the wind
« Last Edit: October 30, 2009, 03:05:40 PM by Erik Janssen » Logged
Peter Germann
Trade Count: (0)
Commander
****
Offline Offline

Age: 73
Location: LSZH
Posts: 223



« Reply #6 on: October 30, 2009, 08:52:27 AM »

I have built a device that controls the ESC output as a function of line tension. The system is promising but needs more time to be developed further. The reaction of the system is fast.

Good to see you're back in business, Eric.

How does your setup react in terms of braking in dive? What happens in the wingover? Does it reduce power quick enough to prevent too fast dive? I mean, if it would actually reduce the airspeed in that particular dive considerably, then tight, flat and level pullouts from wingover dives might actually become easier to fly...

Anyway, thank you for sharing the the result of your research with the community, I look forward to hear more about it.

regards, Peter
p.s. I take your remark re the 16 years as a big compliment for those who, in whichever form, actually run the event.
Logged

Peter Germann
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #7 on: October 30, 2009, 10:04:51 AM »

The system takes away the power very fast, I have NO overshoot in any of the maneuvres, the wingover is quite easy to fly.

The only difficult point is the forward squares, when you fly through the wind you lose line tension and the motor picks up just before going into the dive. That makes you practice to fly downwind. If your position is right this does not occur.

An electric motor has no brake, in the the test diagram you can see the rpm change when the prop load is doubled and released. The overshoot is influenced by the weight of the propeller. CC45 is good, Kontronik very bad. I have not tested other controllers.

Braking a dive can only be achieved by a very light airplane with lots of drag.


* RPM increase decrease.jpg (79.16 KB, 809x551 - viewed 199 times.)
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #8 on: October 30, 2009, 10:41:40 AM »

I agree with Erik about the brake.

In my simple physics plot of the first part of the wingover, during the dive the thrust is smaller than in the climb, but still positive. It is the drag that is actually increasing as the plane is diving. I'll repost the graphic it here since I think it may be lost in the other thread.

The thrust us the blue trace--notice it is still positive even at the 180 degree point --which would be (actually past) the pullout from the wingover. However the drag (yellow trace) is rising and larger and opposing the thrust. Of course gravity is still winning here!

The horizontal axis on the left plot is degrees (0=level lap entry, 180 = inverted lap pullout---but I don't actually include what happens in the corners themselves), and the right plot x-axis is seconds to get a feeling what is going on in "real" time.
added later--the left vertical scale is for line tension, the right side is for all other quantities (sorry I left that out).

added---of course in this case, the rpm is being held constant. If the rpm was allowed to drop, so that the plane got past the nominal pitch speed, I would think you would get some braking action. I base this on my experience from over a year ago, when we didn't have the prop braking feature on the Castle Creation ESC, and so when the timer timed out, the prop would freewheel. The airplane really slowed down fast in that case, much faster than when the prop is stopped by a brake.


* LineTensionCalculation.png (139.31 KB, 1199x743 - viewed 222 times.)
« Last Edit: November 02, 2009, 02:47:35 PM by Alan Hahn » Logged
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #9 on: October 30, 2009, 12:51:15 PM »

My sensor feeds a standard PID controller on a single axis sensor pointing away ftom the pilot.
The P (proportional) factor is linear to the loss in line tension, therefore it reacts immedeately as soon as line tension is back
The I (integrater) factor is not used, as this slows down reaction time, it has to integrate = count which takes time

I tried to slow down the model as soon as it accelerates. For this purpose I used a D (differential) factor. When the next line tension is higher than the previous it is regaining speed. Then I shut down the ESC.

I made a test flight with three vertical eights without D factor and three vertical eights with the D factor. All eights go up in 2,68 to 2,70 secs and come down in 2,38 to 2,40 seconds. There was no difference between the two settings.

It seems that a converted .46 airplane with a weight of 1750 gram is too heavy for the motor to be lifted and simply drops down due to the lack of drag.

It seems that the speed controller is not forcing the prop to run on a lower rpm and still needs to freewheel the rpm away.
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1645


« Reply #10 on: October 30, 2009, 03:28:56 PM »

Well that number in knots looks much better than 26m/s ... I think you used wrong units and it perhaps 26km/h what gives 7.2m/s and it seems appropriate, I expected 8m/s. OK

So now we know conditions. I see that you can keep it on lines even at relatively slow lap times, it was 5.4 if I measured well. But Erik, do you remember what I wrote you last winter? If you use centrifugal sensor, you will always have 1g less overhead. And it will always lead to overspeeding in diving. You use differential input in regulation, so you can suppress it little bit, but it will remain to some exetent whatever you do. You can see it well especially in hourglass, it nicely adds power after first corner till second corner. That is part when model must penetrate wind. But motor must slow down immediately after second corner, but it will not, because you have still -1g on sensor, and that send you with too high speed to the third corner and result is, that radius of that corner ends too low, may be at 5o degrees or so ... and all still too fast, so if you want finish the maneuver, you must do short horizontal flight at 45 degrees to allow slowing down before last corner which anyway too quick.

You are right, the prop does not have enough braking effect. The reason is clear, it pulls at much higher rpm and blades work much better at high rpm and good orienation of the airfoil, compared to low rpm and reversed airfoil. It means that such a system nmust work in advance. This is too late, but not too late from sensor or device point of view. It all works well. It is too late from nature of approach.

