Bob, Thanks for the response. Since the CPU that queries the clock chip is part of the controller, I wasn't sure that the shorter access time of the newer chips would make a difference or not.
Now, about time correction- I am looking at making this as simple as possible. I don't have an IRLP node at the site, so using any kind of network connection is not an option. Requiring a computer to be running at the controller location is not an option. I also don't want to play around with adding or subtracting seconds to get the clock close- that's too much human involvement. I don't want to be tasked with manually changing the time when Daylight Saving Time starts or stops. Since the WWVB time broadcasts automatically adjust for DST, any method of synchronizing a controller at a remote site to WWVB seems to be the best way to go. Here's one possible solution, offered by a friend of mine: Obtain a simple and relatively inexpensive "atomic digital clock" that has an alarm function. Tap into the alarm beeper circuitry so that a logic level is detected when the alarm goes off. Set the alarm so that it triggers at, say, 0400 hours local time every morning. Install a macro in the controller that, when triggered, will reset the controller's clock to 04:00:00. Voila! Every day, courtesy of the NIST, my controller is always on time. If the DS1643 clock chip is at one extreme of its accuracy tolerance, say two seconds per day, the error could be minimized by jam-setting 03:59:59 or 04:00:01, if needed, to keep the clock within one second of exact. Since the execution of a macro takes time, the jam-set time needs to be offset to compensate for the delay. Perhaps the next generation of repeater controllers will have WWVB or GPS time synchronization built in? (hint hint) 73, Eric Lemmon WB6FLY -----Original Message----- From: [email protected] [mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED] Sent: Sunday, November 11, 2007 2:33 PM To: [email protected] Subject: Re: [Repeater-Builder] Dallas Semiconductor Real-Time Clock Hi Eric, >The Dallas Semiconductor "Nonvolatile Timekeeping RAM" found in many popular controllers, including the Link RLC-1 Plus, is Part Number DS1643-150. We have a lot of experience with the DS1643 and its bigger brother, the DS1644. The S-COM 5K uses the DS1643 (8K RAM), and the 6K and 7K use the DS1644 (32K RAM). A second source for the DS1643 is the STMicro M48T18-100PC1. For the DS1644, a second source is the STMicro M48T35Y-70PC1. >Notice that the "-150" indicates 150 ns access time. The replacement device offered by Dallas/Maxim has either 70 ns or 100 ns access time, and I have no idea if the newer device will work properly where a 150 ns device was used. As a rule, a faster memory is okay. Slower isn't. >I see that the guaranteed accuracy of the DS1643 clock is within +/- 1 minute per month, and there is no capability to tweak the crystal to get better accuracy. That's right. We've been using these parts for many years and the reports from the field range from excellent to mediocre timekeeping. The controllers have "clock tweak" commands that add or subtract seconds and can be called from the scheduler program. If one knows how many seconds the clock is gaining or losing in a day, then automatically resetting the clock daily (or more often) makes for a pretty accurate clock. Dallas/Maxim now has a series of timekeeping ICs and temperature compensated crystal oscillators. We're using the DS32KHZ TCXO in the new 7330 with good results, so perhaps the temperature at the repeater site has a lot to do with the accuracy of the controller's clock and calendar. 73, Bob Bob Schmid, WA9FBO, Member S-COM, LLC PO Box 1546 LaPorte CO 80535-1546 970-416-6505 voice 970-419-3222 fax www.scomcontrollers.com
