Dual Sensor vario development

[RightRudder]

If anyone is contemplating a board design I think this sensor would be good for total energy compensation.

http://www.digikey.ca/product-detail/en ... ND/1168436

This would lend itself nicely to also allowing airspeed data as part of the telemetry if desired.
My other wish list item would be a bootloader and usb support.

[rainer]

If anyone is contemplating a board design I think this sensor would be good for total energy compensation.
http://www.digikey.ca/product-detail/en ... ND/1168436
This would lend itself nicely to also allowing airspeed data as part of the telemetry if desired.

I had a quick look at the datasheet for that sensor. it´s operating temperature is only 10C to 60C which is not suitable here.
and to be honest. i´d rather play with a TEC tube for energy compensation. (even though this would probably interfere with the altitude readings.)
One of our main goals for this project should be to keep it as simple as possible to build. So we should try to stick to ready build modules like the arduino pro/nano and the ready made ms5611 modules.

My other wish list item would be a bootloader and usb support.

There is an Arduino bootloader on it of course. if you want a usb port, you can just build it using an Arduino nano instead of the Arduino pro mini.Its only a timy bit bigger but works as well.
Actually i´ve got to different setups here. One nano on a breadboard together wit a 3.3V MS5611 module that i am developing on and one with a Arduino pro mini 5V 16Mhz together with a 5V MS5611 module which is the yellow device on my pictures.
rainer

[RightRudder]

I don't do much soaring when it's below 10 deg.C so I figured it would be alright. That temperature range is only for the guaranteed accuracy, the operation will be fine to much lower temperatures.

If you are thinking of making a TEK tube, a small brass tube blocked at the top end sticking straight up from the tail and having a small hole drilled at the back (straight back from the tail) will work pretty well, but then you are building a netto vario. You'll need to do some data averaging since the A/D happens inside the pressure sensor and you don't have the option of any analog low pass filtering. Yeah, the TEK probe will give a very noisy signal. I think just a straight energy compensated vario gives more information to the pilot which is of value and also is easier to get set up than a netto since you don't know the polar of your model very well AND a forward facing pitot has two advantages. It has a smooth almost noiseless signal due to the natural stagnation point of the air ahead of the pitot opening, and you have an optional new metric, airspeed.

[Flaps 30]

Whilst sitting here pondering the meaning of life and the Universe, my thoughts went onto the Vario that is sitting near me with its red eye looking my way.

I was wondering if it was possible to hook up second pressure sensor that would be in a plastic tube/bottle with a pipe measuring the air pressure/airspeed. In doing so it would be possible to have something approaching a Total Energy Compensation system. That would to a large extent remove false triggers/readings due to atitude (climb/dive) changes due to pilot input to the controls.

Just a quick look around reveals this. ----> http://shop.rc-electronic.com/e-vendo.p ... 403&p=2403

I know that a lot of systems rely upon fairly complex probes/bottles and pipework and that puts a lot off the idea of doing anything along the lines of a conventional system, whilst using electronics and wires + a soldering iron is much more attractive.

[Henning]

Hello flaps30,
at least the developers of the MS5611 sensor had your idea in mind when they provided the device with 2 selectable I2C adresses...

So its easy to connect 2 of them to the same sensor bus in a vario module - pitot compensation could be done via Software then.

As far as i know no lowcost vario did this up to now... Who does the 1st step?

Regards
Henning

Gesendet von meinem Nexus 7 mit Tapatalk 2

[Rob Thomson]

Whip cracking!

So exactly how do we deal with the hardware side?

Rob

Sent from my GT-I9300 using Tapatalk 2

[rainer]

yes, i´ve got one of those drotek breakout boards here. it can be adjusted to 2 different addresses.It would probably require a different aproach to reading the sensors though.
The sensor needs aprox. 10-11ms delay to generate an output on the highest resolution. As we need to read it as fast as possible to reduce noise via the filtering mechanism, a second sensor would reduce the reachable resolution and speed. So we need a new aproach to reading the sensors asynchronously.

The other thing is that this kind of approach will require a huge amount of testing before the calculation will work properly. It might be possible though by creating an adjustable calculation that can be remote controlled via one or 2 pots in the air. we can do that already for the sensitivity, so this should´nt be a problem. it would dramatically improve the possibilities to tune this kind of algorythm.

