Sandaysoft

Hi Alan.
Just a quick update:
Recorder (and La Crosse aneometer) is now in top of my mast and only a little cabling remains.
The last couple of days there was absolutely no reason to measure any sunshine.
That leaves a little time for the theoretical parts - I just went back from buying a glas container (ø 83 x 130mm) for my next prototype.
Further a have made some preliminary work for how the present the sunrecorder data (simulated data !)
The 24 hour bar have 4 different colors:- black=sun below horizon, white= daylight hours, grey= no sunshine, red=sunshine.
any comments / suggestions for improvemnets etc?
Best regards
Ole

Clarification please

Are you creating 'daylight hours' by calculation or measurent ?

If grey = No Sunshine and red = Sunshine, when does it show White ?

_________________

......................Imagine, what you will KNOW tomorrow !

Daylight by calculation; white is in the future (i.e. possible sunshine, yet to be determined), going by the date and time on the image.

_________________
Steve
Sanday Weather
----------------------------------------------------------------------------------------------------------------------------------
Like Cumulus and want to support it? Please donate!

Thank's Steve - your explanation is exactly the one I were just about to write.
In my software I calculate the sunrise and set times.
Based upon that I will start off a new day by creating the band with black and white and thus ready for the sun detector.
The software will further determ sunshine yes/no and update the band every 2.5 minutes.
I am looking forward to have data available for analyzing especially around rise and set where apenwith exsperienced it is hard to discriminate correctly due to the very low light level.
Since I got the stuff up in the mast, the sun has not been present and forecasts for the next few days are not worth while mentioning

Thank you both - I must be having a bad 'bald' day

_________________

......................Imagine, what you will KNOW tomorrow !

Hi All interested,

Now system in mast and running (notice the sun reflection in the picture).
Just now sky is almost clear - sun set in another 1 3/4 hour.
Hope to get some nice recordings of that.
Best regards
Ole

Hi Steve.
For my further interface programming work, I would very much appreciate your comments on the following:
In Cumulus you calculate Sun Rises and Sun Sets times - I have achieved the same results in my calculations.
The Dawn / Dusk times seems to be when sun is approx. 0.7 degrees below horizon - can you confirm that ?
What will be the best approach considering file format for possible future use in Cumulus:
- Continues records every 2.5 minutes with date/time stamp and boolean info for sunshine?
or
- Only records with date/time stamp and boolean info - True for change to sunshine / False for change to No sunshine?
or
- Records with date/time stamp for start of sunshine periode and date/time stamp for end of sunshine periode?
I believe the last proposal is the best one.

Best regards
Ole

Cumulus uses Civil Twilight, which is 6 degrees below the horizon. Well, it should be doing that, but that doesn't agree with what you're saying.



Yes, I think that one will probably be best.

Steve

Hi Ole
I wonder why you need to time and datestamp the sun on/off times. Don't you need the total duration of sun up to a certain update time. What happens if there are fast skudding clouds and the sunny intervals are only seconds long - it would mean an awful lot of timestamped records. Maybe I'm missing the point but the final useful record is the daily sun total. Or are you recording the total sun duration inside each 2.5 minute block. Would that not serve to give you the sunniest parts of the day if you need it.
The device certainly looks beautifully engineered - and the software looks impressive - you have provided the skills that the original idea needed to cary it forward. Well done
Regards
Alan

Hi Alan,
Below I use abreviations as follows:
Th = Threshold level (ambient light after some data manupilation which have yet to be analyzed and desided upon)
Son = Reflected light > threshold (sunshine)
Soff = Reflected light < threshold (no sun shine)
SSon = Sunshine Status at present
Dcount = delay counter
My plan for the software is as follows
- Make an A/D conversion every 2 sec.
IF SSon
IF Soff THEN Dcount = Dcount +1
IF Son AND Dcount > 0 THEN Dcount = Dcount -1
IF Dcount = 75 THEN SSoff AND Dcount=0 (Sunshine disappeared)
If SSoff
IF Son THEN Dcount= Dcount + 1
IF Soff AND Dcount > = THEN Dcount = Dcount -1
IF Dcount = 75 THEN SSoff AND Dcount = 0 (Sunshine came back)

