New rain rate calculation

[goldrush]

I must echo Frank's point.
My interest is primarily in the Wind (I know I should have a station with a faster update time, but 2 La Crosse I have tried were not as good, or reliable overall as the cheapo Fine Offset!).

There is a noticible difference in the wind display with the new build, such that one can hear a gust a long time before it is reported in Cumulus and I have reverted to the previous version.
Certainly, in the extreme, if Cumulus polled the station continuosly, Gusts (and other data) would appear on sceen as soon as they are reported (plus processing time) by the station (agreed only at a max repeat frequency of 48secs)
Now, assuming both timers are accurate, the station can report a change 48 secs before Cumulus recognises it because the timers are not synchronised.
I believe that in practice, the occurance time of gust "spikes" are not recorded by the station anyway, the station only reports the highest peak during any 48sec time period.
Thus, the time such are reported by Cumulus could represent a real time error of up to 96 secs.

I do recognise that generally, it is percieved as of less importance to know the value of wind at specific times, than to record correct rainfall and that Cumulus, as with other such software and "amature" weather sations, is not really designed to represent true "real time" displays.

Guess by the nature of things, it's all a compromise.. you win some.. you lose some.

I suppose in my case the answer is to crack the long update time from the Fine Offset.
Problem is.. "I'm not very clevur.. but I can lif evy fings"

[steve]

This is really why I originally had the update interval at 15 seconds, even though I knew the station only updated every 48s, because I couldn't synchronise with the station. And I realised that in changing it, it would mean that potentially there was a further delay of up to 48s before the data changed. I don't really see this as a problem. I think some people get a bit obsessed with 'real time' rapidly changing data, and in that case, they should buy a Davis for 10 times the price and get updates every 2.5 seconds! I'm not pointing the finger at anyone, just making an observation.

I'm rather more concerned, however, by the possibility that it's missing updates altogether.

The change to 48 seconds was just a cheat, really, as an easy way to fix my new rain rate calculation. I'll have a look to see how to fix it properly without changing from the 15 second update interval. Which in any case, as you will realise, still means a potential delay of up to 63 seconds between a gust occurring and actually appearing on the Cumulus display.

Steve

[steve]

The drawback that I can observe with the reading every 48s is that it sometimes misses gust/wind readings as the timer doesn't always kick in every 48s. It will sometimes take longer and therefore misses a gust/wind reading.

Ah. You mean the timer runs, but late? It should still pick up the latest reading, if the station only updates every 48 seconds, the timer would have to be more than 48 seconds late to miss a reading?

Steve

[goldrush]

I do have a need to get close to "real" time wind.. as opposed to Baked Bean flatulance and realise that the only real way is to spend lots of gold coins (or indeed go back to the old mechanical/electrical dial guage and mark 1 eyeball).

By the way, it was not a complaint and not related to this thread.

1 strange thing I have noticed, (and again not a complaint and I believe that the ftp part has not been written by you) which may well be my set up, or web host. (Vista 64bit, Linux based web host) is that if someone actually requests a page from the site at exactly the same instant Cumulus is updating by ftp, that specific page gets corrupted (usually resulting in a zero bit file) and does not get uploaded/replaced, resulting in the usual 403 error displayed for that page until the next ftp update.
Do not think it is the Host as I am allowed simultaneous ftp uploads and with other softwares I do no seem to get the same problem
Obviously they say it is not a problem with their system.
If I ever get to the bottom of it, I will let you know.

"Save the Planet.. Don't frown.. Smile.. it uses less energy"

[steve]

By the way, it was not a complaint

No - understood, and my response is not a criticism of you or anyone else. Whilst sometimes the requests I get do baffle me somewhat, I do appreciate that we're all different and all have have different wants or needs

1 strange thing I have noticed, (and again not a complaint and I believe that the ftp part has not been written by you) which may well be my set up, or web host. (Vista 64bit, Linux based web host) is that if someone actually requests a page from the site at exactly the same instant Cumulus is updating by ftp, that specific page gets corrupted (usually resulting in a zero bit file) and does not get uploaded/replaced, resulting in the usual 403 error displayed for that page until the next ftp update.

Are you using the 'rename' option?

Steve

[goldrush]

Steve
I did not tke your previous remarks as a criticism.. thats one problem with e-mail.. easy to get the wrong impression

I have tried both using the rename and not.. similar result.
As I say, probably not your problem

Like others, I too cannot really understand how on earth you find time to work on Cumulus and respond to people like me.. thanx

[steve]

I have tried both using the rename and not.. similar result.
As I say, probably not your problem

Odd that it fails even with the rename option.

Like others, I too cannot really understand how on earth you find time to work on Cumulus and respond to people like me.. thanx

Easy. I get up in the morning and switch on this computer, and my work laptop over there --->

I alternate between the two computers all day, occasionally eating, or exchanging a few words with this woman who appears from time to time who I think must live with me, then I switch the computers off and go to bed. But not every day is quite like that. And I am definitely going to have to cut back on Cumulus time soon.

Steve

[Frank02]

Ah. You mean the timer runs, but late? It should still pick up the latest reading, if the station only updates every 48 seconds, the timer would have to be more than 48 seconds late to miss a reading?

Steve

Hi Steve,

I did observe it more carefully and here is what's happening.

