Feint 2

[Tom Harper]

The goals this time:

Use the FEINT KISS navigation software ported to a larger processor

Use the SparkFun MSP430f1232 applications board with the processor already mounted on it ($15)

Provide both RS232 and USB communication to PC

Provide Data Logging

Minimize cables and connectors

Minimize hand wiring

Reduce size

Include leveler and altitude as control operations

Fully autonomous operation

Low cost

Low parts count

The finished system has to be small enough and light enough to fly in my .25 powered foam and paper camera plane.

__________________________________________________ _______________

The MSP430f1232 controller has lots of I/O pins, SPI, UART, and an A/D converter. Plenty of resources for the tasks at hand which eliminates external parts. The problem is that the thing is in an eentsie beentsie package that I can hardly see, let alone solder to. But -> SparkFun <- to the rescue. They have an apps board with the part, JTAG connector, crystal and build area for 15 bucks. Can't beat that!

This system is divided into two physical lumps. The receiver/control board forms one and the servos are glued together to form the other. The aileron servo is outside, but it's connector is on the servo board. Communication with the servos is through a serial link to a DMOS sipo. This resolves any level setting problems, drives LEDs (handy for debug) and requires only 3 communication lines.

The need here is to reduce connections. I've got old Airtronics, new Airtronics, Futaba and stuff I've made myself. I want one connector that does it all. Also much of the room taken up by the servos is in the mounting. By lumping them together the result is smaller.

The same is true of the receiver. Most of the receiver cabling is redundant. Especially when the receiver is not powering the servos. I first considered using only the signal lines without the power runs. With that in mind why not go further back in the receiver and get the serial composite signal before the receiver decodes it into individual channels. That signal is easier for the controller to handle and it only requires one communication wire.

A couple of minutes with a scope probe found the composite signal (arrow photo 1). The scope snapshot in photo 2 shows the composite signal. The control period is the time from one rising edge to the next. By presenting this signal to the interupt line I can pass it through the processor to the DMOS serial in parallel out shift register (sipo). I cut the aileron land on the receiver board and brought the composite signal out as channel one. So I get power, ground and the 7 channel signal on 3 wires.

Photo 3 shows the servo control board with the sipo, the individual servo cables, the 5 wire control cable, the 3 wire aileron connector and LEDs. No resistors yet. Photo 4 shows the 3 servos lumped together with RTV. It makes a compact package. These two had broken lugs. A couple of lugs would be handy for mounting.

Photo 4 is the receiver/control board combination. Only have the receiver and power connectors on it so far.

Photo 5 is the system block diagram.

[Dan_Jones]

Tom,
We are all looking forward to FEINT 2!

I noticed you will be needing a USB to TTL serial adapter. As those happen to be one of the cornerstones of my business, I will be happy to provide one or two for your development.

I know this doesn't help the low cost or reduced part count requirements, but I would recommend SD or MMC cards for the data logging. I have found that the Sparkfun DOSonChip module works, but it is slow. You may be better off using the MMC FAT library routines built into some compilers. I am using MikroC (www.mikroe.com) right now and I really like their support. Of course a dedicated chip is used to manage the data logging.

Dan

[Tom Harper]

Dan,

Thanks for the offer. Let me get a little further along before I take you up on it. I got an Arduino gadget from SparkFun. I intend to put it on a small board external to the model for PC communication. I am sure that it can be improved on.

I'd like to use SD or MMC in the future but I think a NV memory chip will be adequate initially. I want to run some experiments on using the electrostatic field for leveling.

Right now I'm working on getting the sipo to run the servos.

Tom

[Dan_Jones]

Hi Tom,
I've never heard of using an electrostatic field for leveling model aircraft. Does this use the earth's magnetic field? Just let me know when you would like those cables!

Dan

[Tom Harper]

Dan,

It's Ben Franklin's kite experiment. He flew the kite on a cloudy day, but not in a storm. The arc to the key took few thousand volts. The field is about 180 volts per meter. There is over 300 volts from your head to your feet but the impedance is very high and the field wraps around objects, so you don't get zapped. That's a lot of signal for leveling. It takes a very hi Z input amplifier. I have been hauling around some 100M resistors for years. They may be high enough. I hope so because I don't see any hi values in the catalogs.

Maynard Hill came up with the idea years ago. He had a rig that was looking for clear air turbulence. He found a consistent error in the strip charts. Finally noticed that the error followed the control movements of the airplane carrying the equipment. The strip chart was plotting the earths electrostatic field. Old Ben was right after all.

