ETC SA4828

高精度 三 相 PWM波 形产 生器 SA鲳 28在 逆 变器 中的应 用
^i,~
●新 特 器 件 应 用
岛铕 煮王巾 PWM波 形 产 哇器 SA4828
在 迸 变器 中沩 宠用
空军工程大学电讯 工程 学院
张颖超 侯振义
Application of PWM Waveform Generator SA4828
Zhang Yingchao
Hou Zhenyi
摘要 :SA4828是 Mitel公 司生产的三相 PWM波 形产 生器 ,它 可提供 高质量 ,全 数 字的三相脉宽调
制波形 ,而 且编程简单方便 ,修 改灵活 ,文 中介 绍了 SA48⒛ 的原理及其应 周在逆变器中的软、
硬件
设计 方法。
关键词 :脉 宽调制 (PWM);控 制 ;逆 变;SA鲳 ⒛
分类号 :TM呢
1
1~。
52
文献标识码 :B
文章编号 :1~0O6-ˉ 6四7(⒛ 00)09-0003一 Os
对 SA娲 28的 控制是邋过微处理器接 口将数据
引言
送人芯片和两个寄存器 (初 始化寄存器和控制寄存
(PWM)控 制技术在逆变 电路 中得到
器 )来 实现的。初始化寄存器用于设定 与逆变器有
广泛 的应用 ,其 波形产生方法有两种 :一 是模拟方
关 的一些基 本参数 ,这 些参数在 PWM输 出端允许
输 出前初始化 ,逆 变器 工 作 以后不允许改变 。
脉宽调制
法 ,二 是数字方法 。模拟方法 电路结构复杂 ,有 温漂
现象 ,难 以实现精确控制 。数字方法则克服 了上述
MITEL公 司生产 的 SA系 列 PWM波 形产生
缺点 。
控制寄存器在工作过程 中控制输 出脉宽调制波
的状态 ,从 丽进 一 步控制逆变器 的运行状态 。通常
器具有精度高 、抗干扰能力强 、外 围电路简单等优
在 工 作时该寄存器 内容常被改写 以实现实时控制 。
点 ,其 中 ⅣⅥsz8是 主要用于变频调速 、
逆变 电源及
UPS等 工业领域 的高精度 PWM波 形产生器。
2 sA48zs的 主要特点及控制方法
参数是通过 8个 暂存器
R0、 R1、 R2、 R3、 R4、
R15来 传送 的 ”初始化参数先被写人 Rθ 、
・
…。
R1。
R5,然 后通过对 R14的 写操作将参 数送人
初始化寄存器 ”最后再将控制参数写人 R0、 R1。 …
R5、 R14、
2.1SA4828的 主要特点
R5,并 通过对 R15的 写操作将参数送 人 控制 寄存
和 Mitel公 司的先前产 品 SA828相 比 ,SA+828
主要具有 以下 特点
器 。各控制寄存器的地址如表 1所 列 。
:
●具有增强型微处理器 接 口 ,可 与更多的单片
机兼容
3
参数设置
3.1初 始化参数的设置
;
●将调制波频率 的分辨率提高到 16位
● 由于采用 了可 由用户选择 的三相幅值独立控
;
名称
;
m
载
表 1 (弘 鲥 28的 寄存器地址 、
m一 m一
制方式 ,因 而使得 三 相逆变器可用于任意不对称负
●有 三 种可供选择 的输 出波形 ,适 用于多种应
用场合
;
●可提供软件复位功能
”
“
● 内置 看 门狗 定时器 以加强 监控 ,从 雨提蠃
;
了可靠性 。
2。
2 SA娲 28芯 片的控制方法
m
01llll
0101
暂存器
传送初始化参数
传送控制参数
2000年 第 9期
《¤舛电弓竞器件》
ˉ
ˉ・
犀-
⒛00年
9月
表 2 初始化参数 空间分配表
2
7
FRm
R(l
R1
R2
R3
FRS1
RFo
×
PDTC
FIDT5
×
×
×
×
| WD15
∶
R4
WD14
MDT
R§
W“
×
×
σ
CFS1
CF∞
Hr3
PDYs
Pm
PDT1
Pm
PDYs
PDT0
PDY0
PDYz
AC
0
WD13
WDS
WD12
WIy
PDY1
WS1
WD9
WD1
PDYO
WSO
WDs
WDO
f。
当 AC=1时 ,三 个 幅值暂存器独立控制各 自的