Sometimes the timer doesn't kick in and will kick in at the next 48s intervall. Really don't know what causes it to do that.

So here is my understanding of what seems to happen:

- The station reports a 40kmh NE wind
- Cumulus timer doesn't kick in at 48s so doesn't read that value
- The station reports a 10Kmh N wind 48 sec later
- Cumulus timer kicks in at 96s and report the 10Kmh N wind
- It misses the 40Kmh NE wind reading

When you were reading the data every 15sec, if the timer doesn't kick in, it will pick it up at the next timer intervall, 30sec (even at 45sec if it also skip that timer but I only saw it skip the timer intervall once).

I guess the problem is with Windows XP as it should not skip a timer intervall like that.

Should you read the data every 24sec for example, it would pick it up on the next reading even though it skips a timer interval.

As someone else said, I guess that it's a balance between calculating the rain rate and getting the wind readings. As far as other data goes (temp, hum, pressure), it doesn't matter if it skips a reading as it will synchronize on the next reading. In the case of the wind, it's a bit different.

The reason why I notice it is because the station recorded a high gust record while it didn't appear in Cumulus. Since the station records the data every 10 minutes in memory, I guess that this high record was not in the station memory either?

Or maybe there is something that I don't understand in the way those stations work and how Cumulus interact with them as far as reading the data goes.

Thank you.

[steve]

It's a bit worrying that it can miss timer events like that, my understanding of how the Delphi/Windows timers work is that they might be a little bit late if the main thread was busy, but they should happen. There are other timers in Cumulus which would have much more serious consequences if they went missing.

Anyway, I think I'm going to change the timer back to 15 seconds, and solve the rain rate calculation some other way. In fact, I think I'll make it a 10 second timer, to reduce the potential time lag slightly between the station updating and Cumulus reading. I wonder how Easyweather does it - I assume it updates at pretty much the same time as the console? It must have some way of synchronising with the station updates.

The reason why I notice it is because the station recorded a high gust record while it didn't appear in Cumulus. Since the station records the data every 10 minutes in memory, I guess that this high record was not in the station memory either?

Or maybe there is something that I don't understand in the way those stations work and how Cumulus interact with them as far as reading the data goes.

It probably was/is in the station memory, but Cumulus doesn't use the station memory at all during normal running, in the way that Easyweather does. It generates its own log entries from its internal data.

Steve

[Frank02]

Hi Steve,

Easyweather seems to read the station every 20 something seconds. When I check the easyweather log file, the record doesn't show in that log file either as it didn't happen at exactly the 10 minutes interval recorded in the file. It did happen in between recordings. Easyweather shows that record probably because it reads it from the station memory (the station as a memory slot to keep track of the high wind record).

Anyway, just though of letting you know.

Thank you.

[dane]

I wonder how Easyweather does it

I just ran EasyWeather for a little while.
It reads the station every 16 seconds (or thereabouts - 16 to 18 seconds between reads).
It is not synchronised with the station (there is quick flash on the station above the outside humidity every time it updates its memory).
I, too, have the impression that Cumulus misses a read every now and then (only based on observation of the time stamp in the lower right hand corner). So I, too, would suggest you go back to the 10-15 seconds interval.
Do you want me to turn on the debug log for a while and try to analyse the output and look for missed reads?

[goldrush]

Steve.. just for info ref my apparant ftp "problem".
Interestingly it does not seem to be a problem when Cumulus is run on XP, but niether does it appear to be a problem when running Cumulus on Vista when run as a combination of "not administrator", "not in Xp mode" and "not installed into Vista Program files (X86)" folder.

Of course this may simply mask the real problem which I am begining to think is something odd with my Web Host, as if Vista related I assume someone else would also have a similar problem.

Incedentaly, installing Cumulus in a folder e.g C:/Cumulus instead of the normal default also gets over the stupid "Vista Virtual Store" problems.

"Some people live and learn.... me I just live"

[steve]

It reads the station every 16 seconds (or thereabouts - 16 to 18 seconds between reads).
It is not synchronised with the station (there is quick flash on the station above the outside humidity every time it updates its memory).

That's interesting to know. I assume therefore that the actual figures update every three reads?

Do you want me to turn on the debug log for a while and try to analyse the output and look for missed reads?

I'm not sure it's worth it, I don't think we'd be able to work out what the problem was anyway. But please don't let me put you off if you're keen to have a go

Steve

[steve]

Steve.. just for info ref my apparant ftp "problem".
Interestingly it does not seem to be a problem when Cumulus is run on XP, but niether does it appear to be a problem when running Cumulus on Vista when run as a combination of "not administrator", "not in Xp mode" and "not installed into Vista Program files (X86)" folder.

Of course this may simply mask the real problem which I am begining to think is something odd with my Web Host, as if Vista related I assume someone else would also have a similar problem.

Yes, it difficult to see how Vista at 'this' end could cause that problem at 'that' end, but hey, stranger things happen. This is a computer and the internet we're talking about, not something logical and determinate

Incedentaly, installing Cumulus in a folder e.g C:/Cumulus instead of the normal default also gets over the stupid "Vista Virtual Store" problems.

Yes, I added a recommendation about this to the Readme file recently.

Steve

[goldrush]

..........Yes, I added a recommendation about this to the Readme file recently...

I never would RTFM