[Rick Masters]

    Yeah, if you make it too sensitive, you wreck it. So I would recommend building in switchable sensitivity settings on the new prototypes if you are kind of shooting in the dark. Remember, Alan sent me several prototypes before we developed one that worked. Most of the problems were over-sensitivity which provided meaningless output. I tell people again and again that we settled on the "perfect" sensitivity. But that's where the work begins. You have to think of the Snooper and the vario as one instrument and learn how to use them together.
    I remember how frustrated I was with the early prototypes. The first one, I remember, was a glorified draft sensor. But Alan finally got it right.
I'd hate to see you guys waste a whole season just to get to the point we reached in the autumn of 1986 with a well-functioning unit.
    I've been trying to convey that the "magic" is in the pilot's interpretation of what's going on and what must be done. I've seen a lot of pilots fly with perfectly good Snoopers and declare them junk. They either couldn't figure it out or were flying in conditions the Snooper does not work well in (likely humidity, lapse rate, dew point-related. I only tested it in the desert.) Others said it worked amazingly well and they would not part with it. (My camp.)

...length of time of beeps and boops.

    The length of the sounds seemed fixed to me. The interval between sounds was variable. Just like the soaring barometric variometers I am familiar with. Frank might describe how the length of sounds in conventional vaiometers is done.
    When I flew with a Snooper, I focused on the excitement. Lots of BOOPS and BEEPS indicated the presence of a thermal. Lots of BOOPS don't necessarily mean you are descending. The Snooper does that entering the cool air cascading along the sides of a warm rising thermal, right before you enter the thermal. That happened a lot.
    Harry suggested the strength of the temperature differential between wingtip sensors might be useful. It should be easy to build a lot of (weird) functions into a new prototype and experiment with them in the air. That's what I did with the original Snooper. It's quite an entertaining mental challenge. Really sophisticated RC guys could use this thing, as well. That would be hard, but really cool.
    You can also derive the strength of temperature differential along the flight path with higher resolution using a time function as described previously. I don't think the orientation of the sample points is nearly as important as the distance between them. Since the surrounding air is moving much more slowly than the hang glider, a high-resolution differential derived over a few seconds along the flight path should be superior to a low resolution 10-meter fixed sample perpendicular to it. Just my hunch. It may be possible to detect small thermals on light days with wingtip Snoopers. That's what Harry is thinking. I'm jaded by having flown so much in powerful air. I counted good days by how many times I went weightless (slack strap). You don't want a sensitive temperature variometer on days like that. It would drive you crazy. Even when I describe the "mellow" days of autumn in the Owens Valley, there's still some kick-ass stuff out there.

[Harry]

More thoughts on Rick's comments and possible control circuits for a combined vario and TS

1) The primary purpose of the Snooper is to get excited, signaling to the pilot that he should initiate a search pattern. If the thermal is of adequate size, he will encounter it.

I'm trying to visualize this in my head. I would think that the excitement level would be the trigger for initiating the search. If the TS is beeping an audio tone, indicating heating/cooling temperatures in rapid succession, then it would seem that the absolute value of the pulses should be added to create a single sound or indicator to reduce confusion or distraction. Simplifying the alternating audio into a single tone or visual queue may make it more viable for a novice pilot. Level shifting the negative pulses into positive logic, and then say maybe adding or counting the pulses and comparing to some set value could be the gate for a "SEARCH" indicator. Maybe add a knob or switch so pilot can set the trigger level.

2) The secondary purpose of the Snooper is to verify the core of the thermal. When the hang glider centers the core, the Snooper will quiet down in the warm central rising air.

Variometer output > 0.0 volts (positive voltage), triggers the TS to be silent.

3) The third purpose of the Snooper is to tell the pilot when the thermal is topping out. The thermal will begin to slowly BOOP as the variometer indicates decreasing lift. Eventually the Snooper will fall silent as cooler upper air dominates. On racing or XC flights, the decreasing rising speed of the thermal can indicate it is time to go rather than wasting time thinking you have drifted outside the core and searching for it.