幅值 ,该 方式适用于三 相不平衡负载。
看门狗定时器时间设置 (WD)
看门狗定时器的时间 辐d由 下式给出
彦
wd=10zTIM/r血
g。
c。
脲冲延迟时间 莎
pdy的 值 由下式给出
氵
pdy=(63一 k)/512五 arr
幅值控制 (AC)
,三 相 幅值均 由 R相 幅值暂存器
R3控 制 。而 B相 幅值暂存器 R4和 Y相 幅值暂存器
R5内 容无效 。
FRS值 对应 的十进制 自然数 0~6。
脉冲延 迟 时间(PDY)
:
:
式中 ,y∶ k是 输入的时钟频率 ,TIM是 16位 二
进制数 WD(WD15、 WD14… …WD0)对 应 的十进
式 屮 ,k的 值为与二 进制 的 PDY值 对应 的十进
制 洳然数 0~“ 。
制数 。
逋。脉 冲取 消时间(PDT)
瞧冲取消时间 莎
pd由 下式给出
如果在 矽
wd时 间内未对定时器 中的数据更新 (表
明程 序 执 行 不 正 常 ),则 定 时 器 溢 出 9系 统 关 断
:
PWM输 出。
茁
泌=(127-L)/512五 arr
3.2控 制参数 的设置
表 3 波形选择表
WDCl
各控制参数在暂存器
波形
・
R0・ …・
表 4所 列 。其参数说 明如下
a,电 源频 率 (PFS)
Sinusold纯 正 弦形
Triplen三 次谐波叠加
Deadbanded(减 少开关损耗)
R5的 空间分配如
:
电源频率 几唧er的 值 由下式给出
几⒅c Ι彡ange× PFs/65535
为用户预 留
l
PDT值 对 应 的十进
当 AC=0时
:
0
WlDz
输 出波形选择控制字如表 3所 列 。
:
戍arr=ytk/(512× 2n+1)
、
式 中 Ⅱ足k是 输 人的时钟频率 。n的 值为对应 于
Ι遘制 C「 S值 的十进制 自然数 0~7。
b。 电源频 率 范围(FRS)
电源频率范 围 彡angC的 值 由下式给 出
式 中 :m的 值为与 二 进
九 nge=2m五 aⅡ /s“
0
WD10
制 自然数 0~127。
e。
波形选择 (WS)
戴波频率 五aⅡ 的值 由下式给出
0
WD11
WDs
式 中 ,L的 值 为与 二 进 制 的
表。现将各参数作一说明。
a眇
戴波频 率(CFS)
、
VS1
m
×
表 2为 初 始 化参 数 在各个 寄存 器 中的空 间分 配
制的
0
:
表 4 控 制参数 空间分配表
R0
PFS7
PFSC
R1
PFS15
R曰
PFS14
PFSs
PFS13
PF⒊
PFS12
×
×
×
RAMlD5
RAMP+
m
R3
蛳
RAMP6
R4
BAMPT
蛳
R5
Y咖
YAMP6
BAMP5
YAMPs
PFS3
PFS11
吼
P4
E√ 廴
卜明
YAMP0
RAMPs
BAMP3
YAMPS
PFSz
PFS10
CR
RAMPz
BAMPz
Y衄
PFS1
PFSg
Ir`Ⅱ
PFη
PB`
F用
「
I
RAMP1
BAMP1
Y胛 1
RAWO
蛳
⒕ 卜惆
高精度 三相 PWM波 形产 生器 SA4828在 逆 变器 中的应用
式 中 ,凡 nge为 电源频率范 围。PFS是 16位 与 二
进制 (PFS15、 PFS14… …PFS⑴ 对应 的十进制数 的
将4828复 位
值。
b。
电源幅值 (RAMP、
YAMP、 B/矾ⅠP)
各 项 幅值 的百分 比计 算 公 式 如下
往R0~R5中 写
初始化参数
:
写虚拟寄存器 R14
以完成数据传输
∧哪呵 =A× 100%/255
式 中 ,A是 8位 幅 值 选 择 字 (AMP7、
…・
AMP0)对 应 的十进制 的值 。
AMP6・ ・
o。
往R0~R5中 写
控制参数
参数是否需要
相序 选择 (F/R)
三相
写虚拟寄存器R15
以完成数据传输
PWM输 出的相序受控于正 /反 转选择位
F/R,该 位为 0时 ,相 序为红→ 黄→ 蓝 ;反 之 ,相 序为
图 2 程序流程图
蓝→ 黄→ 红 。