Tom

[kd7ost]

Hi Tom,

Didn't Maynards process require some sort of low level radioactive sensors on the wings to pick up the signal? I think I was looking for something like that and maybe they can be found in some of the smoke or Co2 sensors used in home safety?

I dunno. It was a mystery to me but sure would be cool to find a way to do it.

Dan

[Tom Harper]

Dan,

Maynard used ion sources from electrostatic brushes to couple his probes into the electrostatic field. I think you can do the same thing with fine points, like straw needles. Some time back C. L. Stong (Scientific American Amature Scientist Page) built a Tesla turbine that was powered by a wire held aloft with balloons. As I remember he used about 15 feet of Christmas tinsel to couple into the field. Needed a lot of points because he was drawing enough current to run a motor.

If Msrs. Franklin and Stong can do it without Ion sources, perhaps we can too. I want to start with the FMA sensor head and monitor some needle probes with the data logger.

Tom

[Tom Harper]

Programmed the servo control using the 3 wire interface to a serial in - parallel out shift register. It works very well. A single 5 wire flat cable for eight outputs. The outputs can either be PWM (for standard servos) or on/off for LED, camera or other device drive. The max output drive is 150ma per line. Of course the servos are signal level and don't require that much current, but it's nice to have when you need it. The board and program are set up for 5 servos and 3 LEDs at present. A four pin connector allows the LEDs to be remotely located.

These units are cascadable. If you need 8 more drivers, just stick on another servo control board.

Each servo is controlled by a 16 bit word in memory. The range is 510 to 926. The corresponding output pulses get within about 10% of the 1ms and 2ms limits. The PWM, and on/off, levels are output every ~20ms by the interrupt service routine. The 16 bit memory values are set during initialization and are changed by control routines. Any values that are not modified by the control routines become the fixed control positions that are imposed when the transmitter is switched off.

I really like the way this works. The first servo block I made from salvaged stuff. I'll keep it for bench test. Now I'm gluing together some new(er) servos to use in the model. Left some mounting lugs on this time so I have a way to mount them.

Next I need to pass through the conposite control stream from the transmitter. When the transmitter is on the transmitted pulses are passed through to the servos. When the transmitter is switched off, the memory values are used for control position.

[deh6]

I want to run some experiments on using the electrostatic field for leveling. Tom Harper,

(I just ran on to this thread.) I looked into the literature on atmospheric electrostatic fields several decades ago when I was flying fullscale sailplanes. Very interesting part of your project. I've thought about it a lot, but never have gotten around to building something.

Someplace (I just went through one pile and did not find it!) I have an article on using the electrostatic field for orienting a model plane. It may be an IEEE article. One thing I remember from the article is that the vertical field usually reverses under a cumulonimbus. The article said that when the plane passed under the cloud it rolled inverted!

Measuring the electro*static* field requires either something mechanical (the field-mill) or the ion scheme (dig a little of that radioactive stuff out of smoke alarm for it(?)). An ultra high impedance amp is OK for measuring a changing electrostatic field (contradiction of terms!) such as lightning-at-a-distance, but it won't do for a constant field. Basically, the measurement is charging a cap via a very high resistance, so the higher the resistance the longer the time, but eventually the cap is charged so there is no output.

The mechanical schemes (of which there are many) alternately shield/expose electrodes thus generating alternating voltage proportional to the static field, though in one scheme a cylinder with segments rotates and a brush picks up the charge so that the communtated output is dc. The method that appears to be most used today is a disc with "teeth" that is rotated by a motor; the teeth cover/expose small "buttons". The buttons go to a high impedance amp, and the disc is grounded. In a video (maybe PBS program lightining) it had a quick shot of one in operation near the Kennedy Space center where they monitor the field strength before a launch so as to avoid a lightning strike when the rocket takes off.

Another issue to be dealt with is the build up of charge on the airplane. Then there is the plastic film covering on the plane that holds a charge. I don't know just how those would play out. (I think that is addressed in an old out-of-print book "Atmospheric Electricity" Chalmers, where they were taking field measurements from a balloon at various altitudes and they had to neutralize the charge on the balloon...something else to dig out).

I think you can do the same thing with fine points, like straw needles A chap gave a local IEEE presentation at the local power company about lightning. As I remember he said that a one meter vertical wire with brushes (I think the ideal is fine wire such as stainless steel that comes to a sharp point) will generate about 100 microamp on a fair weather day, but this is respect to ground. In the plane something to drain off the charge build up would be needed (and wire brushes would sure have a lot of drag).

