Cumulus Support

Just downloaded the software and it is working great with my Zephyr Instruments PWS-1000TD (once I figured out to use the FineOffset station type). The system read my EasyWeather database and all appears to be working well, with one exception: I am getting strange readings for Average Temperature. On the main page, it shows my current temp as 69F and the average temp as 18F. I live in San Diego...don't think we've ever had 18F, not since the last ice age. I checked the monthly and yearly views, and they show similar errors for the 1-minute mean. My min and max temps over the past 2 months go from 57F to 99F, but the 1-minute mean shows 21F.
Since I am only sampling every 20 minutes with the EasyWeather database, is Cumulus counting a bunch of zeros in the mean? I searched all over the forum, but could not find a way to resolve this or understand why it is coming up with such a low value.
Thanks for the great software and for any help!
-MXC

You didn't start off using Cumulus set to Celsius and then switched to Fahrenheit, by any chance? If not, then it sounds like it's got some bad data from somewhere. The 'this month' display gets its average from the sum of the average temperatures in dayfile.txt; if you have a look in there you may be able to spot the problem. If it's not obvious, upload it and I'll take a look.

I don't understand what you mean by "the system read my Easyweather database". Cumulus doesn't have that facility. Do you mean it downloaded the data from the station logger? Actually, at a 20-minute logging interval, the logger would hold about 2 months worth of data, so I guess that is probably what you mean.

No, I set all the units as soon as the startup screen came on, so it was set to F from the start. My hardware is also set to F. I must admit, if you convert those celsius temps to fahrenheit values, those would be reasonable averages...
I did notice something odd in the dayfile - the initial min temp was zero, so I set it to be similar to the line below it. However, all the average temps in that file appear incorrect. I can't see anything else...file attached, plus screen shots of the s/w.
Yes, sorry...I thought it was reading from the easyweather.dat file, but now realize it is just getting data from the local storage in the device. I have only had the station installed for about a month.

Thanks. That's very odd, it does look like I'm doing the 1-minute mean in Celsius! I'm amazed no-one else has ever spotted this! I'll check the code. If it's true, it means everyone using Fahrenheit has a complete set of incorrect average temps in their data files

I also see from your 'this month' screenshot that there is a problem with the labels being truncated at your resolution; I'll see if I can fix that. What display resolution is that?

Yes indeed, the problem is that when using data from the Fine Offset logger, Cumulus calculates the running average in Celsius, regardless of the configured units

I'll get a fix for this out shortly. Thanks for bringing it to my attention.

Thanks Steve. Glad it was so easy to resolve. As a displaced Brit, I should be using celsius, but after years of hearing/seeing units in F/ft/miles/in etc, it's just easier to digest for me that way, especially when I glance quickly at the display. I guess most of your users must be using celsius...
Regarding the screen resolution, I have a full HD display (1920x1080). This is really nice, but I found it much easier to read with the font size at 120 dpi. For most things, this works great, but occasionally it causes a problem with text lining up in some windows. Also, some icons in the system tray are "ugly" if the programmers don't provide a large icon (Skype, for example). I wish there were a cleaner way around this...I was hoping it might be resolved in Windows 7.
-MXC

I've just uploaded a new build (see Announcements and News) which should fix the average temp problem (for all data recorded from now on). I've also tinkered a little with the layout of 'this month' and 'this year', but not yet 'this period'.

Thanks! Will download and test.
I assume I can just take all those final values in the dayfile.txt file and convert them to Farhenheit values? That should fix my historical 1-min averages, right?
-MXC

mxcrowe wrote:I assume I can just take all those final values in the dayfile.txt file and convert them to Farhenheit values? That should fix my historical 1-min averages, right?

Yes - assuming they are all calculated from logger data. Any that are part logger and part 'live' will be an odd mix of C and F, but it should be obvious if that's the case. If any do look wrong after converting, you could substitute (max+min)/2 as an approximation.

Ok, I installed 1.8.8 (858). I can see several nice improvements.
I modified the dayfile to have all F values, and that appears to work correctly now for "this month" and "this year" values.
The average temp on the "this period" page is still off, but I imagine it still has some C values for this day (it's 10am here and is sampling every 20 minutes). Perhaps this day will end up incorrect, but future days will be correct? If so, I can just fix this day's data in the dayfile tomorrow.
Just to give you the numbers, my overnight low (low for today) is 53.2F and it is now 68.5F (high so far), but avg tmp is showing 26.6F. It does seem to be going up faster than the instantaneous temps.
Regarding the screen alignments, it definitely improved it. I have included a screen shot so you can see the result on my end.
-MXC

I should have mentioned that the average so far today is saved in today.ini (as a total and a number of samples), so you could have corrected that. You still could, I suppose, but as it's constructed from a mix of C and F, that's probably a bit difficult.

Thanks for the screen shot. I'll do some more tweaking for the next build, if I get chance.

I can see how you are calculating it in the today.ini file. Yes, it is mixed up now, so I guess I'll just wait for tomorrow to replace that value in the dayfile with a valid, estimated F temp.
I'll keep playing around with things and report what I find.
Thanks again!
-MXC