It just does counter productive things - exactly like my simulation of 4-2-4 or my gyro.

This experiment shows, that everything is doable, just we need well specify what exactly we need  Devil
Logged
Dean Pappas
Trade Count: (0)
Admiral
******
Online Online

Age: 59
Location: Central New Jersey
Posts: 1190

Welcome to the Stunt Hanger.


« Reply #11 on: October 30, 2009, 05:12:53 PM »

Hello Erik,
This is terribly interesting!
Are you using the governor mode and using the proportional plus derivative terms to vary the governor RPM setpoint, or are you using a simple (RC -style) throttle and varying it?

Regards,
Dean Pappas
Logged

Dean Pappas
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #12 on: October 31, 2009, 02:28:09 AM »

I do not fly PID all the time.
In normal flight I fly governor, this leads to constant airspeed and the model is happy with that.
In wind the prop will act as a windmill and the governor will reach the requested rpm without delivering power, this makes us fall out of the sky.

When the model reaches 45 degrees angle it loses 1/2 sqrt2 of gravity due to the position of the model, in the overhead it is hanging vertically towards the pilot so we lose 1g (at constant groundspeed)

So I add power as soon as the loss is BELOW the minimum line tension (0,7g loss) and use the P controller to add power. The steepness is defined by the Proportional factor and the Full Throttle position defines the max power. The 0,7  P factor and Full are parameters.

As soon as the line tension is ABOVE the 0,7g loss I go back to the fixed throttle position. Governor takes over and the model slows down in the bottom of the maneuvres.

These parameters allow you to influence the amount of extra power and how soon it becomes available.

With this device I reached the limit of my airframe. My model is 16 years old, relatively small has severe overweight and a warped wing. Nevertheless it stil is competative. For further developments I need a better airplane, I started building an Impact as this was the standard in my previous life.  


Modv = the new value to be sent to the ESC
Fliv = flight value
Gmin = minimum line tension
Sensor = sensor
Ppos = Proportional factor

Control loop 20 times per second:
modv = fliv ;
if (sensor < gmin)
   modv = fliv + (gmin - sensor) * ppos ;
if (modv >= maxv
   modv = maxv
ESC out = modv

My CC45 settings:

#######################################################
# Castle Link Data File
# Created: zaterdag 9 mei 2009
# Do Not Edit This File By Hand
#######################################################
Hex55: 85
Brake Strength: 60
Brake Delay: .6 sec (Delayed) (*)
Brake Ramp: Medium
Cutoff Voltage: 120
Current Limiting: Normal (*)
Cutoff Type: Soft Cutoff
Motor Start Power: 59
Direction: Forward (*)
PWM Rate: 12 Khz (*)
Vehicle Type: Control Line
Throttle Type: Governor Mode
Throttle Response: 5
Governor Gain: 21
Initial Spool-Up Rate: 8
Head Speed Change Rate: 8
Auto Rotate Enabled: False
Governor Mode Type: High
Vehicle Setup - Battery Type: LiPo
Vehicle Setup - Number of Cells: 4
Vehicle Setup - Battery Voltage: 14,8000
Motor Timing: 5
Power-On Beep: Enabled (*)
Auto-Lipo Volts/Cell: 3.0 Volts/Cell (*)
« Last Edit: October 31, 2009, 02:45:50 AM by Erik Janssen » Logged
Crist Rigotti
Trade Count: (0)
Admiral
******
Online Online

Posts: 2923


Electric - The future of Old Time Stunt


« Reply #13 on: October 31, 2009, 06:53:53 AM »

Erik,
Great thread!  Have you tried using a pusher prop?  If so what difference did it make with your set up?

Thank you,
Logged

Crist
AMA 482497
Waxahachie, TX
Electric - The Future of Old Time Stunt
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #14 on: October 31, 2009, 10:54:06 AM »

Yes,

Last month I mixed up the ESC wires in my model. I pushed the start button and the timer started counting the Walk to Handle Time. After 18 seconds the motor started backwards and my model drove into my pitbox. Luckily my pitbox has my field battery inside and the model did not flip over and stayed in position untill I could grab it.

Now my leads are red-white-blue so I cannot mix them up anymore.

I prefer the traditional setup, motor up front. I am flying APC-E props, 13x6,5 is my favourite in wind.

Logged
Dean Pappas
Trade Count: (0)
Admiral
******
Online Online

Age: 59
Location: Central New Jersey
Posts: 1190

Welcome to the Stunt Hanger.


« Reply #15 on: October 31, 2009, 12:18:54 PM »

Thanks Erik,

Dean
Logged

Dean Pappas
Crist Rigotti
Trade Count: (0)
Admiral
******
Online Online

Posts: 2923


Electric - The future of Old Time Stunt


« Reply #16 on: October 31, 2009, 08:13:20 PM »

Erik,
When I referred to using a pusher prop I meant a reverse rotation prop.  I think you can get an APCE-13x6.5 as a pusher prop.  You just reverse the rotation of the motor.  I found that  it helps with line tension especially on the V8, hourgalss and OH8.
Logged

Crist
AMA 482497
Waxahachie, TX
Electric - The Future of Old Time Stunt
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #17 on: November 01, 2009, 01:50:27 AM »

Sorry, I mis interpreted your question. With my sensor I have line tension in ALL the maneuvres, even when flying in strong wind with a baseline speed of 5,4 sec/lap.

