.

.

All things hang gliding. This is the main forum. New users, introduce yourself.

Moderators: sg, mods

User avatar
By mgforbes
#47455
Got the FLIR version of this at work. Way cool; sees all sorts of interesting
things in the thermal spectrum. I'm using it for circuit board design optimization,
but they're handy for other things too. For $30 grand, you can have one too. They
do come cheaper....black/white versions for a few thousand.

I've been in a discussion with somebody local about resurrecting the thermal
snooper using modern sensors and processors. Sounds like there's some
interest in such a device.

MGF
User avatar
By saltoricco
#47463
mgforbes wrote:Sounds like there's some interest in such a device.
Yes Mark, I'd love to have one. I don't have the expertise to help though. Can you tell more?

Holger
User avatar
By mgforbes
#47472
I envisioned the current-generation design using three sensors mounted on
the wing, and a centrally-located CPU to do data collection and reduction,
as well as whatever display/beep/light functions might be needed.

Looking around, I found that Microchip makes a thermal sensor with on-board
digitizer and serial bus output. It's very small and lightweight, and with 12 bit
accuracy at 4 samples/second, it has the resolution to detect very small
changes in temperature. The small thermal mass is essential; for good
response you can't afford to have a big sensor out there with a long thermal
time constant. You need something that responds very quickly to changes in
ambient temperature.

http://ww1.microchip.com/downloads/en/D ... 21909b.pdf

You'd need a CPU to do the serial polling of each sensor, and run the
algorithm needed for figuring out where the hot side is, and sensing the
rate of change of temperature. I don't have a good feel for what that
algorithm looks like, but I'd start off by just building it as a datalogger, and
go out flying with it for a while to see what sort of data was found. You might
build the proto with a microphone on it, so you could listen to a vario
beeping and grab that data at the same time the thermal data was being
collected...or modify the vario with an audio output tap and wire that to
the logger. Or even better, integrate a solid state pressure sensor and
read the pressure directly; you don't care about absolute altitude or rate
of climb. All you need is change in pressure to decide if you're climbing
or descending, and how that correlates with temperature variations.

http://tinyurl.com/36qk9b

A PIC18F2580 would do nicely as the CPU to run all this:
http://tinyurl.com/24skg5

A bit of power supply work, a display, and after that it's a Simple Matter
Of Programming. (SMOP) :-) And you'd probably want to throw in a
serial EEPROM to stash the data, along with a port to plug into your
PC so you could do data extraction and analysis. Spit it out in CSV
format so you can dump it into Excel. A 25AA1024 would get you a
megabit of storage for a couple bucks. That's 128Kbytes, figure four
bytes/record so that's 32K records, that's 136 hours at four samples
per second. Overkill, fer shure.....

You'd probably want some sort of beeper to produce two tones, one
indicating "hotter on the left and rising", the other indicating the right
side. Specific conditions to trigger are pending further research.

MGF
#47483
mgforbes wrote:I envisioned the current-generation design using three sensors mounted on
the wing, and a centrally-located CPU to do data collection and reduction,
as well as whatever display/beep/light functions might be needed.[...]
Thank you Mark! That's impressive.
mgforbes wrote:You'd need a CPU to do the serial polling of each sensor, and run the
algorithm needed for figuring out where the hot side is, and sensing the
rate of change of temperature.
So you're trying to capture the temperature gradient field? That should also require at least indicated airspeed as input parameter, I'd think. Otherwise you don't know whether an increased delta is due to an increased gradient or i.e. a deceleration. Or maybe that factor is minor enough.

I'm trying to wrap my mind around this one. If we don't shoot for the gradient field (but instead the change over time at a virtual center position) is there a gain in reading the temperature at different locations? When traveling through a gradient field it takes time X to get a readable delta. How does this time relate to the time that passes when first the front, then the rear sensor reads the same point of the gradient field? Would the second time be greater? Seems only then there is a gain. Gotta think about this some more.

BTW, have you read Rick Masters' article? He says it worked nicely in dry climate (Owens) but failed near the coast. Of course that was back then, with a single sensor and no Mark. :wink:

mgforbes wrote:I don't have a good feel for what that algorithm looks like
We could sit down together with Dan once the data is available.

mgforbes wrote:You might build the proto with a microphone on it, so you could listen to a vario
beeping and grab that data at the same time the thermal data was being
collected...or modify the vario with an audio output tap and wire that to
the logger. Or even better, integrate a solid state pressure sensor and
read the pressure directly; you don't care about absolute altitude or rate
of climb. All you need is change in pressure to decide if you're climbing
or descending, and how that correlates with temperature variations.
None of the varios offers a real-time readout, hm? The total energy-compensated output of a vario would be better than reading the pressure. Maybe one of the vario producers can give you a unit that has a read-out. After all, that's pretty interesting field research you're doing.

