Maxim实时时钟芯片设计指南5413-二进制编码十进制(BCD)格式实时时钟中的状态机逻辑

网上DS12C887的文章涉及到时间的存储格式使用的都是二进制代码,究竟使用BCD码该如何操作?正好美信官网上有一篇文章。美信官网不稳定,先贴到这里,有时间再翻译。

原文链接

State Machine Logic in Binary-Coded Decimal (BCD)-Formatted Real-Time Clocks

摘要 : When developing code to operate a real-time clock (RTC), it is often beneficial to understand how the clock core operating logic has been defined. This tutorial discusses the counter-chain structure used in Maxim‘s binary-coded decimal (BCD)-formatted RTCs. It provides some insight into the expected chip behavior if improper or illegal contents are somehow installed.

A similar version of this article appears on Electronic Design, July 2012.

Introduction

When developing code to operate a real-time clock (RTC), it is often beneficial to understand how the clock core operating logic has been defined. We all have a fundamental comprehension of how a clock and calendar should operate when properly programmed, but this application note also provides some insight into the expected chip behavior if improper or illegal contents are somehow installed.

A basic real-time clock counter chain found in many of Maxim‘s binary-coded decimal (BCD)-formatted RTCs (such as the DS12885 and DS1302) is illustrated in Figure 1.

Figure 1. RTC counter chain structure.

Each counter register has defined minimum and maximum values, and most are preset to their appropriate minimums on initial power application. As a natural function of timekeeping, when the Seconds count reaches the maximum value, the next increment causes a carry to the Minutes register and the Seconds rolls over to the minimum value. A Minutes rollover carries to the Hours; the Hours rollover carries to both the Date and arbitrary Day-of-Week registers; the Date rollover carries to the Month register; the Month rollover carries to the Year register; and if applicable, the Year rollover carries to the Century register. Table 1 lists the registers and associated range of legal values.

Table 1. RTC Register Range and Carry Function

Register Minimum (hex) Maximum (hex) Carry to Register
Hundredths 00 99 Seconds
Seconds 00 59 Minutes
Minutes 00 59 Hours
Hours (12-hour mode) 41 72 AM PM, PM AM + Day & Date
Hours (24-hour mode) 00 23 Day & Date
Day 01 07
Date 01 31* Month (*Month and Year dependent)
Month 01 12 Year
Year 00 99 Century (register or bit)
Century 00 99

Some variations on the counter implementation may include a ‘Hundredths of a Second‘ register (precedes Seconds), and a Century bit to flag the Year register rollover.

Register Descriptions

For each register, the normal state machine logic is to increment the BCD count from the minimum to the defined maximum, and then roll back over to the minimum while applying the carry to the next register. For an example, the Date register BCD count sequence for a 30-day month would be 01h...09h, 10h...19h, 20h...29h, 30h, and then roll back to 01h.

Maxim does not critique the values a user might install during operation, so the ramifications of loading improper values are based upon the specific component and which register(s) was actually involved. Illogical time and date entries may result in unexpected operation.

The Hundredths of a Second register (if applicable) counts from 00h to 99h. The next increment causes the counter to roll over to 00h and the carry is applied to the Seconds register. All 8 bits in this register are read/write capable. Devices containing this register are internally clocked from the 4kHz time base (32.768kHz/8).

The Seconds register counts from 00h to 59h. The next increment causes the counter to roll over to 00h and the carry is applied to the Minutes register. Only the least significant 7 bits in this register are usually read/write capable. Most devices not containing a Hundredths of a Second register are internally clocked from the 1Hz time base.

The Minutes register also counts from 00h to 59h. The next increment causes the counter to roll over to 00h and the carry is applied to the Hours register. Only the least significant 7 bits in this register are read/write capable.

The Hours register accommodates either a 12-hour format (with an AM/PM designator) or a 24-hour clock. A control bit (usually bit6 = 12/%-overbar_pre%24%-overbar_post%) is used to identify the clock operating mode. Only the least significant 7 bits in the Hours register are read/write capable. In the 24hour Mode (bit6 = 0), the Hours register counts from 00h to 23h. The next increment causes the counter to roll-over to 00h and the carry is applied to both the Date and Day registers.

