Pyranometer with modbus connection

[John Gorman]

Hi,

I was interested in collecting some data from a pyranometer, and wondering if anyone had any experience with any pyranometers that are solid. We like Modbus in general so that would be the preferred method but interested in any recommendations for quality instruments.

Thanks, John

[John Gorman]

We're trying out the Rika RK200-03 pyranometer and it seems to be working well. It's pretty easy to add it to a SolarNode (running on a Raspberry Pi) as a modbus device and then chart the data.

[John Gorman]

some more technical info about this test implementation:

http://solarnetwork.net/v4/testing-the-modbus-device-plugin/

[Rob Beckers]

John, this is interesting info! I didn't know that the R-Pi had a project for PV data collection (and can talk MODBUS).

Regarding that pyranometer: The type you use looks to be one that collects all radiation in a 180-degree arc. For predicting or verifying solar PV output this is less useful, as you need radiation in-plane with the panels to do that. From the output of your pyranometer you would have to calculate from the position of the sun what the flux is that is in the same plane as the panels.

But maybe that's not what you're using this for at all...

-RoB-

[John Gorman]

Hi Rob,

Thanks for your reply! Yes, we have a open-source platform that has two parts so far - a low-powered client (SolarNode) that can use the Rpi or a few other SBCs to gather data and a cloud repository (SolarNet) that captures all the data and makes it available for charting/analysis/action etc.

What this pyranometer integration is about is getting some more data about the "context" of how an array produces kilowatt hours each day. There are a ton of variables clearly, and irradiance in GHI measurments are only one of them - local shading, wind and brand of gear, kind of DC/AC wiring etc. all contribute to the final number of watt-hours produced arguably - but in different weights really.

a new module we are developing (SolarQuant) is aiming to take measurements of the "context" that a solar array operates in - time of year, time of day, sky conditions, temperature, possibly panel temperature, barometer, humidity etc. etc. and feed those to a neural network. this technology is new is getting very common now and there are some multiple open-source platforms to work with. We know that there will be some very good science to determine production numbers based on angle and time of the year and type of PV module, and DC/AC losses from the inverter and wiring. But we also think that from an empirical point of view, there might be details in the context data that help direct which elements of the "context" matter more than others. This is really what the neural network does - sort through ALL the possibilities and zero in on what works, in a non-linear way. (needs a ton of processing power and time currently but this is improving I believe.)

Do you think that a POA device mounted in the plane of the actual array - maybe a silicon based one rather than a GHI thermopile mounted horizontally and open to all frequencies - is going to get us a more relevant "input" for our calculations? I am reading up on the GHI-to-POA conversion so thank for that advice, but we might have to add a second pyranometer to this test rig! (which is easy as you can see :bigsmile: )

Thanks, John

[Rob Beckers]

John, very neat! :cool:

The norm for monitored arrays is to use a silicon based cell mounted in-plane (with glass over it, just like a mini PV module and in fact a small PV module would do this beautifully). From that it's more straight forward to calculate if the array output is in line with irradiation, and an alarm can be fired off if a string of inverter falls below the limit band.

This way of measuring is more relevant because it behaves much more like the panels will. Panels do nothing with sun hitting them from behind (happens in summer here both morning and evening, if the panels face south), while a pyranometer will still register lots of light.

-RoB-

[John Gorman]

Thanks Rob, yes I hear you about the light happening anyway even though the sun is behind the panels - that makes me think we need to test a POA sensor. these tend to be a lot less expensive than the GHI thermopile ones I hear? In our experiment, we will see what inputs are going to contribute to a good guess on the performance so I'll be updating the forum, once we have more results. the computational requirements to run SolarQuant are pretty high, but the advances with neural networks made from using the chips on video cards (such as nVidia) are rapidly progressing so we'll see what's possible.

[Rob Beckers]

This is what I had in mind, it even talks ModBus: https://imtsolar.com/products/silicon-irradiance-sensor/

Looks expensive though! Maybe there are cheaper ones, in fact, in the spirit of the project you'd use just a cheap cell, plus temp sensor, and put an arduino or Pi in to read it and convert to ModBus. Would be a nice project... If only I had more time... :unsure:

-RoB-

[John Gorman]

I will be interested in how much these cost as well but they look very high quality - thanks Rob.

I think it is a balance between instrument cost and data quality but not sure where that line lies. we definitely think that peer performance review of and array or panel located nearby should have some value, but there are a lot of variables that change what you're measuring. so a calibrated "off-the-shelf" precision device like this one, located right at the array location, as a repeatable standard would be useful I think.

[Rob Beckers]

Drop 'm a line, let me know when you find out! The whole device exudes a high cost number, but who knows, we may get pleasantly surprised. Looks incredibly well made, and as you say it's hard to pass up on something that can produce high quality data out-of-the-box.

I tried Googling for price, but none seems published on the Internet. I did find that some monitoring companies use these IMT Solar products as their sensors.

-RoB-

[John Gorman]

Cool, I will let you know what we find out - there seems to be another light sensor here that's silicon based:

http://rikasensor.com/product.asp?bm=BANK201312910626

but do not know if it has the same type of sensitivity, certainly not going to beat the IMT in quality but we're impressed so far.