Zero alternating pulses. Variometer output <= 0.0 volts, acts as logic gate to turn on Cooling or "TOPPED OUT" indicator.

4) The fourth purpose of the Snooper is to indicate the presence of eddies. These atmospheric ripples do not have a changing temperature component so the Snooper will remain silent. Shears and the path of shears can also be recognized by the astute pilot.

No output from the TS. What is the vario reporting in this case? If nothing is triggering the TS, then no output.

I have a Ball 500H vario that would be perfect for matching up with a TS. I can add a "SEARCH" indicator above the needle gauge, and a "TOPPED OUT" indicator below the needle gauge. Radio Shack sells these little audio kits that can be programmed with a human voice to announce "SEARCH" or "TOPPED OUT", thus reducing all the beeps and boops competing for the pilots attention.

[Bill Cummings]

"No output from the TS. What is the vario reporting in this case? If nothing is triggering the TS, then no output."

This can mean you are perfectly coring the thermal.
___________________________________

"Zero alternating pulses. Variometer output <= 0.0 volts, acts as logic gate to turn on Cooling or "TOPPED OUT" indicator."

I think this can happen with just more than topping out. (Rick )
___________________________________

" programmed with a human voice to announce "SEARCH" or "TOPPED OUT", thus reducing all the beeps and boops competing for the pilots attention."

Oh boy - NOW YER TALKIN!!

[Rick Masters]

"Zero alternating pulses. Variometer output <= 0.0 volts, acts as logic gate to turn on Cooling or "TOPPED OUT" indicator."

Maybe you'll see me as a sourpuss, but too much embellishment of data can lead to no improvement - or worse.
Remember, the most powerful computer in the Snooper/vario mix is the pilot's brain.
He doesn't need to be told anything by a talking box.

Yes, you will get zero alternating pulses from the Snooper when there is little in the way of temperature difference in the air you are flying through.
That can be in areas of low dynamics like still air, wind shadows, inversions and plain vanilla days with little lapse rate. And it can also happen in specific parts of high dynamic environments like centering thermal cores, topping out thermals and travelling along the crests of horizontal wind shears.


For instance, the Owens Valley is like a pipe that connects the high Mojave desert to the northern basin and range. At night, the cool northern katabatic flow runs south like a river towards the San Gabriels. Day brings warming and the flow reverses, with warm southern air pushing north against the cool southern flow. In the afternoon, a gentle shear often sets up south of Bishop as the warmer air rises over the cooler air, creating an invisible highway from the Sierra across the valley to the White Mountains.


This air is not rising from thermal activity. It can be constant temperature at zero sink as you cross the valley on the crest of the shear. But if you drift to either side, both the vario and the Snooper will begin making sounds. In the image above, imagine you are sitting on Black Mountain at the southern end of the White Range. You are looking West toward the Sierra along the crest of the shear at 6000 feet in the center of the image. To either side, the vario will indicate sinking air and the altimeter will concur. The Snooper will BOOP or BEEP occasionally if you slip off the southern side (left), still in warm air, but if you drift off the northern side (right) and into the cooler air, the Snooper will BOOP excitedly as it admonishes you for making such an uncorrectable error - all the way to the ground. [use ctl- to view the entire image]

[Harry]

"Zero alternating pulses. Variometer output <= 0.0 volts, acts as logic gate to turn on Cooling or "TOPPED OUT" indicator."

Maybe you'll see me as a sourpuss, but too much embellishment of data can lead to no improvement - or worse.
Remember, the most powerful computer in the Snooper/vario mix is the pilot's brain.
He doesn't need to be told anything by a talking box.

Yes, you are clearly a sourpuss.

The greatest engineers in the world are in this forum and you are scoffing at them. We are here to analyze you and reverse engineer your brain on how to make the TS work.