Anyway, a great idea. Keep us updated on this.

[Tom Harper]

deh6,

Thanks!

A PHD type up at Langmuire Lab told me about the 'shaded pole' scheme for measuring the absolute value of the field. I could never figure out were the ground reference was. I think it was measuring the incremental change in charge because it was mounted on a rising balloon.

In the stabilizing case you are measuring the gradient that exists when the wing tips are poking into the field at slightly different altitudes. Ground is the electrical center tap between the two. That should be a static voltage rather than a differential charging a capacitor.

The 100 microamps from a meter of brush points is interesting. That is the method used by CL Stong. A rough calculation indicates he was coupling into an impedance of 100K Ohms per meter. That's probably a lot of bristles. Another rough calculation says that I could match my 100 meg Ohm resistors with about a half inch of bristles on the wing tip. That's doable.

Another possible coupler would be the pin plate out of an Ionic Breeze device. One of the car models would be ideal. Anybody got one laying around?

Thanks again!

Tom

[deh6]

Tom,

I dug out Chalmers, Atmospheric Electricity, 2ed (1967). (It's old, but most of the work was done in early 1900's.) Using it for orientation looks rather bleak. Here are some notes of the more significant items that I jotted as I went through it--
- There are diurnal variations in the field. At one location it varied from +100V/m or so, to -100V/m (night) during the winter. In the summer that location did not go negative.
- The field can be negative in the lee of hi-tension wires from ions being carried by the wind.
- Pollution (mostly smoke) affects the strength of the field.
- Clouds, rain, drizzle, fog, sea air, snowfall, change the field.
- Dissimilar metals can add a volt or two (*)
- Ionizers used to equalize a conductor with the surrounding field:
..Effective resistance of a polonium equalizer approx 2E10 ohm.
..film forming over the polonium reduces the strength.
..half life of polonium is 136 days.
..voltage measured depends on the wind across the point (*)
..radioactive material should be one that only emits alpha particles; beta and gamma creates a space charge at a distance and changes the measurement.

* - If the installation is "differential" so that it is equal between wing tips the effect should cancel out.

Overall, I found it rather discouraging, especially the wide variation in the field strength and that it does not have to be a thunderhead for it to be negative. Clearly, angle of bank cannot be extracted, at least with a simple setup, since the strength of the field varies greatly.

Chalmers' chapter 5, The Vertical Potential Gradient, is worth looking at before spending too much time on the experiment. A library in a state university system most likely can get it via interlibrary loan. (Be sure it is 2nd edition).

I think it was measuring the incremental change in charge because it was mounted on a rising balloon. I found the section in Chalmers about balloon measurements (sec 5.37, p 157). The problem was charge on the balloon distorted the field, so they used a water droplet method to neutralize the charge. Since absolute field measurement is not the issue here, neutralizing the field would not be an issue. However, to the extent that charges accumulated unequally on the plane (the plastic film covering) that could have a big effect.

the 'shaded pole' scheme for measuring the absolute value of the field. I could never figure out were the ground reference was. I think it was measuring the incremental change in charge because it was mounted on a rising balloon. Chalmers mentions a double-field-mill used in airplanes, etc.

"Vonnegut, Moore, and Blume (1957) arranged two similar installations of radioactive equalizers at the wing tips. In level flight there is no differential effect, but on banking the aircraft the vertical potential gradient can be measured." The citation--
Vonnegut, B., Moore, C.B. and Blume, M. (1957). Preliminary investigation of the distribution of space charge in the lower atmosphere, Art. Stim. Rain , pp. 131-41. [5.38.,6.12.] Obviously, there has been more recent work done on this, but the principles remain the same.

About the issue of using points on the wing tips without radioactive ionizers--assuming you have internal combustion engine running, there is an ionic current generated. At least that is what I read. I tried measuring the current between my car and ground and got zero; the literature said it should be on the order of 10 ua. Assuming the literature is correct, the charge build up from the engine might be sufficient to cause ionization at the points on the wing tips, hence removing the need for radioactive stuff. Unless the plane is charged to a potential different form its surroundings, (Chalmers goes into detail of how an insulated conductor (the plane) placed in a field picks up ions from the air until its potential is the same as the electrostatic field), the points will not ionize the air.

Interesting stuff.