I never tried a pusher prop.

If you feel better with this rotation, then it helps. Very small changes may influence the overall performance, some of them can be explained while others cannot. Stick to the things that work for you.

Without sensor I cannot fly in wind, with the sensor I can. With the parameters I can shift the 4-2-4 stroke point up and down. I found that a classic 10 o-clock > 2 o-clock makes my loops wider than high so I moved it up to 12 o-clock so it runs faster in the top loop of the vertical eight and it does nothing in the bottom loop.

In the forwared squares I get in trouble if I fly it through the wind, the motor picks up and then I have to go down.. Flying at the right position solves this problem.

So I found something that works for me, so it helps. The next improvement will be started here.

So if you feel hapy with the rotation stick to it and move on to the next level.
« Last Edit: November 08, 2009, 02:45:35 AM by Erik Janssen » Logged
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #18 on: November 01, 2009, 12:52:03 PM »

Igor,

I will try to explain two major differences between you gyro experiment and my setup.

As far as I understood you flew the gyro indoor. Without wind constant groundspeed and constant airspeed are the same. I flew the sensor from Wolfgang Nieuwkamp in constant groundspeed outdoor. The wind made the model to fly in different airspeeds in a lap, groundspeed + windspeed against the wind and groundspeed - windspeed with the wind on the tail. I found it very difficult to fly level, the model kept climbing and descending all the time. Therefore I prefer to fly level in governor without any extra power.

You explained that the gyro reacted in both plus and minus direction. My setup only adds power, I do not close the throttle to reduce the speed.

The 1g overhead you mention is only there if I would put my gmin value at flying speed. In normal flight we pull approximately 2,7g If I put my sensor to 2,7 it will react the same as your gyro with the 1g overhead. I prefer to put it on 2,0g so the power comes on only after a loss of 0,7 in return I get half the down leg of the hourglass to bleed off the speed which is enough to make a proper corner at the bottom.


Is this underasstanding correct?
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1645


« Reply #19 on: November 01, 2009, 04:22:31 PM »

Erik, I flew regulators with all 3 ways to control speed also outside. I wrote that I use it indoor, because I found good and satisfying way how to use indoor, but it does not work safely in all conditions outside (yet). So that is why I never wrote "I got it" :-)

So yes I tried it and I know which and when make troubles. I wrote it several times, may be my English is not good enough to explain it. I will try to take the hourglass as an example, may be it will be more clear.

I would say, from my human experience, that the power (rpm) needs to be higher immediately after first corner, when model goes up, means thrust of the prop which is in that point equivalent to drag of the model, cannot handle gravity. It means the model is slowing down and if we want to keep the speed, we must add power - immediately after that first corner.

If model reaches the top of the sphere, we should come back with the power to value in level flight. Means we should do it after second corner. Model flies second segment of hourglass in the same height, means it does not slow down anymore with standard power.

And we have to brake after the third corner, because the model is down the nose and it will ovespeed if we do not.

But your system with acceleration sensor does something else. It does not do anything after first corner and it allows slowing down. At 45 degrees starts to accelerate, what is ok, but strongest acceleration happens on top of sphere, where it sees -1g, instead of – 0.7g at 45 degrees, and that is wrong – as I wrote we should be bact with the power. And it still pulls also after 3rd corner down to 45 degrees elevation. Your trick with derivation will help to some extent, but cannot compensate systematic shift of its effect. It just acts 45 degrees later than it should. Result is that high figures make tight lines as you say, but it prevents you to make good bottoms in figures with corners, just because it is too quick for human maneuvering. 3rd and 4th corner of hourglass is good example. Evryone who fly seriously knows, that improper motor run will damage flying properties also on very good trimmed model.

If you compare it to the usual wet 4-2-4 setup, it adds very soon  when you point nose up, because it goes lean in that position.
Logged
Dean Pappas
Trade Count: (0)
Admiral
******
Online Online

Age: 59
Location: Central New Jersey
Posts: 1190

Welcome to the Stunt Hanger.


« Reply #20 on: November 01, 2009, 07:23:53 PM »

Hi Gang,
It amuses me when non-native English speakers apologize for their excellent facility with the language! Thank you both for making yourselves well understood.

May I suggest another open-loop correction.
We see from the instantaneous power utilization curves that a momentary increase in drag and momentary drop in RPM result from every square corner ... even the first turn in the outside squares!
How about putting a sensor (potentiometer) on the bellcrank, and differentiating its position and then gating that signal so that it adds to the RPM setpoint ONLY when the handle is moving away from neutral? Tuning this will also be a compromise between the climbs and dives, but my instinct tells me that the compromise is not so great, because the handle movements differ.

Dean P.
Logged

Dean Pappas
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #21 on: November 02, 2009, 12:38:45 AM »

As I think it will be no fun posting in Dutch I will continue in English.

I agree that improper motor runs will damage your flying properties, in the IC age I did all sorts of experiments with tank layout, position, prop pitch, diameter, number of blades, compression, muffler system, plugs and even the inside diameter of the fuel feed line. Try 1/8 inside and your 4-2-4 will be more crispy.

Another thing that damages your flying properties is a bad model, in my case the electric conversion added too much weight. I do agree that it would be easier to fly if it would come down slower but I feel that the primary reason is the weight of my current model. I do not experience the effect as severe as Igor mentions, I expect the 123 cells converted model to be rather heavy too.