输 出禁 止位 (INH)
“
该位有效时 (为 0” ),所 有 的 PWM输 出变为低
电平 ,但 不影响其它操作 。一旦设置无效 ,输 出立 即
离 电路 )后 输 出 ,在 控 制 电路 中 ,单 片机不但 用来 完
恢复 。
数据 复
的逻辑 检测 。 由于 51系 列单 片机都是地址 、
d。
e。
计 数 器复位 (CR)
“
当计数器复位位 (CR)为 0” 时 ,红 相相位计数
“
器设置为 0” 。
软件 复位 (RST)
“
当该位有效时 (为 1” ),芯 片复原为初始时的默
f。
用 总线 模 式 ,故 将
MUX、 RS引 线 连 到 高 电平 。
SETTRIP用 来 快速关 断 PWM输
TRIP端 输 出高 电平 ,指 示 灯亮 。
出 ,当 其 有效 时
。
,
4.2系 统 软件设计
软件设 计 是 整个 逆 变控 制 的核 心 ,它 决 定 着逆
变 器 的输 出特 性 。 图
认状态 ,它 的效果与硬件复位脚相 同。
看 门狗定时器选择 (WTE)
“
当该位有效时 (为 1” ),看 门狗定时器被启用
SA鲳 28的 初 始化 、输 出脉 宽 和频 率 的控 制 ,还
要处 理采样 数据 以形成 闭环控 制 ,完 成对倮 护信 号
成对
2给 出 了本 系 统 的程 序 流 程
图。
g。
从 程 序 流 程 图 中可 看 出 :单 片机 先 将 SA4828
;
,在
向其 传 送 初 始 化 参 数 和 控 制 参 数 之 后
反之 ,看 门狗定时器被禁 止 。
复位
4
工 作状 态 ,这 时单 片机应 不 断查 询输 出状 态 ,以 便 随
硬件的设计与实现
系统软、
SA4828即 可输 出 PWM波 形 ,逆 变 器 随后 将 处 于
PWM输
出特性 ,以 满 足 系统要 求 ;只 要 系统
4.1系 统硬件 连接方案
SA4828用 在逆 变器 上 的连接 电路 如 图 1所
工 作 正 常 ,看 门狗定 时器 就不 断被更新 ,以 防止其溢
示 。图中 ,AC交 流信号经整流 、逆变 、隔离 (对 于
出而 中断
UPS和 逆变器而言 ,若 用于变频调速 ,则 不需要 隔
5
时调整
PWM输
出。
结论
SA4828可 以提供高质量 、全数字化 的三 相脉
取样A/D
宽调制波形 ,并 能实现精确控制 ,以 构成性能优异 的
逆变系统 。该系统设计简单 ,控 制 电路使用器件少
,
因而可降低成本 、提高可靠性 。另外 ,芯 片提供 的
sETTRIP sA4828
M【
氵
ADo.7 ALE RD WR TR IP
SETTRIP端 在异 常情 况 下 可越 过 CPU的 控 制 而
直接关断 PWM输 出 ,因 而进 一 步提高 了系统 的可
靠性 。
收稿 日期 :2000-04-05
咨询编号 :000gO1
图1
系统 原 理 框 图
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
This Material Copyrighted By Its Respective Manufacturer
SA4828
SA4828
3-Phase Pulse Width Modulation Engine
Replaces November 1999 version, DS4204-3.2
DS4204-4.0 January 2000
The SA4828 PWM engine has been designed to provide
waveforms for the control of variable speed AC induction motors, static and uninterruptible power supplies and other forms
of electronic equipment which require precise power waveform
generation.