The next thing would be the mechanical component. we would require a good way to attach pressure tubes to our sensors. Anybody has a 3d Printer here? it might become handy to create some custom parts.

If we start something like that it should be seperated in a different branch, not to interfere with the single sensor aproach.

I´m going to be away for the next week, but i could do some tests with dual sensors when i´m back and have more sensors at hand.
can anybody find out how to modify the GY-63 module to a different I2C address?

[Flaps 30]

I´m going to be away for the next week, but i could do some tests with dual sensors when i´m back and have more sensors at hand.
can anybody find out how to modify the GY-63 module to a different I2C address?

From what I see it seems that the GY-63 Module is like mine with a CSB pad... The data sheet on the MS5611 says "In I2C- Mode the complement of the pin CSB (Chip Select) represents the LSB of the I2C address. It is possible to use two sensors with two different addresses on the I2C bus. The pin CSB shall be connected to VDD or GND (do not leave unconnected!)." My Sensor board 'CJMCU' has a CSB pin, so I guess that most MS5611 boards will do the job in dual sensor mode.. I also guess that I should do something with my CSB pin. as it is floating at present.

[Rob Thomson]

Could I throw in a curve ball...

Theoretical an triaxial acceleration sensor could be of use at detecting /filtering stick lift. (this is just my musing, and not a hard fact)

No doubt complex maths would be required - but may be a simpler hardware solution?



Sent from my GT-I9300 using Tapatalk 2

[rainer]

We require true air speed or better change in air speed. i don´t think a triaxial sensor would give us the required information.

[Rob Thomson]

Ok.. So another curve ball just thinking out the box.

The iPhone has an app that measures airspeed based on the sound level of the wind against the microphone. Supprisingly accurate.

Would something similar make a potential sensor?

Just thinking out the box to save some machining issues for end users

Sent from my GT-I9300 using Tapatalk 2

[Kilrah]

Hmm, no quick-reply here?

The most affordable Pitot tube I know of is the Eagletree one (bottom of page):
http://www.eagletreesystems.com/Standal ... dalone.htm

Ideally the differential sensor used for speed should have pressure ports...

[rainer]

here are some that might be interesting:
http://www.buildyourowndrone.co.uk/Air- ... s-02dp.htm
http://www.wishyourgift.com/hobby-equip ... 002dp.html
or the sensor on its own ( see rightrudders earlier posts): http://www.digikey.ca/product-detail/en ... ND/1168436

[Rob Thomson]

Or...

http://www.wingedshadow.com/howfast.html

[Rob Thomson]

Looking at the sensor used.. there are a range of commercial ones around.

http://www.xicoy.com/catalog/product_in ... ucts_id=93

A really magic resource would be this:

http://code.google.com/p/ardupilot-mega/wiki/Airspeed

The sensor can be purchased for $20 from

http://store.diydrones.com/Breakout_Boa ... 004-03.htm

Data sheet is here:

http://www.freescale.com/files/sensors/ ... XV7002.pdf

Rob

[Rob Thomson]

here are some that might be interesting:
http://www.buildyourowndrone.co.uk/Air- ... s-02dp.htm
http://www.wishyourgift.com/hobby-equip ... 002dp.html
or the sensor on its own ( see rightrudders earlier posts): http://www.digikey.ca/product-detail/en ... ND/1168436

Looks like the same sensor?

Initial check seems that it is a simple voltage output from 0.5 -> 4.5v to show speed? (Not sure I am reading the guide correctly!)

[Henning]

Hello vario fans,

these "2-in-1" sensors look rather big to me - so I would spend some more seconds about the dual concept, based on the great work on the MS5611 vario rainer already did...

Multiple sensors on the I2C bus are common practice (most MultiCopter IMUs work this way), so I think the "dual MS5611" hardware wouldn't be so difficult. I just did a Eagle layout sketch, based on my "Jeti-EX-Vario concept" (with a single MS5611) I have running on the breadboard. Assume these dual MS5611 should work in this constellation. I2C experts: any idea?

Regards,
Henning

[Flaps 30]

My vote would go to the dual MS5611 arrangement. Ideally that would still work with the Arduino Pro Mini. That way it is easy for people to 'update' from the simple vario to the dual system without too much outlay. I have a lovely Tic Tac plastic box that would serve as a pressure chamber for the airspeed sensor.