[Tom Harper]

deh,

Maynard Hill did not mention problems with charge build up on the model. The rapid motion of the model may prevent it. There is a charge difference between the LE and TE of the wing. The resulting current flow over the wing may disrupt any electrostatic field.

To detect a local static field you may need to use the integral approach. Connect a vertical wire to a mylar or polyethelene dialectric cap through a 10 meg resistor. Over a period of time the capacitor should take on a charge due to the atmospheric field. I have observed this as a start up error on a data logger.

A possible substitute for the radioactive source might be a voltage multiplier connected to a needle. Easy to build and not chemically hazardous.

I have a pdf of a very complete Chinese article on this topic. You do not need to read Chinese to understand it. The signal seems to be very large and somewhat noisy. He used a 408 Hz lo pass filter to limit the rate of change to the slew rate of the model. I tried to attach it but RC Groups rejects files larger than 3MB and this one is 3.6MB. Send me a PM with your email address and I will email it to you.

This is fascinating stuff that contains an element of magic.

[deh6]

Tom, thanks. I pm'ed you.

The txt file on the ionizer is along the lines of the unit in our central air conditioner. Rather than use 60Hz it had a small one transistor/ferrite core oscillator, then a a long string of diodes/caps to multiply. The high voltage fed an aluminum screen sandwiched between aluminum screen and insulated with fiberglass filter material. It worked for about a year and then the condensation ruined the pc board.

This is fascinating stuff that contains an element of magic. One can see how it captured people's minds around 1900, when they were relatively uneducated in electrical theory. (When they discovered that the dead frog leg that would "jump" when a voltage was applied it had some people thinking they had discovered the source of life!)

What makes electrostatics tricky is that we are used to dealing with voltage, current, and capacitance, which are derived from the basic phenomena of charge, but we don't usually deal with charge itself. Complicating the outside air situation is that ions in the air make it very slightly conductive (about the same conductivity as porcelain).

To detect a local static field you may need to use the integral approach. Connect a vertical wire to a mylar or polyethelene dialectric cap through a 10 meg resistor. I'm not sure I understand; the resistor has me puzzled. (Whether the wire is vertical or horizontal makes virtually no difference; what counts is the capacitance of the wire/plate.) If I connect a wire/plate to one side of a *discharged* cap of very high quality (low leakage, and absorption), and the other side to ground, then after a period of time, the cap will charge up to the potential of local electrostatic field. This charging time will be a function of the ion density in the air, and the air movement. The charging ceases when the potential of the wire/plate matches the local electrostatic field. After enough time has elapsed I then switch the cap to a high impedance voltmeter and read the initial voltage (which will decay as the cap discharges into the voltmeter impedance).

[Tom Harper]

The field is an orthogonal gradient of a nominal 180V per meter. If you run a vertical wire up 3 meters you enter a field on the order of 540V. A capacitor connected from there to a spike driven into the ground will evenutally take on some or all of the charge. The resistor provides a bit of safety if you touch the wire.

[deh6]

OK, I get the picture. (Regarding the resistor and safety--they say don't pee on a fence in the dry country.)

Thinking about this--then the voltage would be approximately 1/2 the 540 v since those parts higher than the 1/2 way point are at a voltage lower than the surrounding field and the lower half is higher than the surrounding field, so as a result when the transient has died out, the top half is gaining (losing) ions (gain/loss depends on the polarity of the ions) that the bottom half is losing (gaining).

The Chinese article looks well done, but being able to read the text would be useful. Clearly, the FET scheme depends on ions (as does the wire example above). The plane flying through the air would help the charging/discharging of the terminals.

The field is an orthogonal gradient of a nominal 180V per meter The key word is "nominal." Some of the data in Chalmers shows this varies a lot, so it doesn't look too useful for anything more than wings level on a fair weather day. To get any sort of angle or roll rate would require some sort of constant calibration.

I think I'll do a little experimenting. I don't hold much hope that it will be useful for UAV'ing, but it is an interesting subject.

[Tom Harper]

deh,

Good plan! I'm going to run some experiments - but, after I've finished a couple of projects.

I think it holds promise. I assume it was used on Maynard Hill's successful transatlantic flight.

Keep us informed.

Tom

[kd7ost]

Hi Tom,

I was able to correspond with Maynard Hill a few times after his transatlantic flight. I asked him how he managed roll control and he stated he used a gyro. It wasn't an off the shelf hobby gyro though. It was something they got from Japan for a $1,000.00 per copy or so. It was much more sensitive than a hobby gyro according to his reply. They placed it so that it would detect yaw, but used it to correct roll. They had to build the signal amplification and servo driver circuitry themselves to make it work.

