high resolution rain sensor idea

[zac]

I'm looking at the feasibility of developing a moderate cost (<$50 manufacturing cost in 10K quantity) high resolution rain sensor.  The typical sensors used in weather stations are usually bucket tipping types that offer limited resolution.  The collection vessel can be enlarged and/or the tipping bucket be made smaller, but the resulting resolution is still limited within practical size constraints.  A typical tipping bucket rain gauge will only read to 0.01"/hr (or to 0.005"/hr for expensive models) and will not even register small amounts of rain. 

I wonder if it would be practical to use a linear optical sensor array  (like those used in inexpensive scanners) to accurately measure rainfall.  Basically, have a coherent light source (laser diode?) project a flat several inch wide beam to illuminate the linear sensor array and then process the data from the sensor array to measure the "profile" of the falling rain drops.  Then, calculate the amount of rain from that data.  It should even be possible to determine the size and distribution of the rain drops. 

What do you think? 

PS.  Some car companies are using an opto sensor to control automatic wipers:

http://en.wikipedia.org/wiki/Rain_sensor

http://www.egr.msu.edu/classes/ece480/capstone/spring10/group06/Documents/ECE480_Design_Team6_Final_Report.pdf

http://www.fujitsu-ten.com/business/technicaljournal/pdf/2-7E.pdf

[Gallymimu]

you might be able to make this work when you have sparse rain droplets, but as soon as you run into scenarios where multiple droplets are in the same beam path I'm not sure how you'd ever be able to back out the droplet size.

[intel]

This job can be done cheaper by using piezo speakers. Arrange a sensor 1 m2. Then you counted the raindrops can be calculated. Refine the circuit will take some time, but you can have cheap accurate rain gauge. Have you need to go into detail, I don't know. But, I hope this idea interest you.

[zac]

This job can be done cheaper by using piezo speakers. Arrange a sensor 1 m2. Then you counted the raindrops can be calculated. Refine the circuit will take some time, but you can have cheap accurate rain gauge. Have you need to go into detail, I don't know. But, I hope this idea interest you.

That's an interesting idea.  If I understand correctly, you're suggesting a piezo speaker with a large surface area and each drop of rain hitting it will generate a spike.  Compensating for the surface getting wet and determining the size of the drops could be challenging. 

There is a problem with the optical scan method during heavy rain too as Gallymimu mentioned.  This could also be an issue for the piezo sensor if multiple drops hit at the same time. 

The signal processing algorithm would be something that counts the number of peaks (corresponding to each drop of rain) and uses the amplitude of each spike to measure the size of each drop?

[intel]

Dear zac, can be a good designer. I thought these problems. No need to write much. Wide-area issue: 125cm2 (25 cm x 25cm) of space to create, the results solved by multiplying by 4. Of course, the lower the sensitivity. This is due to the required sensitivity.

The other problem you mentioned, I never used the laser array does not occur in the system for less than $ 50, I am suggesting the system may be cheaper. For sensitivity, each sensor separately counts. I mean, every sensor before the digital "1" to produce, is designed. These are then counted. Number of integrated do it, depends on the design. But in general cheaper ICs ...

Drop size: we need to think it over. My first the solution. To do this, use a separate sensor, according to changes in the peak intensity of the final calculation will change.

Also, the problem of accumulation of rain water on the sensors, there is. This incline forming around or sensors to be grilled, designed and can be fixed.

[solutions]

a) evaporation will ruin your 12 sig figs of rain measurement "accuracy"
b) rainfall varies over distances even as short as tens of meters, so more accuracy than what's out there now is overkill.



If it's exercising your neurons, go for it, but you need to realize that real men build radar satellites to measure rainfall rates:



 

[intel]

This system can be a bit pricey at $ 50.

[zac]

a) evaporation will ruin your 12 sig figs of rain measurement "accuracy"
b) rainfall varies over distances even as short as tens of meters, so more accuracy than what's out there now is overkill.

I disagree.  The commercially available sensors are only good to 0.01"/hr (or to 0.005"/hr for very expensive units).  They don't even pick up a small amount of rain at all, even enough to make the ground wet.  Being able to at least detect a trace of rain would be useful.  I'm also considering a conventional tipping sensor for the larger amounts and a separate sensor to pick up small amounts. 

[solutions]

You're blind to the fact that IT DOESN'T MATTER.  Rain varies in its distribution.

Get out of the basement and look up. Those puffy white things are where rain comes from. Now imagine rain falling straight down from the puffy white things and none from the spaces between.

What's the average rainfall in the geographical area as a result of Billy having a rain-o-meter accurate to 1 micron and sampling in one location with a sample window of say 10x10cm?

Measuring rain to 0.01" is a marketing gimmick. It makes ZERO science sense.