In the 12-hour mode (bit6 = 1), the Hours register count sequence, starting at "12AM", counts 52h, 41h...49h, 50h, and then 51h. The next increment causes the counter to roll over to 72h with the carry applied to the %-overbar_pre%AM%-overbar_post%/PM bit (PM = 1). The register then counts 72h, 61h...69h, 70h, 71h and the next increment causes the counter to roll over to 52h as PM is cleared, with the carry applied to both the Date and Day registers. For clarification, 12 midnight is 12:00:00 AM and 12 noon is 12:00:00 PM.

The Day (Day-of-Week) register is the most simplistic counter, with an intended content of 01h to 07h. The next increment causes the counter to roll over to 01h with no carry. Should a user accidentally deposit a 00h content into this register, there is no adverse effect upon the other timekeeping registers, but the first week of clock operation will have 8 days (0 7). The Day register only has three functioning bits, so even though it initially seems impossible for the user to load a value greater than 7, any hex value > 7 written would store the bad value logic-ORd with the three functioning bits (07h) (e.g., writing 1Fh would store a 07h, or writing F5h would store a 05h). Normal calendar convention is to define Sunday= 1, but that definition is arbitrary and fully under the user‘s control.

The Date register counts from 01h to the defined end of the present Month. The end of the month is dependent upon the present Month and Year content (in the case of February). Only the least-significant 6 bits in the Date register are read/write capable. Table 2 lists the three options.

Table 2. Last Date of Month
Calendar Month Numeric Month End Date
January, March, May, July, August, October, December 1, 3, 5, 7, 8, 10, 12 31
April, June, September, November 4, 6, 9, 11 30
February 2 28 (typical year), 29 (leap year)

The Month register counts from 01h to 12h. The next increment causes the counter to roll over to 01h and the carry is applied to the Year register. To address Y2K issues in the late 20th Century, a Century bit was included in the MSB of the Month register on components that did not include a full 8-bit Century register. This Century bit facilitated easy detection of the century change, as the bit changes state upon a Year register rollover from 99h to 00h. Care should be taken when loading new calendar values to avoid corrupting the Century bit logic condition. Please refer to the register map for the specific component in question; conventional register placement leaves bits 7 and 4:0 as read/write capable.

The Year register counts from 00h to 99h. The next increment causes the counter to roll over to 00h and the carry is applied to either the Century register or the Century bit, whichever is applicable to the component in question. All 8 bits in this register are read/write capable.

Leap Year compensation is based upon the era of the specific chip design, and applicability is normally annotated on the cover page of that data sheet. For most chip designs that do not include a Century register, the Leap Year compensation functions properly through 2099 and is assumed to be the result of this expression:

If MODULO (CALENDAR_YEAR/4) == 0

Due to the era of existing chip designs, many of today‘s RTC components containing a Century register apply the same algorithm solely upon on content of the Year register, and therefore will incorrectly designate 2100 as a leap year.

The DS1347 RTC utilizes the full Gregorian algorithm to determine Leap Year adjustments for February:

    Leap_Year = 0                            ; default to No
    If MODULO (CALENDAR_YEAR/4) == 0                ; divisible by 4?
        If MODULO (CALENDAR_YEAR/100) == 0            ; divisible by 100?
            If MODULO (CALENDAR_YEAR/400) == 0        ; divisible by 400?
                Leap_Year = 1
            end            ; if
        else
                Leap_Year = 1
        end                ; if
    end                    ; if

The Century register (if applicable) counts from 00h to 99h. All 8 bits in this register are read/write capable.

Results When Illegal Data Is Written

Up to this point, we have discussed the proper counting methods used by the RTC, given the legal range of values for the register in question. Occasionally a question may arise concerning expected counter behavior when improper values were initially installed. The answer is deterministic, and only requires a copy of the component‘s register map to arrive at a conclusion.

To access the result of loading an improper value, perform a Logic AND of the written value with the write-enabled bits in the register in question.

If the resulting content is less than the register‘s maximum counter value, the register should subsequently increment from that AND-ed value. Rollover will occur when the bit states in the register eventually coincide with the expected maximum value for that counter (12h or 00010010b as shown in Example 1).

Example 1: Write 2Ah into the Month register of a DS1337 serial RTC

From the product specification, the Month register content was verified to match the register description above.