For next year I am building an Impact, this has a thick wing profile. I expect it to lose the overweight. Hopefully this will make me lose some speed in the downleg, if not it will provide me with a better platform to work with.

The sensor gives excellent reading on X,Y and Z axis, an ADC (analog comparison) gives a reading of 335 at 0g and 270 on -1g and 400 on + 1g I cannot simulate the 2,7g we are flying at but from the model behaviour I deduct that it will be around 515.

There is a lot of resolution and we can do anyting with it, adding a potmeter is not needed as we already have the reading
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #22 on: November 02, 2009, 07:59:49 AM »

Erik,
I applaud your measurements and am amazed that you are going after the wind problem. I am having problems just trying to understand flying under perfectly calm conditions!

So for a brief moment, if we forget about wind, one thing I have thought about---if I was really trying to figure out where to pour on the power--- would be to to use the 3 axis accelerometer in a finite state machine. We know (I think!) that it is in the bottom corner that we face the biggest deceleration force (again I ignore the airspeed loss in the corner itself, but I think it will also be covered by what I want to do). Just before you pull up, the y axis accelerometer has -1g, the x axis accelerometer has ~3g outward, and the z axis is basically zero. Now once the corner is pulled, the y axis jumps up to be very large negative g's, and the z axis begins to show a deceleration growing to -1g when the plane is just attaining the vertical direction. Probably the x axis is decreasing (maybe a lot if you lose speed in the corner itself).

So if you look at these conditions, are they unique enough to know to add a power burst, and as the plane flies past 45o in the vertical, start to back off. The point is that it is in this part of the climb that we are slowing down.

Like I said, you would need to explore the other corners etc. in the pattern to see if you can find unique states where it is clear to add power, and other states where it is not.

Of course this doesn't address the wind problem, except that if it keeps the speed up as the plane climbs (and I am guessing that you will want ~2000 rpm pulses to fight gravity and corners), it is easier to handle the wind. Also maybe it would make the algorithms to handle the wind a little less critical.

I also think you have a point about the airfoils. I was surprised by your calculation in the Design forum on the wing drag (we need to make a pointer to that thread, or bring part of it here). So if you make the airfoil thicker, and that basically adds no drag to level flight (efficiency issues), but allows a nice sharp corner with minimal airspeed loss, it really sounds like a win-win situation and will help the situation in the upper part of the hemisphere.
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1645


« Reply #23 on: November 02, 2009, 09:59:37 AM »

Hmmm Dean, I do not think it is good idea to add power for exapmple in first corner of inverted square  Devil

The only think which I wanted to say, that making sich device is not trivial. I do not mean technically, I did all three types. The problem is "what and when" - means the difficult is to make rules when to accelerate and when to brake. Means the question which Peter wanted to solve by manual operating with throttle.

I think the only way is to combine several imputs and make "something clever" and easy to adjust device.
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1645


« Reply #24 on: November 02, 2009, 10:08:03 AM »

Alan,

>>>Just before you pull up, the y axis accelerometer has -1g<<<

This is what we WANT (means no speed to air change), but in reality you see somewhat less (because of speed to air acceleration), question is, how you want to figure up, how much it HAS to be (depending on its position) during the wingover for exmaple.

Tis is exactly influence of Gravity, which you cannot compensate without other sensor.
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #25 on: November 02, 2009, 10:42:41 AM »

Alan,

>>>Just before you pull up, the y axis accelerometer has -1g<<<

This is what we WANT (means no speed to air change), but in reality you see somewhat less (because of speed to air acceleration), question is, how you want to figure up, how much it HAS to be (depending on its position) during the wingover for exmaple.

Tis is exactly influence of Gravity, which you cannot compensate without other sensor.


I am not sure I completely understand what you are saying, but if it is basically that it is complicated, well I would agree. So it would be interesting to fly a flight with a 3 axis accelerometer to see if you can find ways to discriminate.

So just in case I wasn't clear in my previous post, by x,y,z I mean relative to the plane in level flight. So in this first corner, the "y" accelerometer would go from ~-1g to -10g (a guess for the centripetal acceleration during the corner) to basically "0g" during the wingover until the pullout, and then to +10 g for the inverted pullout corner to ~+1g for the inverted half lap (of course not quite a half lap).

So yes, you need to track what each accelerometer is doing during the maneuvers for what I would want to do. Since I haven't put any time into carefully looking, the question is whether you could find ways to uniquely (and easily) define the corners, mainly the bottom corners which go into vertical pullups.


Added: probably should change my x,y,z designations to "pitch", "yaw", and "roll" axes respectively to mean axes that are frozen in the airframe.
« Last Edit: November 02, 2009, 11:37:43 AM by Alan Hahn » Logged
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #26 on: November 02, 2009, 11:04:49 AM »

Sorry for the confusion, I only use ONE sensor, this sensor is pointing away from the pilot, in your explanation the x sensor. I do NOT use the Y or Z axis. So for Line tension I use a Line Tension sensor.......

The sensor AWAY from the pilot detects 0g before takeoff, roughly 2,7g in level flight, 2,0g at 45 degrees and 1,7g in the overhead position. In sensor readout values, 335, 515, 465, 430, loads of resolution on these readings.

In my solution there is only added power in the sphere above the 45 degree angle, NOT in the low maneuvres.

