Wind sensor schematic

[dave_in_NJ]

I have been trying to understand the schematic so that I can understand the output as it relates to wind speed.
I plan on using an Arduino with this as an input.

it appears that the actual sensors are connected to individual op-amp inputs. and one sensor is temperature only and outputs to the micro controller. the other is the wind sensor. This appears to be part of a Wheatstone bridge. half is the sensor in the wind, the other is part of the op-amp output/voltage divider after Q1

in most of my reading, the two sensors in the wind are on the same Wheatstone bridge. am I reading the schematic correctly ? any help with my confusion would be most welcome !

[oz]

You are correct that in the most standard schematic both sensors are part of the wheatstone bridge. The NTC thermistor though, has enough issues with non-linearity that we found locating a thermistor that even began to approximate the response in the bridge while being 50C cooler was a tough order to fill. The other issue we had is that it is easy to push the thermistor in a zone where it draws a lot more current, which is also not desirable.

So we just decided to go the hard way with software compensation. Although getting the data through a temp-controlled wind tunnel is lots of work.

Hope that helps,

Paul

[dave_in_NJ]

50c ? wow, that is quite hot ! I was thinking 50°F, max. I guess that more power means faster cooling, more time to see the changes.

my need is to get a sensor that is accurate to wind speed from about 2 to 35 mph. I consider slight breezes to not really be wind, so the low-low end is not as important. over 35 mph and patio furniture starts to turn over.

I noticed the filtering on the last stage. is the raw signal a pulse ? it seems the output could fluctuate under any wind as wind has a turbulent flow.

I can visualize the output trying to compare the heated element with the non-heated on. the thermistors are on separate curves and therefore would not be linear. some correction in software has to be made.

what ADC chips are suggested for higher sensitivity ? looking at some, a 16 bit, offers about 13 bit resolution.

[oz]

my need is to get a sensor that is accurate to wind speed from about 2 to 35 mph. I consider slight breezes to not really be wind, so the low-low end is not as important. over 35 mph and patio furniture starts to turn over.

I think the sensor as it is now with the RV ouput for signal and the TMP hooked up as per my wind sensor sketch, should do fine for what you want. Unless you leave your patio furntiture out in the winter - below about 40 degrees is mostly unknown and not likely to be very accurate. See my sketch on the sensor page and in Github https://github.com/moderndevice/Wind_Sensor

I noticed the filtering on the last stage. is the raw signal a pulse ? it seems the output could fluctuate under any wind as wind has a turbulent flow.

No just a voltage - although there are schematics that generate pulses - then the freq is the output.

I can visualize the output trying to compare the heated element with the non-heated on. the thermistors are on separate curves and therefore would not be linear. some correction in software has to be made.

Correct - the biggest rap on thermistors is that they are fairly non-linear. Also the equations are non linear. If you think about it, the wind speed is going to be proportional to the power you're putting into the thermal element to keep it hot. P = E^2 / R So it's a non-linear relationship to begin with. Most of the accounts I've read suggest just calibrating from known wind speeds and doing regressions to determine your sensor's formula.

what ADC chips are suggested for higher sensitivity ? looking at some, a 16 bit, offers about 13 bit resolution.

Anything is going to work. IMO more accuracy isn't going to buy you much because of high freq components (you're measuring air currents after all) and the issue of calibrating the thing. Unless of course you have access to a nice NIST variable temp wind tunnel. Quite a nice place for a nerdy date in my opinion - only don't ask my spouse. She shakes her head when I start taking about wind sensors.

[Dalhimar]

I am currently working on a project to determine the flow rate of custom designed mini turbines. I am planning on using the wind sensor, and have tested it out without a handheld anemometer for reference. So far it appears to be decent at sensing the lower speeds, but I will be generating much higher velocities than normal ambient conditions. In order to do this there are a few things that I'd like to ask about the wind sensor so I might be able to better understand the way that it, and your sketch work so that I might be able to modify or calibrate the sensor/sketch to specific conditions where my testing will be.
In your sketch you have multiple static values used when determining final values such as in this line:

    TempCtimes100 = (0.005 *((float)TMP_Therm_ADunits * (float)TMP_Therm_ADunits)) - (16.862 * (float)TMP_Therm_ADunits) + 9075.4;

If it would not be too much trouble, might you be able to give a definition/explanation of what these (and others present throughout the sketch) static values are and how they were derived?
I know there is on sensor temperature sensing, however this seems to be somewhat effected by the windchill effect with breezes dropping the temperature readout noticeably. With this in mind and since the equations in the sketch use the temperature readout, might it be more prudent on my part for accuracy reasons to use a external temperature sensor that would not be affected by windchill effects inside of the testing platform I am making?
Also you have stated that the angle of the sensor effects the readouts, as expected due to surface area exposure to direct airflow, but what orientation have you seen the greatest accuracy from, or what orientation did your write the code to be used with? Acting as a blade, or flat of the sensor facing the oncoming air?

Worst comes to worst I'll be able to calibrate based off of a hand held anemometer and miniature wind tunnel such as the blog post on calibration used.

[oz]

Dalhimar,

Sorry for the slow reply I'll get around to getting my forum to notify me. There are few features on the forum that aren't built in and one is posts pending moderation.

The line

TempCtimes100 = (0.005 *((float)TMP_Therm_ADunits * (float)TMP_Therm_ADunits)) - (16.862 * (float)TMP_Therm_ADunits) + 9075.4;

Was derived from regressions with data from the temp sensing thermistor.
The regression was from a range of data about 30 degrees C in range. I haven't really checked to see how accurate it is with
the current build of sensors, we really haven't changed anything in the temp schematic.

Hope that helps

Paul