BCD Binary
2Ah 00101010 Write to Month
9Fh 10011111 Enabled bits in Month register (with Century bit)
0Ah 00001010 Resulting content of "0Ah" after writing the Month
12h 00010010 Logic match for "maximum value"

If the resulting content is greater than the register‘s maximum counter value, the register should subsequently increment from that AND-ed value to some unpredictable rollover condition. In this case, not only was the hour value illogical, but setting bit6 also changed the functionality of the Hours counter to 12-hour Mode. Prediction of the roll-over point is illogical and results may vary, based upon that specific chip design.

Example 2: Write FFh into the Hours register of a DS1390 low-voltage RTC

From the product specification, the Hours register content was verified to match the register description above.

BCD Binary
FFh 11111111 Write to Hours
7Fh 01111111 Enabled bits in Hours register
7Fh 01111111 Resulting content of "7Fh" when reading the Hours

Example 3: Write 38h into the Hours register of a DS1341 low-current RTC

From the product specification, the Hours register content was verified to match the register description above.

Since bit6 (12/%-overbar_pre%24%-overbar_post%) was not set, the component would be running in 24-hour mode, and 38h is obviously past the normal 24-hour mode register maximum of 23h. The resulting time error is difficult to estimate, given an unstated period of operation between when the value was corrupted and when it was detected.

BCD Binary
38h 00111000 Write to Hours
7Fh 01111111 Enabled bits in Hours register
38h 00111000 Resulting content of "38h" when reading the Hours

Example 4: Write AAh into the Seconds register of a DS1302 trickle-charge timekeeping chip

From the product specification, the Seconds register content deviates from the register description above; the inclusion of the CH bit (bit7) requires some further understanding of that bit‘s functionality.

Since only 7 bits are used for the Seconds value, the counter was actually written to 2Ah. "CH" stands for Clock Halt.

With CH = 1, the oscillator is stopped and the counters do not increment, so it will never increment or roll over with CH set. The resulting time loss is difficult to estimate, given an unstated period of this idle (noncounting) condition before detection/correction.

BCD Binary
AAh 10101010 Write to Seconds
FFh 11111111 Enabled bits in Seconds register
AAh 10101010 Resulting content of "AAh" when reading the Seconds

Example 5: Write 2Dh into the Date register of a DS1338 serial RTC

From the product specification, the Date register content was verified to match the register description above.

BCD Binary
2Dh 00101101 Write to Date
3Fh 00111111 Enabled bits in Hours register
2Dh 00101101 Resulting content of "2Dh" when reading the Hours

Summary

In conclusion, we have discussed the counter chain structure utilized in Maxim‘s BCD-formatted RTCs, the derivation of unique calendar events, and Leap Year/Century handling. Additionally, using published register information and the deterministic nature of the state machine logic, some behavioral traits resulting from errant programming can be recognized during the system debugging phase. This symptom recognition can reduce a product‘s time to market.

原文地址:https://www.cnblogs.com/jqdy/p/12307731.html

时间: 2024-08-24 13:06:53

Maxim实时时钟芯片设计指南5413-二进制编码十进制(BCD)格式实时时钟中的状态机逻辑的相关文章

Maxim实时时钟芯片设计指南5791-关于编写健壮的实时时钟控制代码的提示

用DS12C887设计一个万年历,虽然反复查看说明书,还是出各种的错误. 因此,从美信官网查询资料,翻译的不太通,凑合着对照看. 原文链接 Tips for Writing Bulletproof Real-Time Clock Control Code 关于编写健壮的实时时钟控制代码的提示 摘要 : With the use of proper device initialization and code sequencing, accurate timekeeping and clock a

(3)I2C总线的字节格式,时钟同步和仲裁

字节格式 发送到SDA线上的每个字节必须是8位.每次传输的字节数量是不受限制的.每个字节后必须跟着一个ACK应答位.数据从最高有效位(MSB)开始传输.如果从机要执行一些功能后才能接收或者发送新的完整数据,比如说服务一个内部中断,那么它可以将时钟线SCL拉低来强制使主机进入wait状态.当从机准备好新的字节数据传输时,释放时钟线SCL,数据传输便继续进行. ACK和NACK 每个字节后都有ACK发生.ACK应答位允许接收器通知发送器字节成功接收了下一个字节可以发送了.主机产生所有的时钟脉冲,包括