[RightRudder]

Guys since Rainer got the Kalman filter working so nicely we may only have to put the openXvario in a small bottle sealed where the wires come out through the lid and also a tube. Hook the tube to a vertical cylindrical probe with a hole in the back (this is a TE probe) and presto. a 3mm aluminium tube will work well. I posted references to NASA papers detailing this type of probe in the main thread of the openxvario here:

viewtopic.php?f=86&t=2780&p=38703&hilit=TE+probe#p38703

If you put the vario in a bottle with this kind of probe you now have a vario that does not respond to airspeed changes. This is known as a netto vario or airmass vario. The audio response of the vario is directly related to the air you are flying in and has nothing to do with the energy changes of the aircraft due to up/dn elevator deflections. I was initially skeptical of this approach because the TE signal is noisy (turbulence on the back side of the probe) but the kalman filter may do wonders for this and it will sure simplify things. We will try it and see!

[Flaps 30]

My eyes glaze over when presented with a whole stack of maths, combined with having to make holes that have to be a specific size or angle in the back of a tube before it has any chance of working. Are there any other holes in the tube? Sorry if I sound a bit cynical or reluctant to go the way of doing things mechanically when (with modern devices) it is easier to do it by electronic means and by doing so have more control over the precision of the vario. By measuring the relative airspeed with the other sensor and throwing that into the equation will give you the same result and probably better. If nothiing else. You have control over the final result by just changing code.

Yes doing it with a TE probe might be the way that some people wish to go. Just not for me. Mind you. I cannot code for love or money..

[RightRudder]

Here's the thing. You can certainly compensate the vario by measuring airspeed and then scaling that value with a certain gain factor to tailor the response. The eagletree vario does exactly this but is not as sensitive as oXv. But what should that factor be? If you think making a probe will give you a headache then get out your extra strength tylenol for the digital approach. It ain't easier. The probe does the math for you. It isn't hard to make and isn't all that critical even. The last page of the first paper gives you the drawings. Or you could make a probe with a 33mm length of 3mm tube with a 1mm hole drilled 6mm from the end. Block the end and keep it square. For best compensation, angle the probe 20 degrees forward of vertical. Hook the tube to a chamber and put the vario inside. Seal it well. Done. If it's too noisy a small capillary tube in the line between the chamber and TE tube will do wonders. I could do this in an evening but coding a digital compensation with dual sensors...mmm I might need to use up my vacation for that one.

[Flaps 30]

Well.. I hear what you are saying. Reading through some of the information revealed that things were not that simple..

From the paper "As shown by wind tunnel results , the position of the orifice with respect to the end of the tube was very important. For final adjustment to a given installation , it was found that drilling the hole slightly farther from the end of the tube (approximately 1/32-inch) than twice the diameter, allowed easy adjustment a f t e r initial flight test by filing off the end of the tube in small increments and retesting until achieving good compensation. One or two iterations usually sufficed.

As mentioned earlier, the shape of the end of the tube affected compensation and care had to be exercised to bevel the edges very slightly. It was also imperative that a good seal exist after filing; it was found very worthwhile to check for leaks after any such filing adjustments because a tiny hole in the soldered end produced drastic effects on compensation."

So from that I feel that it really isn't that easy to get things to work and they also mention foot long pipes between the probe itself and the remaining pipework to the instrument. They also mention probe heights of six inches above the fueselage... So that is more rather restrictive items to take into account.

It also mentioned that the tests showed that they got good total energy rate information over a flight range from 40 to at least 150 mph. Most of my gliders tend to be at the bottom of that range and known to go below it. So I'm not so sure how good the mechanical system would be at the lower speeds we tend to be at.

As for how to do the calculations... Not sure on that. Some more reading needed here to work that out and I am sure others have a better handle on this than me. I do see a bit of empirical testing required to make this work using two sensors. What we do know, is that when the airspeed falls in normal flight that we have put in up elevator and that decrease needs to be fed into the system to compensate for the gain in altitude to keep the output at the same value as before the elevator input and same applies for down elevator with the corresponding increase in airspeed.

[RightRudder]

Yes you can read that in different ways I guess. One or two iterations doesn't seem like much to me. Dr. Drela seems to think that the probe isn't too critical for an RC glider according to what he said on another forum. These paragraphs have to be read considering competition sailplane pilots tend to obsess over minutia and be total perfectionists. WRT your comment about the probe length etc I have already taken into account the lower reynolds numbers for the model with the smaller 3mm tube and kept to the general guideline of probe length 11 diameters, hole distance from end two diameters and hole size 1/3 of tube diameter. According to a thread on RCGroups like this one, the compensation works well!