In the low maneuvres the wing alpha will get you the lift, the speed loss is around 45 degree angle not at the bottom, see the pics where I put the line every 0,24 secs. The wingover is flown crosswind in light conditions with a Super Tigre .46 just before I converted this model to Electric. The yellow V lines are for references so you can compare the travel in 0,24 secs.

It starts losing speed as soon as the weight of the model starts pulling the model down, so lighter is better but we knew that already.

The first 10 seconds after the motor started I stop the sensor from opening up the throttle, then I start controlling BELOW a 0,7g loss. As soon as the model is back ABOVE the 0,7 loss I stop adding power.

Hopefully you now understand how I shift my control line to 45 degrees angle.


* 46 measurement.jpg (8.88 KB, 465x348 - viewed 185 times.)

* 46 climbing.jpg (16.65 KB, 477x377 - viewed 214 times.)

* 46 speed in top.jpg (16.97 KB, 477x344 - viewed 202 times.)

* 46 down.jpg (17.12 KB, 458x356 - viewed 201 times.)
Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1645


« Reply #27 on: November 02, 2009, 01:28:04 PM »

hmmm ... sorry Alan, I was completely confused what the x, y and z is. I expected that your pull up is pull up from vertical to horizontal flight and thus y axis is longitudinal axis and thus I thought that -1g is what you expect to see just before that pull up  Layingdown

It did make sense, so I thought I understand.  Grin

Ok, I was too quick, so I will explain what I mean on that MY example. So imagine the model does a wingover. We have one sensor in longitudinal axis. I note that if its axis is perfectly tangent to the flight path and radius of maneuvers and we expect no drag changes in corners, then it can see only aerodynamic forces – in this case prop thrust minus drag.

So the question is what such a sensor sees during the wingover:
- after pull up – nose points up, prop thrust = drag sensor says 0
- little later – model is slower (because of gravity), prop thrust is little higher, drag little lower, senso see – say +0.1g (for sure more than 0 and for sure far less than 1g)
- on top of the hemisphere -  model is slow, accelerating, sensor will see “some number”,  could be easily close to 1g
- end of diving – modes is quick, but probably still accelerating and sensor sees number lower that -1g … say -0.5g

Now take data and try to interpret them without knowing that we are in wingover:
Sensor = 0 = constant speed
Sensor = +0.1 = model is accelerating, needs less power (hups, wrong)
Sensor = -0.5 = model is slowing down needs more power (hups, wrong again)

Other sensors:

Sensor in vertical axis say all the time 0 it can be sign of wingower, but it can be easily curved flight close to the ground, so it does not mean too much.

Sensor in lateral axis says from 3g to say 2g on top and then again 3g. Those 3g are also on begin and also on end of wingover and also in level flight – so question is how to recognize diving from level flight (or better that curved), because -0.5g longitudinal axis needs braking in diving but accelerating in level flight. Etc

And now take to the game the wind.  Devil

If you want realy know what the model does, you need 3 sensors for linear acceleration and 3 sensors for angular acceleration (or angular position). And only then comes the question when to add and when to subtract the power
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #28 on: November 02, 2009, 02:42:31 PM »

Of course at some point, with a complete inertial guidance system, the controller should just cut the guy holding the handle out of the loop  y1. He would be there just to apply some line tension!

By the way, the violet trace in my graph would represent the force you would see along the longitudinal axis of the plane. I called it the "tangential" force. Also remember that I didn't put the corner forces into my calculation.
Logged
Howard Rush
2016 supporter
Trade Count: (0)
Admiral
*
Offline Offline

Location: Bellevue, WA
Posts: 5614



« Reply #29 on: November 02, 2009, 04:39:07 PM »

You could use longitudinal acceleration (+ forward) + yaw acceleration (+ about z axis pointing down through the airplane for airplanes rigged to go counterclockwise), but you should only try this indoors.   
Logged

The Jive Combat Team
Making combat and stunt great again
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #30 on: November 03, 2009, 12:07:44 AM »

I prefer to make things easy, governor for the low Maneuvres, one sensor above 45 degrees.

In governor the motor is challenged to deliver power as soon as the airspeed is reduced, it will reduce power as soon as airspeed is above the requested rpm.

We use speed controllers that are developed for heli use, these mechanisms do not like a brake nor sudden increase of power.

When we go up the model slows down and governor will add power, at least that is what is is supposed to do. Some manufacturers deliver a spool up curve and maintain rpm power curve that is softened up a lot in order to preserve the gears. This makes these controllers to react too slow for the use we have.

It is no use building a 6 sensor operated computer controlled device with such a speed controller. It will not follow no matter how hard we push the paddle.

In control systems we call this the Time Constant or Tau of the system. We can measure this by flying at 5,5 sec/ lap and then open up the throttle to full power and measure how long it will take to reach 5,0 sec/lap. You will notice that there will be a dead time first and then it will start to accelerate. If this Tau is too high the it will be impossible to achieve a proper reacting system. If you want to learn more about this then go to wiki and read about the Ziegler-Nichols Method

We measured the Kontronik speed controller and found out that is is very slow on spool up and does not maintain rpm when put under pressure. A kontronik spreads around 400 rpm differences, the CC45 within 100.

So if you lose speed in the low maneuvres there is a chance that your speed controller does not support governor the way we need, try a CC45 set it aggressively and feel the difference.


* Kontgronik 1 spool up.jpg (82.19 KB, 822x555 - viewed 187 times.)