I finally figured out today what is really happening. AS WE ALREADY KNOW, the TS-I essentially sniffs out the various temperature bubbles you call the "sloughing thermal." The instrument simply beeps and boops the changes in temperature as you carve through the sky like the blind man you are, looking for invisible lift. The beeping noises are being processed in your head trying to make sense of the information. What your brain is doing, is processing the information the same way a sample-and-hold circuit works, in other words, an analog peak-detector. It dawned on me today what is really missing from the original circuit to make it all work for the mere mortal hang glider pilot. No one has changed anything for the worse. All I'm trying to do is make improvements. Having figured out what to do next, I can now move forward. The circuit I propose to add to the TS is a peak-detector. Dan will most likely agree with me that the peak-detector can be added without harming the original design. By adding the peak-detector, you won't have to guess when to start your search. You won't have to count so many seconds to decide when to turn and search. The detector will do that for you IF YOU LIKE. It is a feature that can be turned on or off. With my purely analog design, it will be real-time and damn-fast.

Refer to the Thermal Snooper Basics doc, figure 8, at the very top of the diagram you will see a bell curve, labeled "Temp". That little curve says it all. The closer to the core, the steeper it gets, further from the core, the flatter it gets. A peak-detecter will look at that curve (in terms of voltage) and measure the rise and fall of the temperature as it passes a thermal. When it finds the peak, THAT is the trigger to start your SEARCH. Your brain does the same thing listening to the beeps and boops. I've been sitting here all this time trying to figure out how to normalize all that noise you learned to do from experience. You shouldn't have to do that. The circuit can do it for you and MUCH MORE EFFICIENTLY.

Thus, I will add a peak-detector to my circuit and will not have to guess when to start my search. It can trigger my SEARCH indicator on my vario and turn on the audio announcement to begin looking for that thermal. Just to make things more interesting, I can also add audio messages that can be used at launch such as: "Thermal detected", "topped out", "thermal passed", for when sitting at launch kicking rocks and waiting for a good cycle, or the best time to launch.

[Rick Masters]

The circuit I propose to add to the TS is a peak-detector.

Don't forget the trough detector...

[Rick Masters]

We are here to analyze you and reverse engineer your brain on how to make the TS work.
-- one of the best engineers on the planet

Be careful...

This guy may very well be demented.
-- Paul Murdoch, President, United States Hang Gliding and Paragliding Association

GIGO.

[danstrider]

Some progress to report. I've consulted with two colleagues and have two separate paths going.

First is a rough rebuild of the snooper's sensing elements (Alan's Figure 3). The same temperature to voltage relationship of a diode is being used with an opamp to provide amplification gain (instead of a series of inverters). Using the calculations and gain numbers from Alan's document resulted in a snooper with significantly different time response. The diode's temperature temporal response doesn't make the same high-pass filter as Alan presents, so blowing on the sensor provides a much faster response than an actual snooper. Also, the R-C filter that Alan presents isn't making sense, since the diode isn't really a resistor, and the order is backwards. All of this is getting deeper than my electronics depth, so that's all I know to regurgitate.

Second is a bit of a break away from the original snooper, but is working on the same principles. A very small thermister (~20ms time constant) is being over-sampled at >100Hz and differentiated with the output rate at 10Hz. At my aircraft's cruise speed of 12m/s (24kt), this results in about 3m (10ft) of physical lag. Software is then providing additional filtering so we are able to play with the band-pass filter time constants to get roughly the same temporal response as the original snooper. Right now they don't sound exactly the same, again with the new implementation being much faster than the original snooper, but it is easy to slow down the time response now to better mimic the snooper's response. The physical circuit is being converted from breadboard form to a small PCB that is flyable. Also, it is being set up to have multiple input channels to get readings from both wingtips simultaneously, per earlier discussions.