[Flaps 30]

Okay. Between the two of us we could bang on about the mechanical method vs the dual sensor method till the cows come home. Both IMO have advantages and disadvantages. This thread hopefully will lead to a way to get the two sensor method to work. Failing that it may help in some way to improve the single sensor vario.

My thoughts today just rested on how to get what would be a neutral starting point whilst flying a specific model glider with a clean configuration in (ideally) still air. At the so called neutral point we will have a known airspeed and if you really want it a known sink rate. That sink rate figure could be used as a zero/centre setting for the output of the unit.

If the airspeed increases the pressure on the airspeed sensor will increase to a set point if the speed stays the same. The pressure on the Vertical Speed (VS) will also increase and continue to go up as the aircraft looses altitude. So we now have a situation where both pressures have risen and the VS pressure rises all the time if nothing changes. So the ouput after doing the maths between the two should be such that no lift or sink is indicated.

I have this picture in my head of a slope that has its angle set by the airspeed (higher airspeed=faster descent) and the output of that slope angle taken away from one of the sensors to get the final result. This logic is on the basis that both sensors increase pressure when in a dive, so take one away from the other to get the final figure, which should be zero if the line slope does its job.

The other way around for when the airspeed reduces and the aircraft climbs reducing the VS pressure should also ouput zero.

That is as far as my thoughts have gone. I'm sure they are flawed as they are thoughts before I was really awake and before the first coffee.

[RightRudder]

An airspeed indicator displays the difference between total pressure (measured at the pitot) and the local air pressure in the free stream (static pressure). The total pressure is the pressure at pressure coefficient Cp=+1. If you use an absolute type pressure sensor rather than a differential type then you are measuring total pressure directly, not airspeed for which you need a differential type sensor. The TEK probe has a hole at the back in the suction area behind the probe and measures the pressure at Cp=-1. So all you need to do is hook up a pitot to an absolute type pressure sensor and subtract its output from the static pressure measured by the MS5611 in order to compensate the vario. If you scale the value by a gain factor so that you get a compensation value which ranges say between -0.8 and -1.2 you can adjust the compensation to how you like it. It is better to be slightly under compensated than over compensated. This is the electronic analogy of moving the location of the hole on the TEK probe relative to the end of the probe. If the vario gives an indication when the elevator control is moved then it is under compensated and therefore you should move the hole closer to the tip or in the electronic case make the scaling factor a larger negative value. The calibration is where it gets ugly. Both pressure sensors should be the same type and have identical response characteristics.

[RightRudder]

Actually now that you've got me thinking on this you could do it with just one differential type sensor but just hook the pitot up to the negative input port of the differential sensor! That should avoid the calibration problems but you don't have the ability to adjust the compensation. Shouldn't need to though because you should get a nice clean noise free unity value compensation signal from the pitot. I've been using some HSC type honeywell sensors for another project which would be perfect for this. A bit expensive but you get 14 bits of resolution for a full scale sensitivity of 2.5mbar (that's about 40 times the sensitivity of the MS5611 !) They have nice little hose barb connections that work perfect with soft silicone fuel tubing. Here's an image, that's the HSC sensor at the bottom.

[Rob Thomson]

What sort of price are they?

Get the feeling there are to distinct systems here.

OpenXVario - easy and cheap
OpenXVario++ - easy, little more expensive, super accurate?

Rob

Sent from my GT-I9300 using Tapatalk 2

[RightRudder]

They are about cdn$50.00 each! Ouch right? But they are a different breed of sensor and are quite well internally compensated against temperature and non linearities. They seem to be quite immune to 2.4GHz RF as well. Very sensitive, it can detect normal air currents in a room! This is not another vario I'm building but something completely different. The sensors would be ideal for compensating the vario if you wanted to do it by the two sensor method. I feel a little sheepish as I think I was the one who started this thread but that was back before the kalman filter which should make the TEK probe work nicely. If that's the case I probably won't bother doing anything else.

[Rob Thomson]

Well.. Guess that is relatively.

Consider... A good competition f3j model can cost you upward of $1500

So while $50 may seem like allot... I would probably not use one in more than one or two models.

Sent from my GT-I9300 using Tapatalk 2

[RightRudder]

Yeah I wish I could afford a Maxa or an e-Supra