* kontronik 2.jpg (81.59 KB, 822x555 - viewed 184 times.)
« Last Edit: November 03, 2009, 01:27:29 AM by Erik Janssen » Logged
Igor Burger
Trade Count: (0)
Admiral
******
Offline Offline

Posts: 1645


« Reply #31 on: November 03, 2009, 02:14:04 AM »

Erik, I do not know if you write it to me, but I certainly know lot about regulation, since I studied it 5 years  Grin  ... you are right that "some" ESC can make problem, but for sure I use controller which is quick enough, it is lost time to make regulators with slow ESC on end.

on this video you can see that the regulator is able to react even during first corner ... the best example is houglass:

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

Posts: 1645


« Reply #32 on: November 03, 2009, 02:19:18 AM »

and now as I see that video - it is clear how the model needs strongest pull on the begin of vertical eight, when the gravity has strongest effect, and not on top of hemisphere when model goes flat Devil
Logged
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #33 on: November 03, 2009, 07:35:10 AM »

I think we are looking into two different challenges

a.   Loss of speed as soon as the model goes up
b.   Loss of groundspeed due to headwind in governor

My sensor helps to compensate for the loss of groundspeed in governor when the model encounters headwind. This allows me to fly rather slow in windy conditions. As it switches around 45 degrees it throttles down in time even in the hourglass.

I do not experience the loss of speed outdoor as severe as the effect indoor, therefore I have not been searching to solve that. For symmetry reasons I want the model to climb faster and come down slower. Hopefully I will achieve that by a lighter model with more drag.

I will look into this area with a closer look as soon as this model is finished.
Logged
Tim Wescott
2016 supporter
Trade Count: (0)
Admiral
*
Online Online

Age: 53
Location: 122.4767W, 45.3116N
Posts: 8689



« Reply #34 on: November 27, 2009, 10:44:00 AM »

The system takes away the power very fast, I have NO overshoot in any of the maneuvres, the wingover is quite easy to fly.

The only difficult point is the forward squares, when you fly through the wind you lose line tension and the motor picks up just before going into the dive. That makes you practice to fly downwind. If your position is right this does not occur.

An electric motor has no brake, in the the test diagram you can see the rpm change when the prop load is doubled and released. The overshoot is influenced by the weight of the propeller. CC45 is good, Kontronik very bad. I have not tested other controllers.

Braking a dive can only be achieved by a very light airplane with lots of drag.

Correction:  An electric motor itself can act as a brake, but you'd need a speed controller that would make that happen.  This effect is quite apparent if you have a high-ratio gearbox on a motor, such as you find in some industrial servomechanisms (or RC servos, for that matter) -- just try turning the shaft with the motor leads shorted vs. the motor leads open.  But you'd need a speed controller that's designed to actively brake the motor when it over-runs; I have no clue if there's one out there for the brushless 3-phase motors we use.

Are you measuring line tension at the bellcrank, or are you inferring it from the radial acceleration of the plane?  I've been thinking of doing this with a gassie and an R/C carb, rather than learning how to finagle the motor indirectly as is usually done.
Logged

AMA 64232

The problem with electric is that once you get the smoke generator and sound system installed, the plane is too heavy.
Dean Pappas
Trade Count: (0)
Admiral
******
Online Online

Age: 59
Location: Central New Jersey
Posts: 1190

Welcome to the Stunt Hanger.


« Reply #35 on: November 27, 2009, 10:58:19 PM »

Hello Tim,
None of the ESCs anywhere on the market offer an RPM governor that will transition from driving, through freewheeling to either regenerative or "plugged" braking. Folks like Erik know this all too well. At best we get freewheeling, though the loop gain and transient response of even the best setups never seems to hit a true freewheel anyway, so a real brake would not help us.

Even the geared F3A Pattern setups need the regenerative brake for good down-lines in some places, while most of the schedule works well with a 500 RPM "idle" as the lowest setting.
Regards,
Dean P.
Logged

Dean Pappas
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #36 on: November 28, 2009, 01:49:54 AM »

True, connecting the leads will brake the motor. Sadly my ESC does not do this in rpm change. We tested Castle Creations Phoenix 45 and Kontronik Jazz. The Kontronik is software delayed so bleeding off rpm goes even slower.

Technically it should be possible to make an electronic brake on the prop when we want the rpm to come down. Just glue two magnets to the back of the spinner plate, put a coil just behind and let the processer short circuit this coil when it thinks it is neccesary to bring the rpm down.

What we do not know is whether we need it, if the weight of the model equals the drag there is no need for such a feature. The amount of drag is a function of speed, so if we go up faster the drag will be higher and the acceleration in the downleg smaller.

In 2006 when the shut off was not yet available we used a disc brake to stop the prop. The braking force was spring loaded. It is in the museum now.


 


* reminmodel3.JPG (949.66 KB, 1632x1224 - viewed 415 times.)
Logged
Tim Wescott
2016 supporter
Trade Count: (0)
Admiral
*
Online Online

Age: 53
Location: 122.4767W, 45.3116N
Posts: 8689



« Reply #37 on: November 29, 2009, 07:20:41 PM »

Hello Tim,
None of the ESCs anywhere on the market offer an RPM governor that will transition from driving, through freewheeling to either regenerative or "plugged" braking. Folks like Erik know this all too well. At best we get freewheeling, though the loop gain and transient response of even the best setups never seems to hit a true freewheel anyway, so a real brake would not help us.