Please keep us updated, that's very exciting!

Holger
User avatar
By mgforbes
#47484
I'm thinking of three sensors; one on each wingtip, and one at the CPU
position on the basetube. What you're looking for is not only changes in
temperature, which can be simply rate-of-change-per-unit-time, but also
changes from one side of the glider to the other which may give you
an indication of which way to turn. The center reference sensor may be
unnecessary....but at a couple bucks per, if you've already wired up two
of 'em on a serial bus, adding another one (or two, or five) is easy/cheap.

Zero'th order project is to build something that works as a datalogger. I'm
fairly confident that once we have some sampled data that includes the
current altitude and temperature for each sampling interval, we'll begin to
see some patterns that we could use for prediction. At very least we'd get an
idea of how much scatter there is in the data, which will guide the tuning of
damping algorithms.

This is all predicated on having (1) some free time to work on it, and (2)
somebody willing to write all these fancy algorithms. I'm a hardware guy,
not a programmer.

MGF
User avatar
By saltoricco
#47485
If we're super lucky we may be able to get an indication by just reading the temperature difference between the two wing tips. Without much calculations. When you're scratching close to the terrain you're changing flight direction frequently. That may give a greater integrated delta in unstable then stable conditions.

Holger
User avatar
By saltoricco
#47486
mgforbes wrote:What you're looking for is not only changes in
temperature, which can be simply rate-of-change-per-unit-time, but also
changes from one side of the glider to the other which may give you
an indication of which way to turn.
Ah, so you aren't trying to give the sensor more time to react but do try to what I called "capture the gradient field."
mgforbes wrote:This is all predicated on having (1) some free time to work on it, and (2)
somebody willing to write all these fancy algorithms. I'm a hardware guy,
not a programmer.
OK, once the data is in I bet Dan is going to want to see it and I do too.

Holger
User avatar
By gerg
#47490
Hmm, interesting...

I've got a few PIC's lying around that I'm not doing anything with, and can do some simple programming... Not much would be needed for something like this. I've done mostly PICBasic (makes it easy...) but little bits of ASM where needed...

I would be curious though... whether 4 readings per second would be enough resolution to be useful. Don't most Varios work at more than double that?
User avatar
By saltoricco
#47493
gerg wrote:Hmm, interesting...
Great, maybe you can help Mark build data loggers to collect data? 4 reading per seconds...let's see. If you fly 40 MPH that's 18 m/s. Divided by 4 is 4.5m. So you get a reading every 4.5m. I'd think a little faster would be helpful. OTOH, you're probably slower when you're snooping around, so it's probably good enough.

Holger
User avatar
By peterb01
#47767
Just stumbled on this discussion. You can thank JeffO for intoning my name and experience. Yup, I've got prototype #13 and it works like a charm, helping map out the patterns and some movements of the air I never knew existed prior to the Snooper. I've got a number of stories / experiences which show the advantage of the unit. The biggest problem was a steep learning curve because it's not very intuitive; it'll signal one thing when you think something else "should" be happening.

For example, it really, really helps when following a thermal "train", upwind from the caboose to the engine (so to speak). Its variable beeping rate tells me it's not time to turn just yet but then I hit the main core, continue into the wind until it just signals I'm entering cooler air... then it's time to crank and bank. Once did the above, finding 50fpm bubbles and then hit the engine at 1000fpm. The particular run took 20 minutes from hitting the first bubble until I hit that cooling point, where the main body of the wind compresses against the thermal column.

Noisy and irritating to everyone else while on the ground but mostly silent in the air, unless there's a lot of hard lift when it sings, "Turn NOW!"

They're real, guys.