It worked along the lines of the BTA-AS06/07 units but was much smaller.

Dan

[Tom Harper]

Ah ha, so even Maynard abandoned the electrostatic scheme for his big flight.

Interesting.

[deh6]

They placed it so that it would detect yaw, but used it to correct roll I presume it was based on the principle of the old primary instrument panel (turn & bank: a yaw rate gyro ("needle"), inclinometer ("ball"), and airspeed) handles roll/bank. Keep the ball centered (no slip/slide) with rudder; keep the needle centered for straight flight, or offset for a turn with aileron, and hold the airspeed constant with elevator.

It wasn't an off the shelf hobby gyro though. Maybe he was also integrating the rate for heading and needed low drift between updates.

...so even Maynard abandoned the electrostatic scheme for his big flight. As I remember the story a number of the attempts never made it (maybe they used electrostatics?).

I looked up Maynard Hill's patents--
http://patft.uspto.gov/netahtml/PTO/search-bool.html

3,868,074 Shows the circuit and layout in a plane.

3,873,050 Compensation for the charge accumulated on the aircraft.

4,067,520 Compensating for mountains, towers that will disturb the field out to about 5 times the height of the protrusion.

The last two suggest he was beginning to see the problems with the basic scheme.

[kd7ost]

As I understand it,

All the planes were set up the same. All first attempts simply quit working and telemetry was lost. I don't recall reading anywhere that they knew what happened. I never asked him about those either. Could have been battery failure, engine failure or what have you.

I don't understand how the roll control works either but it has something to do with a conventional high wing plane and the way that you can't have roll without inducing yaw and vise versa. But since the rate gyro will accumulate position error, somehow placing a highly sensitive gyro on yaw, and using it to correct roll works. I believe it may be due to the fact that if there is any heading error being induced by a roll error, yaw will continue to shift off axis. That way the gyro error doesn't null unless the plane is returned to a flat roll which makes the yaw stay on track.

My explanation is poor. I don't really know how it works but it seems simple. (Thats how my mind works :D )

I had tried doing the same thing with just a rate gyro once but couldn't get it to work.

I have a BTA-AS/06 and an 07. I can state from observation that when installed in a plane, and I taxi., As I steer the plane and make turns on the ground I see aileron deflection come into play. It's proportional to how hard I turn or induce error into the gyro. Even though the plane is flat in roll while sitting on the ground in motion, the unit will try to deflect aileron till the yaw error returns to neutral.

It also does level the plane in flight as stated. And quite well I might add.

Don't know if this helps or merely muddies the waters. I admit to being a system integrator and not an engineer. You guys are way over my head.

Dan

[Tom Harper]

I doubt that electrostatic sensing is a cure all for leveling problems. But, it may have something of value for AP systems. We only fly on clear days in dry New Mexico air and avoid thunderstorms. It's worth some investigation.

Tom

[kd7ost]

I agree,

I tried doing something with it once but got no where. I don't have the tools to make it work. It would be small, light, and very miserly in the power consumption department when compared to a gyro.

Dan

[deh6]

I don't really know how it works but it seems simple. I think you have the idea. If the flight is coordinated (high wing, polyhedral do this naturally; some planes require rudder to counter adverse yaw when starting a turn) then the angle of bank relates directly to the rate of heading change (and diameter of the circle) for a given airspeed. Rather than depend on the aerodynamics, one way to automate coordinated flight is with an accelerometer:
http://www.rcgroups.com/forums/showthread.php?t=354497&highlight=yaw
I've currently have a scheme using a vane working on the bench now (and a of household project that has to get finished first!).

We only fly on clear days in dry New Mexico air and avoid thunderstorms. Sounds good. I'm in Florida. A lot of humidity. Except in the winter, thunderstorms are almost daily and the high level blowoff from a thunderhead a long distance away is strong enough to affect the (IR) security sensors on the flood lamps outside; they begin "beeping". Once I was flying a sailplane (full size) in a sea breeze front. About 5 mi up the coast there was a thunderstorm. The lightning was coming out of the blowoff (anvil) at least 5 miles from the cloud inland and that means the field strength was around 30,000v/m, so it had to be pretty high another 5 miles further. A field-mill & logger has been on the project list for some time, more from curiosity of lightning strikes than model airplane flying.