The equipment to measure the netto variometer and either of the two temperature rate methods is in place. The aircraft state (position, altitude, airspeed, attitude, etc) is also measured and the winds (direction, magnitude) at altitude are estimated. Everything is time correlated and recorded. I'm quite optimistic the snooper problem will now boil down to flying the two snooper sensors in good thermal conditions to get a nice data set, then spending time analyzing the data set to figure out what the correlations between the snooper and variometer might be (i.e. how to translate Rick's understanding into math), and finally creating a candidate heuristic to steer the aircraft based on the snooper measurements to validate if thermals can be located with the snooper.

Fun so far, more to come.
Dan

[danstrider]

I made a few good steps toward a new snooper and wanted to get a little feedback from folks.

The new snooper sensor itself is 1x0.5 inches. It uses a diode for the sensing element, but the whole rest of the electronics path has been updated to the 21st century ... real instrumentation amplifier, high-pass filter, and differentiation circuit. This new setup is also more straight-forward to calculate the analog signal processing characteristics, so it can be better documented. I'm not ready to share the actual numbers or circuit, but those will be coming eventually.

Interestingly, running the sensor on an o-scope, you can put your hand a few inches away from the sensor and it gives a signal. Similarly, from a couple feet away, you can blow a puff of air and watch as the warm breath eventually gets to the sensor. Pretty awesome actually.

I made a quick system with a microcontroller and piezo speaker to get a feeling of the processing to make the "beeps and boops" sounds. It uses the same philosophy of sounding after the temperature rate passes a threshold (digitally programmable). The length of the pause between beeps/boops is again the operative signal. Beeps indicate rising temperature, and beeps closer together in time indicate a faster rise in temperature. Boops are decreasing temperature. If the temperature isn't changing, the sensor is quiet.

Here is a (very boring) video testing the system with a glass of hot and cold water.
https://youtu.be/GHbBhHSSsBc

Listening to the original snooper prior to takeoff, it seems more sensitive. It can't be sensing more temperature change compared to my glass of cold water, so it must be tuned for more sensitivity...
https://www.youtube.com/watch?v=uCsGjg3bFdU

I'm looking for feedback on what the snooper should signal in practice prior to takeoff. What can I do to dial in the sensitivity of the sound on the ground? Everything is tunable except the actual raw sensor sensitivity.

Enjoy and thanks!
Dan

Code: Select all

/*
  Beeper/Booper for the thermal snooper.
  Dan Edwards, 2016-09-04

  circuit:
  * 8-ohm speaker on digital pin 9
  * themal snooper on analog pin A0
*/

#define speakerPin 9
#define snooperPin A0

#define beep 1200   // Hz
#define boop 400    // Hz
#define durate 100  // miliseconds
#define thresh 8    // counts out of 1023 total range of snooper

void setup() {
  // sensor warm-up time
  delay(250);

  // initialize serial communications for debugging
  Serial.begin(9600);
}

void loop() {
  static int lastReading = 0;

  // read the sensor
  int snooperReading = analogRead(snooperPin);

  // print the sensor reading
  Serial.print(snooperReading);
  Serial.print("  ");
  Serial.println(lastReading);

  // play tone: beep is rising temperature, boop is falling temperature
  if (snooperReading >= lastReading + thresh) {
    tone(speakerPin, boop, durate); // beep
    Serial.println("BOOP");
    delay(durate); // delay while tone goes
    lastReading = snooperReading; // save

  } else if (snooperReading <= lastReading - thresh) {
    tone(speakerPin, beep, durate); // boop
    Serial.println("beep");
    delay(durate); // delay while tone goes
    lastReading = snooperReading; // save
  }

  delay(25);
}


[Bob Kuczewski]

Very cool Dan!!!!!!

For testing, I would take it to my favorite launch site and see how it responds to thermals coming through. Maybe you could rig up a very light streamer on a long pole as a visual cue. Then watch for correlations. It would be interesting to see the relationship (lead time, etc.) between the snooper and vertical airflow.

Thanks for sharing the code. It looks like C. What processor are you using?