Said in other words: The change of state (either sunshine or No sunshine) must be present at least 2.5 minutes before the actual state changes.
I think this way of doing it will more og less eliminated the problem with fast changing conditions.
You are correct about that the goal is to measure the actual sunshine hours per day (and month/year), but I intend to use the bar graph presentation which will continiously (or more correctly with 2.5 minutes interval) show the actual sunshine result during the day.
Mind you - the above is only my thoughts so far - at the end of the day result might come up some what different.
The system is as mentioned now up and will record the situation per day with 2 seconds interval.
Results then into Excel for further analyze.
At present I am strugling a bit with noise on my data - the next prototype might end up with some analog interface in the unit.
Thank's your kind remarks - because your started with the basic idea, I will not make only one prototype next but two - one for you and one for myself.
Best regards
Ole

Project Status Update.
The phototransistor with gain interface is a problem because of the very low photo current versus induced noise even if using screened cable.
Have found an alternative from Osram - SFH-5711 which have a logaritmic characteristic which I now think is exactly what is required.
It means for illustration that the signal level change is the same for the range 1 to 10 Lux as for the range 10000 to 100000 Lux.
Further I am now studying the so called 1-wire concept.
Maxim has a four channel A/D converter directly useable on a 1-wire system.
Only a 3-lead cable is required for 1-wire system (+5V, GND and data) meaning all the required electronics can be placed in the sensor unit itself.
In the PC end only a 1-wire/USB adapter is required (found one on http://www.hobby-boards.com/catalog/product_info.php?products_id=1503)

(Submit instead of preview)
The only snack about the photosensor is that it is only available in a SMT package 2,4 x 2,1 mm with four terminals.
So I need a good magnifying workshop lamp with light and the smallest possible tip for my solder iron (I am 67 and eyes have been better).
Best regards

Ole

Hi Ole
At 66 I find the same - I use a headband magnifier - a bit like a sun shield with lenses which I modified with a flip down jewellers eye glass for really close work. We just go to prove you don't lose the enquiring mind even if the other bits of the body don't work quite so well !!
I guess to get a good horizon to horizon line of sight a fairly high mast and long cables are neccessary (my mast is about 20 metres away above a barn roof to see over some trees) - would a wireless link for the data prove any better assuming the electronics can be built into the sensor. Or if in the end the data comes down to on or off could the sensor switch a 5V line ?
As I've said before my electronics knowledge is severly limited so that might be a completely useless idea.
Regards
Alan

Hi Allan.
Good to hear we both are 'products' of the last war - fringe benefits of our age is we don't have to interfere with the ugly phenomena called 'work' (as far as I remember thats the word ?)
I just briefly checked what is mentioned about maximum cable length for the 1-wire system.
Ordinary 4 leads telephone cable can easely achive 30 meters and using better cables like Cat-5 much longer length can work.
I am not still fully confident if 1-wire is the way to go, but it looks attractive in terms of simplicity just with a 1-wire / USB adapter in the PC end.
I do think that making a wireless link is a bit more complicated.

That is a possibility, but actually my plans for the software part is to display the results as shown in my entry of Mon Oct 26, 2009 3:31 pm which will also count the sunshine hours per day, week, month.
At present counting sunshine hours here in Denmark is not an urgent issue - rain counting is more popular.
Best regards
Ole

Hi Ole
I stll do a bit of work to keep the little grey cells working.
Maybe I'm oversimplifying but if you had a 'stopwatch' counting the on time during the 2.5 minute blocks (then reset) the software knows the date and time so your graph would still work.
On another subject I wondered about condensation misting up your glass jar.
Like Denmark Cornwall is grey and days seem short.
Regards
Alan