Accuracy of the clock memory byte in Siemens S7-300

Does someone know how the clock memory in Siemens works? I tried to search specifications from S7 programming manual, but there was no information of the function of clock memory. Is it sycnhronized with real time clock?

In my application I need an accurate one second pulse. Now I'm comparing the seconds of the real time clock to make a pulse, but it's not very sexy way to do it.

I believe that the Clock memory is very accurate, the problem comes when you have additional program scantime added between each pulse.

STL??? said:

I believe that the Clock memory is very accurate, the problem comes when you have additional program scantime added between each pulse.

Click to expand...

Sorry, my question wasn't clear enough.

I my case the clock pulse is used to count seconds in over 24 hours surface treatment, so more important is a minimum discrepancy comparing to real time clock in the end of the treatment. Starting and ending time are saved on scada screen at the beginning of the treatment. Remaining and elapsed time is also shown in the screen.

So if in clock memory pulse is 1 millisecond lag to real time clock, it makes 4 seconds lag in a day. It doesn't look good on the screen.

The clock pulse (or any other S7 timer) is not accurate enough to make a 24hr clock. What is wrong with using the real time clock for this? That's what it is there for.

Also, there are two ways to access the clock. The first is via an SFC, and the other is to directly address the time in OB1. Unless you care how much time has elapsed since OB1 started (and in your case, you don't), then the OB1 time will work fine. Do a search right here, and you'll see it has been discussed several times already.

You haven't mentioned which CPU you are using, but real time clock accuracy isn't a major feature of most PLC CPUs in any case. In the S7-300 series I've just checked a spec which says the 312 clock variability is "<15 seconds / day" and the 319 clock variability is "<10 seconds/day".

There is nothing to suggest what influences accuracy (temp, power-supply irregularities ...?) or what the repeatability is. Also, I assume these are worst-case limits. There is no indication I've seen of what typical figures may be achieved.

Don't go seeking levels of accuracy which are unachievable. How "real" does the real-time need to be?

Ken

If ur using a scada system why not use the scadas time functions. Pretty sure it is linked yo your operating systems time, which is very accurate

chud said:

If ur using a scada system why not use the scadas time functions. Pretty sure it is linked yo your operating systems time, which is very accurate

Click to expand...

Real time clock in PLC (CPU 314) is synchronized via NTP once in every minute. So the real time clock in PLC is very "real time". For communication with scada I use cp 343-1 lean ethernet card.

There is no indication I've seen of what typical figures may be achieved.

Click to expand...

My experience with a 317 - 2 DP and a number of 315 - 2 DPs suggest that the typical value is not far removed from the 10 - 15 sec / day quoted. I could get the time reasonably accurate used the adjustment feature in CPU properties, but sooner or later the time would start drifting off again. In the case of 317 the CPU was in an air-conditioned control room, so I wouldn't have thought it was responding to changing temperatures - no idea what the reason was, I eventually gave up and let it drift.

How OB35 will respond if it is only made to count seconds?

A M 0.0//start
JCN qw1
L 1
L MW 50
+I
T MW 50
L 10//if OB 35 set at its default value
==I
JCN qw1
L 1
L MW 52
+I
T MW 52
L 0
T MW 50//seconds over here
qw1: NOP 0


But Ithink execution time of OB35 can make it drift over a period of 24 hours.

Ok how about synchronizing with nuclear clock via cp343-1 lean;-)