I figured that -- I was only being nit-picky about the wording, not about what you can do with what's currently on the market.

Back in the brushed-motor days it was common for an ESC to have a brake, to work well with a folding prop on a powered sailplane -- I take it that that feature has gone by the wayside?

But control line stunt is such a huge market, surely someone will take this up!!!  Wink
Logged

AMA 64232

The problem with electric is that once you get the smoke generator and sound system installed, the plane is too heavy.
Alan Hahn
Guest
Trade Count: (0)
« Reply #38 on: November 29, 2009, 08:29:03 PM »

Tim,
There is a brake which is applied with zero throttle. It will stop the prop.

But let me comment. My motor never overspeeds the set rpm in the down leg of a maneuver--e.g. the rpm stays constant. Also the power in the downleg is far from being turned off, so it isn't a problem with our motors at all. Now It is possible that I might choose a prop that in the down leg makes the applied motor power go to zero, but I haven't seen it yet.
Logged
Dean Pappas
Trade Count: (0)
Admiral
******
Online Online

Age: 59
Location: Central New Jersey
Posts: 1190

Welcome to the Stunt Hanger.


« Reply #39 on: November 29, 2009, 09:43:36 PM »

Exactly, Alan!

later,
Dean
Logged

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

Posts: 1645


« Reply #40 on: November 30, 2009, 12:54:10 AM »

Alan is right, but this thread is about controlled rpm, not constant rpm, we want to brake the model downhill ... but in any case, it is true that we do not need to brake THE MOTOR we need to brake the MODEL, and since we have fixed pitch prop, the only way is to set lower rpm if we want actively brake, and that is the problem, the thrust of the prop (positive or negative) depends on RPM, so braking cannot be so effective as pulling, and that is the problem of difficult braking, not recuperation, or shaft brake etc. ...

that system is doable, but you will see that the limitation of such device is not hidden in "electricity", it is in aerodynamics, or better word can be "controllability"  Devil
Logged
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #41 on: November 30, 2009, 03:51:58 AM »

Motor never overspeeds rpm is not true, a motor will change rpm a little between up, down, head and tailwind. This can be seen when you enlarge the rpm graph.

In this example we used an AXI 2826/10 and 12,5x6 Graupner CAM, rpm dropped about 300rpm in maneuvers with the CC45, using the Kontronik Jazz changed this to a dramatic 600. The more stable the rpm the sooner governor reacts to pull you out of a low square. (graph shows loops wingover and squares) We tried a 13x6,5 APC-E and found out that it was too big for the AXI, rpm drop is very big and unflyable.

A Spitz 3020 has less copper and build quality and loses 570 rpm in a square, the AXI performs a lot better in flight.

The batteries can be limiting too, if they are not true 20C or cold you may lose more than 500 rpm in the first square of the maneuvers. And the faster it goes up the more drag it will have at the top so less braking is needed.

More drag is an option. Igor's new wing profile may be helpful although I do not understand the theory behind it. I have chosen to increase the size of my model by 10% wing area and reduce the weight with 10%. So 10% more drag and 10% less weight that pulls it down. That should be a step forward.


* rpm drop in detail.jpg (89.09 KB, 843x584 - viewed 183 times.)

* AXI prop limits.jpg (35.95 KB, 481x355 - viewed 165 times.)
« Last Edit: November 30, 2009, 06:36:46 AM by Erik Janssen » Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #42 on: November 30, 2009, 09:22:35 AM »

Erik,
I clearly am NOT seeing this on my Castle Creation Phoenix ESC with governor. Here is a plot, somewhat busy of a wingover. Wingover starts about 236.2 s on the x axis, and ends at ~245 s on the upright pullout.

White is airspeed (mph --sorry!), red is altitude (feet --sorry again!), violet is power input to motor, and green is the rpm. It looks like the minimum rpm bit is about 50rpm. So in the first vertical corner I might see a ~100 rpm glitch downward, but in the dive I don't see any real evidence that rpm is increasing at all. Ignore the glitch at 240.6 s---there is an issue with my Eagletree that I am trying to sort out with the company!.

The other thing to remember, quick glitches don't really do very much to affect items like airspeed. They don't operate for long enough to matter. I believe I am running a moderate governor gain on my CC ESC.


* ENoblerWingOver.png (121.09 KB, 1082x662 - viewed 173 times.)
Logged
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #43 on: November 30, 2009, 12:01:06 PM »

From the pattern I would say white is rpm and green is volts.
My logger only has rpm, volts and amps so less chances of confusion
7860 to 8160 matches my numbers
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #44 on: November 30, 2009, 02:26:27 PM »

Erik,
I have all the quantities (volts, amps, rpm, airspeed, altitude, watts, mAHr from the Eagletree, plus one (I call Tangential Acceleration) I derive from the airspeed difference in my grand array of datapoints. I just blanked out (made the trace transparent) everything except airspeed, altitude, rpm, and power, as I mentioned.

Yes the plot is busy, but if you try to correlate cause and effect, it is useful to have everything on the same plot.

I do still have some synchronization concerns--I try to use a video taken of the flight to offer visual markers for when the maneuver actually starts. However I discovered that the Eagletree seems to count 59 seconds for every 60 seconds of the video. Adds some complications to the effort.

