MOD54415 RTC Month Error

[gavinm]

Folks, the 25MHz processor crystal is likely to be far from perfect! Probably less accurate than the 32kHz which are specially manufactured for RTC applications.
Like you my application may never be connected to the internet or even a LAN where there might be a time server. The best I can expect is that there may be a PC connected to a very local network of a few instruments using NB modules and this will be used to configure them via a web interface and possibly collect results data (that has been time/date stamped by the instrument). So on the assumption that the clock on the PC will be correct (probably a notebook that has been connected to an IT LAN recently), it will be my source of setting the RTC in the NB module.

Good luck with your projects.

[Forrest]

We offer an NTP server that could work for you. If you can receive a GPS signal, you are set.

http://www.netburner.com/products/netwo ... k70-ex-ntp

Folks, the 25MHz processor crystal is likely to be far from perfect! Probably less accurate than the 32kHz which are specially manufactured for RTC applications.
Like you my application may never be connected to the internet or even a LAN where there might be a time server. The best I can expect is that there may be a PC connected to a very local network of a few instruments using NB modules and this will be used to configure them via a web interface and possibly collect results data (that has been time/date stamped by the instrument). So on the assumption that the clock on the PC will be correct (probably a notebook that has been connected to an IT LAN recently), it will be my source of setting the RTC in the NB module.

Good luck with your projects.

[gavinm]

Hi Forest,

Yes thanks. I've seen that. I think GPS signal is a bit limited inside a factory!
More likely to get access to some internal LAN time server, or just have a weekly / monthly procedure to manually check and correct the time.

All depends a bit on the likely accuracy. I'm running a longer term test right now. I'll report back later today.

Thanks for the help.

[mx270a]

Gavin, you are correct that GPS doesn't work well inside a building, especially anything with a metal roof. You would want to put the GPS antenna on top of the roof, and have NetBurner NTP server just inside the building. Then you connect it to your LAN via CAT5 like any other device.

[gavinm]

See related post on RTC accuracy
http://forum.embeddedethernet.com/viewt ... f=5&t=2097 on MOD54415 RTC Accuracy.

The original reason for this post was to highlight the errors setting / getting month data to / from the RTC in the MOD54415. This has been well covered here and we're drifting a bit "off topic". I expect eventually NB will issue a fix to the functions in C:\nburn\MOD5441X\system\mcf5441x_rtc.cpp.

In the mean while we can just use local copies with the fix, as has been suggested here ..

Thanks for everyone's help and input ...

[Forrest]

Dragging out this old post. Release 2.7.3 still had a problem with the MCF5441X RTC rollovers involving months. This has now been fixed. If you are using the RTC, you can confirm the issue with the attached unit testing application that will run on the device. If you need a fix now, go ahead and create a support ticket and I'll send you the new version of the code. Otherwise, await 2.7.4 release for the fix.

Source: https://gist.github.com/fstanley/c13b04d8ca7510320135

[gavinm]

Wow - Forest, 9 months - that simple fix took some time!! Have you all been on an extended holiday...

Gavin

[gavinm]

By the way, I've had 3 x 54415 modules in instruments on site for over 6 months now with the RTC correction "tuned" as in my previous posts, and I visited site this week to find they were all with 30s of the correct time. Two were a little fast, one was a little slow.

[Forrest]

Wow - Forest, 9 months - that simple fix took some time!! Have you all been on an extended holiday...

Gavin

Ha! The problem was fixed, but the fix introduced a new bug that was pointed out by a user on support (which was basically the same bug as before, the month conversion between a struct tm and RTC code)

[Forrest]

By the way, I've had 3 x 54415 modules in instruments on site for over 6 months now with the RTC correction "tuned" as in my previous posts, and I visited site this week to find they were all with 30s of the correct time. Two were a little fast, one was a little slow.

Not bad on that drift. You may be interested in a new 5441X NTP product we are releasing soon. It uses GPS, can use GPS initially just to set time, set via an NTP server, or can be set manually. I believe it includes a crystal which is accurate to .5 PPM, which means at absolute worst case (without a GPS lock or network connection) it would drift less then 16 seconds a year.