- Peter
User avatar
By UnTuckable
#47905
It tells you warmer vs colder, but harder to learn to use than a vario.
Attachments
TS2.jpg
TS2.jpg (153.38 KiB) Viewed 4550 times
TS1.jpg
TS1.jpg (145.37 KiB) Viewed 4550 times
User avatar
By gerg
#47913
Now that's a nice little piece of history!
User avatar
By JR
#47926
I had two of them and ultimately decided they were either completely useless or I just wasn't smart enough to interpret what they were telling me. After they fell off my DT in flight, it became a moot point.
User avatar
By UnTuckable
#47928
gerg wrote:Now that's a nice little piece of history!
It's seen a lot of thermals, still works (I think) just don't have a 9v to test it,

the power off/on was unplugging the battery

I added a safety line too = no lose.
User avatar
By gerg
#47937
Hmm.. More I think about this, the more the geek side of me is interested.

It'd probably be a good start to just get a data collecting rig set up to get some idea of what the numbers would look like, and what sort of number-crunching you'd need to do (if any) in software. Only trick would be tying this to actual indications of lift, e.g. You have a whole bunch of temperature information, but absolutely zero in terms of whether it actually had an effect on your glidepath.

Now, my GPS outputs coordinates once-per-second over a simple serial connection, along with a timestamp. So maybe I could log that timestamp with the temp data, and I can then easily run through my flight track and see what the temp difference actually looks like when entering/leaving a thermal.


Then there's the non-geek, free-flying side of me that says.. Well, I've got great temp sensors all over my body in the form of my skin, not to mention that if I actually hit lift, I can feel it (e.g. one wing will lift...), so why bother with this stuff? I guess the one real advantage would be that I can't feel the temp on my wingtips, so that might be useful info that I normally wouldn't have...
User avatar
By mgforbes
#47952
See my design outline, above. If you're already datalogging anyway, adding an
absolute pressure sensor is fairly easy. That gets you pressure changes,
which correlate well with changes in altitude. You don't need to temperature
compensate, or even convert to altitude, because all you really care about is
the differences and how they correlate with temperature changes.

MGF
User avatar
By UnTuckable
#47962
I've made suggestions similar to what MGF said,
2 sensors one on each wing tip, or 4 an include the nose and keel end,
wireless would be nice, you could wire them into a wheatstone bridge,
and just display which side was warmer.

I used mine more for staying with a thermal, indicating where the thermal
was moving towards (lots of warm beeps) an when I was getting close to the edge (lots of colder boops).
User avatar
By mgforbes
#47984
UnTuckable wrote:I've made suggestions similar to what MGF said,
2 sensors one on each wing tip, or 4 an include the nose and keel end,
wireless would be nice, you could wire them into a wheatstone bridge,
and just display which side was warmer.
The analog method (bridge) would work, but it has lots of analog-ish
calibration and drift problems. The current-generation sensors integrate
all the analog stuff on-chip, along with the A-to-D converter and serializer
for easy communication with a microcontroller. All the fiddly complicated
stuff is already done inside the sensor, for only a couple of bucks. From
there, it's all digital comms and processing.

The sensors are very small, and can be a plug-in module that fits onto
a connector, mating with a permanently-mounted cable on the glider.
Four wire shielded twisted pair would work nicely; V+/Data/Clock/Ground.
Use Belden 8723 or equivalent and it'll hold up for years if done right.

Looking into this a bit further, I see that a +/- 6G accelerometer module
is only $27 in singles at DigiKey. Put that on the board along with the CPU
and the pressure sensor, and now you can tell what the glider's doing in
real time, along with the thermal and altitude readings. It also comes in
3G and 2G ranges, but I figured 6G would do the trick. PG acro guys might
want one just for datalogging....you pull some serious acceleration even in
an asymmetric spiral. I've always wondered just exactly how much. Other
maneuvers are much more energetic.

So far, I see a total BOM cost for the main hub of about $60, of which half
is the accelerometer. The sensor modules at the wingtips are $5 or so
each. That implies a total cost to build of about $90 by the time you figure
in cable and connectors. As a commercial product, it would need to sell
in the mid-300's to be viable, though dropping the accelerometer feature
could kick that down a ways. (You need to figure in packaging, shipping
and dealer markups.)

MGF
User avatar
By Lobido
#48013
:welcome: Untuckable. I recognize the name. Tell us about yourself!
User avatar
By UnTuckable
#48015
I'm just a programmer (hardware if it's PNP ;)

Temperature is one variable, humidity in humid regions and static charge for very dry areas, could be used for additional thermal detection.

Sensors / sensible ?

I started out being very conscientious about keepi[…]

JD, have you flown a newer U2 lately? Just wonde[…]

Windy season

I've used my fpv RC craft for getting more familia[…]

A little bird told me.

....In addition to video, better quality audio re[…]