Cheap pyranometer/irradiance/insolation sensor thoughts

[StBarnabas]

Hi All
If there is one thing I would really like it is a cheap sunlight sensor. By cheap I probably mean around £10, though I personally would happily pay up to £25 for a decent one. We are now pretty close to having a 1-wire interface and could for example produce a pseudo temperature of 100 degrees for 1kW/m^2 and pro rata for other irradiances. It might be nice also to have both a vector (cosine theta) sensor and a scalar sensor (sensitivity independent of angle). I have my own ideas on this and will be contacting various people in the new year.
It would be nice however to get a group going on the forum as there are a number of people interested in such a device and ideas are always welcome. I have put up a poll to see if there is any interest and how much people would be willing to pay.

Sean

[desperate]

Hi StB, hows things up there?

This may be a silly point but isn't a PV panel an insolation meter, surely the output is related to the insolation, isn't it?

You can probably tell I know almost nothing about PV

Desp

[StBarnabas]

Desp
a short circuited PV panel is a pretty good irradiance sensor. The problem with working PV is that the maximum power point tracker continually changes the operating point which makes things a bit difficult.

Regarding weather - very cold was -6.3 centigrade first thing this morning, but no wind and very little snow. Was sunny all day and the PV made c 7kWh of 'leccy  which is not bad given the day length at the moment.

Sean
 

[wookey]

have you considered the hobby-boards sensor:
http://www.hobby-boards.com/catalog/product_info.php?cPath=22&products_id=82

I've got one of these, but have failed to stick it on the roof so far. I have now got the missing ping-pong ball (thanx kristen) so I just need to get round to it.

[StBarnabas]

Hi Wookey
I looked at these about a year ago. I though too expensive and not good enough. I am confident we can do far better.
StB

[EccentricAnomaly]

I'd have thought the two problems would be:

1) getting the right angular response (hemispheric or cosine) without various reflections and refractions messing things up and

2) getting a calibrated source of light to compare against, particularly one with a suitable spectrum.

[StBarnabas]

EA
Yes 1) is a big problem. 2) fortunately is OK as I have access to a good source - though the best minds I think say that sensors should be calibrated by sensors and sources by sources. Have access to both so we should be OK here
Sean

 

[wookey]

The H-B sensor certainly isn't the cheapest you could make. But it is conveniently availa le so I thought I'd try one. I've seen people recommend fitting the sensor inside a ping-pong ball in order to avoid errors due to sun angle and the sensor getting dirty on condensed-on. Seems plausible. I have no idea at all how you'd calibrate this thing. I was assuming it'd only really be any good for relative measurements. I was stalled on 'not having a ping pong ball and being too mean to pay for one I was going to drill a hole in' for about a year :-)

[guydewdney]

A lot of cars with climate control have solar sensors - my old S class (W140 series - 1990's) did - and my old Navara did - I suspect the sensor is quiet cheap.

A quick fleabay search reveals them for Saabs at 4 quid.
http://cgi.ebay.co.uk/SAAB-SUN-SENSOR-4541728_W0QQitemZ370247232580QQcmdZViewItemQQptZUK_CarsParts_Vehicles_CarParts_SM?hash=item5634746c44

[JoergenB]

Hi all,

Did someone end up building a pyranometer?  

I have ordered the relatively cheap (part list available, or $20 + European shipping) pyranometer design by David Brooks:
http://www.instesre.org/construction/pyranometer/pyranometer.htm  

Apparantly someone benchmarked it against a 10x more expensive Apogee unit with good results (http://www.builditsolar.com/References/Measurements/KitPyr/KitPyr.htm).  

Only "problem" I've run into is that it seems to require 12 bit resolution in the ADC; I was planning to use a 1-wire DS2450 which is available locally, but realised that it offer only 8 bit real resolution, despite a 16 bit claim on the first page of the datasheet. I have not managed to find a alternative, so will be grateful for any hints.

Regards,
-JB

[wookey]

I installed mine on the roof for a while but failed to record any data. It's currently sat on my desk again waiting for more important stuff to get done. I realise that's not very useful info.

That apogee pyranometer (and kit) looks very interesting. The bit about 12-bit resolution is because his decide was 12 bits over 2.5V, and the largetst voltage this device puts out is 0.2V, so you only get less than 1/10th of the nominal 12-bit resolution, so a scale of about 0-320. So if your 8-bit device can be arranged to be 255 at 0.2V then you'll only get slightly worse resolution. Hmm, I see the DS2450 is FSB at 2.56V so no help there, and what you mean about effective 8-bit resolution. Amplifying the sensor output to 2.5V would be one way to get better results with this A/D

[JoergenB]

Thanks for your swift reply, Wookey.

Amplifying could be an option, yes. I already feed +5V around my 1-wire network, and there seems to be some space available in the casing for some more electronics.

If I understand Brooks correctly, there is no point in designing for a higher voltage as it will affect linearity: "There is a trade-off between the desire to increase the output voltage (by increasing the resistance) and the need to make sure that the pyranometer output remains linearly proportional to the incident sunlight".

My hypothesis is that I need high resolution, as I'm mostly interested in studying the solar potential on cloudy days, and towards the beginning/end of the season.

By 8 bit effective resolution I refer to the table in the datasheet[1] of DS2450 page 22: "In the shaded areas the accuracy is less than the resolution. The conversion results may include random noise.". I therefore assumed that DS2450 was not suitable for this task.

Some more research have led me to the DS2438, according to OWFS voltage table[2] it supports ±250 (mV) range and 0.2441mV of accuracy if using the Vsens input.  Although the pyranometer resistor is much higher than the Rsens the datasheets mention maybe it can work after all? As I find it on ebay for $10, I think I will give it a try.

BR,
-JB

[1] datasheets.maxim-ic.com/en/ds/DS2450.pdf
[2] http://owfs.org/index.php?page=voltage-measurement

[StBarnabas]

Hi sorry this is a bit of a bookmark. Will try to discuss further tomorrow but there are other options

[SimonHobson]

a short circuited PV panel is a pretty good irradiance sensor. The problem with working PV is that the maximum power point tracker continually changes the operating point which makes things a bit difficult.

Could you just get a cheap (tiny) panel and sit it alongside everything else on the roof ?