二进制转BCD

什么是BCD码? BCD码称为二进码十进数或BCD码(Binary-Coded Decimal‎)亦称二进码十进数或二-十进制代码.用4位二进制数来表示1位十进制数中的0~9这10个数码. BCD码这种编码形式利用了四个位元来储存一个十进制的数码,使二进制和十进制之间的转换得以快捷的进行.这种编码技巧,最常用于会计系统的设计里,因为 会计制度经常需要对很长的数字串作准确的计算.相对于一般的浮点式记数法,采用BCD码,既可保存数值的精确度,又可免却使电脑作浮点运算时所耗费的时 间.此外,对于其他需

使用Socket通信实现Silverlight客户端实时数据的获取(模拟GPS数据,地图实时位置)

原文:使用Socket通信实现Silverlight客户端实时数据的获取(模拟GPS数据,地图实时位置) 在上一篇中说到了Silverlight下的Socket通信,在最后的时候说到本篇将会结合地图.下面就来看看本文实现的功能: Silverlight 与服务器利用Socket通讯,实时从服务器获取数据(本文中的数据是地理坐标),由于没有GPS,所以本文在服务器写了一个构造新坐标的函数(本文是一个三角函数),然后利用Timer组件,实时调用,得到新的坐标,并将新的坐标发送给客户端,客户端接收到发

输入的数转化为二进制序列,并统计序列中1的个数

★输入的数转化为二进制序列,并统计序列中1的个数 描述:普通的模除取余后数直接除二的办法易于理解,但是对于输入的数只限于正数和零,对于负数则不适应,所以采用与后移位的方法以此来扩大数的输入范围. #include<stdio.h> int main() { int m,b,c,i; int count = 0; char a[32]; printf("请输入一个数:\n"); scanf("%d", &m); for (i = 0; i <

ASP.NET中让图片以二进制的形式存储在数据库中

今早有个网友问到我这问题,以前我都是直接在数据库中存文件名的,还没有试过存储整张图片到数据库中,上网搜索了一下,自己又测试了一番,代码如下:建立保存图片的表的SQL语句: Sql代码   USE [niunantest] GO /****** 对象:  Table [dbo].[picdata]    脚本日期: 03/30/2010 14:51:58 ******/ SET ANSI_NULLS ON GO SET QUOTED_IDENTIFIER ON GO CREATE TABLE [d

二进制转BCD码

应用: 用fpga实现对数码管显示,以前通常的方法是进行整除和取余进行运算,但是fpga并不擅长乘法除法运算,所以可以用BCD码来转换. BCD码:通俗的可以理解为用四位二进制数表示一位十进制数字.例如,256就可以用bcd码表示为:0010_1001_0110 因此在数码管显示中,也就是把256各位分出来,就可以用bcd码来表示,下面说一种二进制转换bcd码的方法. 加3移位法: bcd码中只有0~9十进制数,但是在四位二进制中是16进制进1,因此在移位过程中要对二进制进行判断,当在移位之后的

【FPGA】【verilog代码】二进制转BCD [转]

BCD:Binary Coded Decimal 即用4位二进制编码表示1位的十进制数. 定义:BCD码这种编码形式利用了四个位元来储存一个十进制的数码,使二进制和十进制之间的转换得以快捷的进行.这种编码技巧,最常用于会计系统的设计里,因为会计制度经常需要对很长的数字串作准确的计算.相对于一般的浮点式记数法,采用BCD码,既可保存数值的精确度,又可免却使电脑作浮点运算时所耗费的时间.此外,对于其他需要高精确度的计算,BCD编码亦很常用. 参考百度百科对BCD码的解释:http://baike.b

将文件以二进制的形式保存到数据库中

主要应用HttpPostedFile类的InputStream属性.Stream类的read方法和Byte数据类型.首先获取上传数据文件的名称.大小和类型,建立一个访问客户端上传文件的对象HttpPostedFile和一个数据流对象Stream,然后使用数据流Stream对象将上传文件以二进制形式的数据写入Byte类型的数组中,最后将二进制数据保存的数据库中. 代码:try        {            if (this.FileUpload1.PostedFile.FileName