[Gallymimu]

This job can be done cheaper by using piezo speakers. Arrange a sensor 1 m2. Then you counted the raindrops can be calculated. Refine the circuit will take some time, but you can have cheap accurate rain gauge. Have you need to go into detail, I don't know. But, I hope this idea interest you.

I think this may be flawed as you don't know the velocity of the raindrops which will change the force delivered per raindrop.  Drop size, shape, and turbulence in the storm will markedly change the velocity for drops of the same mass (I think).  I don't see how you would ever draw a correlation between an impact magnitude and a volume of water collected.

just use a bucket and put it on a precision scientific weight scale.  Just see how heavy your water is! 

To solutions point, what are you trying to accomplish with the improved accuracy

[intel]

Gallymimu moving rapidly towards solving the problem. "solutions", the (for the solution),gives all the necessary information.. To solve this problem, a $ 50 Good luck ...

    http://en.wikipedia.org/wiki/Rain_gauge

[Vineyards]

I also thought about piezo sensors. I also elaborated the idea by calibrating it by obtaining noise signatures according to droplet size and and speed finally totalizing the masses obtained this way. The sensors may need a hydrophobic coating. PTFE or a nano tech coating might do the job (coating life may be a problem though)

[optikon]

Zac, I can appreciate your drive to develop something that is better than what exists today. But you have to ask yourself some difficult questions that Solutions was getting at. First of all, who out there has this problem of needing better than 0.01" / trace detection? Have you talked to people in the field about this? Are they telling you that they wish they had something like this? And what would they pay for such a beast? Secondly, a device that can measure very small trace amounts of rain is not difficult from a technology /conceptual standpoint so here's the thing, if there was a real market for it, it would already exist - in droves, if the need was great. That's how this world works. Plus, this competes with a cheap rain sensor.

If you are after a product to sell idea, I am telling you do your research so you understand that you are actually solving problem and addressing a market need. I think you will find there is no/small demand in part, due to some of the reasons Solutions has pointed out.
If you are just trying to build something neat & cool, I like that and am mildly interested in what you will come up with, plus I enjoy readings Solutions rant/sarcasm.

[Ichan]

This discussion make me do some search, i guess not the resolution of tipping bucket is the problem but the minimum threshold of the measurement.

This one is very-very interesting http://www.rainsensors.com/ and just on the budget ($49 for three or more).

How it work?

-ichan

[zac]

This discussion make me do some search, i guess not the resolution of tipping bucket is the problem but the minimum threshold of the measurement.

This one is very-very interesting http://www.rainsensors.com/ and just on the budget ($49 for three or more).

How it work?

-ichan

http://www.rainsensors.com/how_it_works.php

I haven't seen this method before.  This looks like a good cheap way to detect trace amounts of rain, but not so good at quantifying it. 

"Not as accurate as a properly maintained Tipping Bucket, but a better choice for many applications.
The RG-11 totalizes a statistical phenomenon, so we do not specify an accuracy limit.  Sometimes the RG-11 will read a bit high, and sometimes a bit low, and most of the time these readings will be within 20% of the readings from a tipping bucket.   Over a long period of time these generally average out.  A few notes:

    A tipping bucket will typically miss the first couple of hundreds of an inch of rain as the collecting funnel fills.  The RG-11 starts totalizing right away, and is thus more accurate than a tipping bucket for small rainfall accumulations..
    If the tipping bucket mechanism gets stuck for any reason, the tipping bucket will not read at all.  If the tipping bucket becomes clogged it may not read.  The RG-11 is more accurate in these conditions.
    The RG-11 will generally not detect snow.  A tipping bucket will read the snow after it melts.
    We know of no above-freezing condition that will cause the RG-11 to fail to detect rain.
    We know of no dry condition that will cause the RG-11 to falsely indicate rain."

http://www.rainsensors.com/tipping_bucket.php


Posted on: August 30, 2013, 05:25:26 17:25 - Automerged

Zac, I can appreciate your drive to develop something that is better than what exists today. But you have to ask yourself some difficult questions that Solutions was getting at. First of all, who out there has this problem of needing better than 0.01" / trace detection? Have you talked to people in the field about this? Are they telling you that they wish they had something like this? And what would they pay for such a beast? Secondly, a device that can measure very small trace amounts of rain is not difficult from a technology /conceptual standpoint so here's the thing, if there was a real market for it, it would already exist - in droves, if the need was great. That's how this world works. Plus, this competes with a cheap rain sensor.

If you are after a product to sell idea, I am telling you do your research so you understand that you are actually solving problem and addressing a market need. I think you will find there is no/small demand in part, due to some of the reasons Solutions has pointed out.
If you are just trying to build something neat & cool, I like that and am mildly interested in what you will come up with, plus I enjoy readings Solutions rant/sarcasm.

A manager at a company that makes weather stations mostly for the commercial/government markets asked me about this so apparently there is interest.  But, the companies that make this equipment are quite backwards and often don't keep up with technological developments.  His technical people were telling him it couldn't be done except at outrageous cost (>$500 manufacturing cost in 10K quantity, not including NRE). 