I actually have found it easier to make my Fourier analysis of glow engines, since the two sources (video camera and a separate audio recording from the center of the circle) synchronized quite well by using sound. With the Eagletree I try to video the prop starting and comparing it to the Eagletree rpm recording. But there do seem to be some problems there, at the 0.1 to 0.2s level.

However I think I'm going to have to build my own processor/datalogger/throttle-timer for that. Just trying to work up the courage to make the plunge! That will include a 3 axis accelerometer!

Logged
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #45 on: November 30, 2009, 02:45:58 PM »

If you want to build your own logger I suggest you start with the Logomatic V2 from www.sparkfun.com it has a parameter file on a micro SD card and 8 analog inputs. My timer starts and stops the logger. On the logger itself I attached 3 axis analog accellerometer and a shunt from my ESC for U and I. Shunt is too from Sparkfun

It works great although I have not figured out how to filter the input. In my timer I take 512 samples and then divide to get an average value to work with. It seems that they wait 1/100 of a second and then take one reading. The red line is Amps and the green a calculated consumption of mAh.

The source code is available so that should help to save some time if you are smart enough to understand how they did it.

I use a powerLED on the model that switches on and of and tells me what I am doing, so I can see points where it kicks in and when it is at full throttle. I only use video for look and feel and counting symmetry and laptimes.



* Logomatic.jpg (31.12 KB, 566x330 - viewed 201 times.)

* CC45 with shunt.jpg (187.31 KB, 872x584 - viewed 258 times.)
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #46 on: November 30, 2009, 05:58:43 PM »

That's exactly what I've been looking at. It looks like I can get at the I2C interface by sacrificing the serial input LED's. That will let me readout the 3 axis accelerometer (+16g's and the EagleTree Airspeed and Altimeter sensors. If I really work, I can get an input to count the rpm counts. If I really get ambitious, I could conceivably program a PWM output for the ESC! But that is thinking ahead.

Will take some programming to set it up, but winter is closing in on us here in the Chicago area.
Logged
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #47 on: December 01, 2009, 02:58:53 AM »

The example comes from a pololy 6g sensor jumpered to 4g wired directly to the logomatic data logger. With 8 analog inputs there are enough. XYZ UI what else?

The technology to use a txt file for reading parameters is easy to use and can be extended with other parameters.

I tried to filter the signal with coils, resistors and capacitors but did not succeed to get rid of the noise. So you have to do someting in the software. Can you find where they determine the frequency, if there is a waiting time and one comparison we can change that into 512 sum and devide to get an average. At 20mHz you will get 22 values per second this way.

I do not know how eagletree measures speed, but if it is a pitot tube I wonder if the reading is accurate enough. We hardly fly straight lines. From the pylon boys I understood they had to fly at least 50m flat before the reading was stable.

Winter brings more wind than I can handle, even with the sensor. I will focus on a new plane not software this winter. After 16 years of service my model is falling apart and needs retirement.
Logged
Alan Hahn
Guest
Trade Count: (0)
« Reply #48 on: December 01, 2009, 07:39:27 AM »

Erik,
One thing to remember is that averaging is a low pass filter! The only thing of importance is that you sample it fast enough to avoid aliasing (which will generate fake low frequency signals). I think you are doing that.

The particular accelerometer I am looking at does the filtering and digitizing internally, based upon the readout rate. So if I choose 50Hz, then it will have a 25Hz filter on it. But since it is already digital, I need the I2C readout. The chip on the Logomatic does support I2C, but you need to setup the device so that the mutiplexed I/O pins support it. I'm doing this since I'd also like to get my Eagletree sensors into the device too.

Yes I don't know how accurate the airspeed indicator is. It seems to be off by about 15% in level laps, indicating 48mph instead of the 54mph (24 m/s) I calculate from lap speeds. I'm not sure if it is the low speed calibration, the crab-out angle, or something else. This speed is the same as I measured with a similar device a couple of years ago, so it is something systematic. However I am guessing that the relative airspeed isn't too far off. It certainly follows the maneuvers quite well, but I am guessing that during the actual corner, the pitot is getting misleading information due to the angle of attack change.
Logged
Erik Janssen
Trade Count: (0)
Lieutenant
***
Offline Offline

Posts: 58


« Reply #49 on: December 01, 2009, 09:33:04 AM »

I sample 512 conversions to take an average at a rate of 22 per second, that is 11264 per second. If you want the logic, go to www.pololu.com, select the 168LV robot controller, go to resources and pick Demo3, here you have the ADC conversion, LCD readout and a pwm setup. Copy paste the code and modify the adresses following the data sheet from the processor you want to use. Before I learned about timers I used a multiplication of 300x22 cycles to make my timer run for 5 minutes.

In the old days you had to learn how things work, nowedays with the internet you can copy paste this knowledge. I found it very useful to follow the tutorials from www.avrfreaks.com specially the eeprom memory.

Building a logger was a bridge too far, I succeeded in getting the device of Roland Riegel operational but was unable to write the logic due to a lack of C knowledge. Then I found the Logomatic logger which is rather cheap and has all the functionality. If I can find the time I will look into the sampling part of this device.

We also built a device with a digital accellerometer from parallax Memsic 2125 Dual-axis Accelerometer, the logic works fine but we have not flown it yet. As the analog device works very well and sounds very smooth I do not think we will change it.
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 1.1.21 | SMF © 2015, Simple Machines

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