If someone in the US doesn't do it, I suspect someone elsewhere (china, eastern europe, etc.) will.  I was looking at some medical gear (pulse oximeter logger that does perfusion index and ECG logger) and was shocked at how cheap the stuff coming out of china is (Contec brand).  A complete wrist mounted data logging pulseox with windows software is only about $100 delivered.  12 lead ECG machines are about $400-900.   Ultrasound machines in laptop format are $1500.  They're selling them on websites like dhgate.com and on ebay.  I got a logging pulseox and an 12 lead ECG holter logger and both were pretty good though their software on the PC needs some work.  Support is pretty horrible though. 

[Gallymimu]

If someone in the US doesn't do it, I suspect someone elsewhere (china, eastern europe, etc.) will.  I was looking at some medical gear (pulse oximeter logger that does perfusion index and ECG logger) and was shocked at how cheap the stuff coming out of china is (Contec brand).  A complete wrist mounted data logging pulseox with windows software is only about $100 delivered.  12 lead ECG machines are about $400-900.   Ultrasound machines in laptop format are $1500.  They're selling them on websites like dhgate.com and on ebay.  I got a logging pulseox and an 12 lead ECG holter logger and both were pretty good though their software on the PC needs some work.  Support is pretty horrible though. 

That would be pretty ballsy to put a chinese ECG on yourself.  I wonder how many safety standards it complies with.

[zac]

That would be pretty ballsy to put a chinese ECG on yourself.  I wonder how many safety standards it complies with.

Well, it supposedly has a 510K FDA equivalence.  But, I suspect a device powered by a small battery is pretty safe. 

[Gallymimu]

We'll if it's got a 510K approval then it must have gone through FDA so SAFE IT IS!

[zac]

We'll if it's got a 510K approval then it must have gone through FDA so SAFE IT IS!

Keep in mind that 510K equivalence is a much less rigorous process than the process to get the original approval.  Stuff like this for routine monitoring should be ok, but I don't think I would want a chinese brand pacemaker, at least not in the near future. 

[kermit]

solutions made a valid comment about distribution.
Perhaps a network of sensors spread over an area would improve the accuracy of the measurement but not for $50 I fear.

[Langley]

Zac,
Below are a couple of commercial examples of the optical and piezo technology sensors being discussed, including a nice little overview of the piezo signal processing.

Both are capable of detecting appreciably smaller precipitation amounts than tipping-bucket sensors, albeit at considerably higher cost. I’d be skeptical of the accuracy claims for the WXT520 piezo sensor if it were from anybody but Vaisala, but I hope to see for myself soon as I should receive a loaner unit next week. I plan to compare it to a couple of conventional tipping buckets (and a NWS Standard  Gauge) for a couple of months before I really trust it, though.

I have no experience with Thies designs, though they’re well regarded, but Vaisala is the world leader in high-rel weather instruments, providing many of the sensors in the U.S. ASOS/AWOS networks, for example.

http://www.thiesclima.com/disdrometer.html
http://www.wmo.int/pages/prog/www/IMOP/publications/IOM-82-TECO_2005/Papers/1%2806%29_Finland_4_Salmi.pdf
http://www.vaisala.com/en/products/multiweathersensors/Pages/WXT520.aspx

[zac]

Zac,
Below are a couple of commercial examples of the optical and piezo technology sensors being discussed, including a nice little overview of the piezo signal processing.

Both are capable of detecting appreciably smaller precipitation amounts than tipping-bucket sensors, albeit at considerably higher cost. I’d be skeptical of the accuracy claims for the WXT520 piezo sensor if it were from anybody but Vaisala, but I hope to see for myself soon as I should receive a loaner unit next week. I plan to compare it to a couple of conventional tipping buckets (and a NWS Standard  Gauge) for a couple of months before I really trust it, though.

I have no experience with Thies designs, though they’re well regarded, but Vaisala is the world leader in high-rel weather instruments, providing many of the sensors in the U.S. ASOS/AWOS networks, for example.

http://www.thiesclima.com/disdrometer.html
http://www.wmo.int/pages/prog/www/IMOP/publications/IOM-82-TECO_2005/Papers/1%2806%29_Finland_4_Salmi.pdf
http://www.vaisala.com/en/products/multiweathersensors/Pages/WXT520.aspx

Do you have any idea what the cost is on these?

[Langley]

I was quoted USD $2140 on the WXT520 Weather Transmitter (which includes an ultrasonic anemometer, etc.). A quick Google of the Thies Laser Distrometer turned up a price of 2600€.

Clearly these prices are way out of reach for home/hobby use, but are typical of professional quality sensors, with the WXT520 actually pretty cheap as such things go.

My reason for posting the info was just to illustrate that either of the technologies could be, and in fact, are used for high accuracy precipitation sensing, though.