AD3
1
28
AD2
AD4
2
27
AD1
AD5
3
26
AD0
AD6
4
25
VDD
The six digital PWM outputs control the six power semiconductors in a three phase inverter bridge, usually via an external
isolated gate driver stage.
AD7
5
24
MUX
WR
6
23
RS
RD
7
22
ZPPR
Information contained within the PWM sequences controls
the wave shape, frequency, direction and amplitude of the output waveform and also the phase relationship between each of
the outputs. Parameters such as the carrier (switching) frequency, pulse underlap and pulse deletion times may be defined during initialisation of the device to allow the SA4828 to be
used with all power semiconductors, irrespective of switching
times. The carrier frequency may be selected up to 24kHz to
allow silent ultrasonic operation.
ALE
8
21
WSS
RESET
9
20
RPHT
CLK
10
19
SET TRIP
CS
11
18
YPHT
TRIP
12
17
BPHT
RPHB
13
16
VSS
YPHB
14
15
BPHB
The device operates as a stand-alone microprocessor peripheral and it may be controlled by virtually any microprocessor or microcontroller with little or no additional logic. Configuration pins are provided to allow the device to be adapted for most
data bus formats, including multiplexed and non-multiplexed data
busses and RD/WR or R/W control formats. Microprocessor
overhead is kept to a minimum since the device requires intervention only when a change is made to the running conditions.
Steady-state operation imposes no microprocessor overhead.
The PWM implementation is fully digital, giving predictable
accuracy and temperature stability in hostile environments which
are common to motor drive applications. An internal ROM holds
the power waveforms in compressed form. Fully digital operation also provides for operation at or around zero speed, enabling DC injection braking to be implemented.
Rotational frequency is defined to 16 bits, giving a resolution better than 0.05 rpm for a two pole motor running at 3000rpm.
SA4828
Fig.1 Pin connections (top view)
FEATURES
■
■
■
■
■
■
■
■
■
■
■
Fully Digital Operation
Configurable Microprocessor Interface
Power Frequency Range up to 4kHz
16-Bit Speed Control Resolution
Carrier Frequency Selectable up to 24kHz
Three Selectable Power Waveforms held in
Internal ROM
Selectable Minimum Pulse Width and Underlap Time
Separate Amplitude Registers for Unbalanced Load
Compensation
Deadbanding Technique to Reduce Losses in Power
Semiconductors
Watchdog Timer
Bootstrap Driver Precharge
The effective rms amplitude of each of the three phases may
be controlled individually if required. This allows compensation
of unbalanced loads which is particularly useful with static inverter and uninterruptible power supplies. In addition the independent amplitude control can be used to create two phase
output for use in single phase induction motor controllers.
APPLICATIONS
Three power waveforms are provided to cover various applications: pure sinusoid, Triplen and Deadbanded Triplen for
reduced switching losses.
ORDERING INFORMATION
The SA4828 is implemented on sub-micron CMOS technology for low power consumption. It is available with an Industrial
temperature range, and in either DIL or SOIC plastic packages.
■
■
■
■
Three Phase Induction Motor Speed Controllers
Uninterruptible Power Supplies
Static Inverter Power Supplies
Power Waveform Generators
SA4828/IG/DP1S 28 lead plastic dual-in-line package,
industrial temp range.
SA4828/IG/MP1S 28 lead plastic small-outline package,
industrial temp range.
Note:- For simplicity reference is made throughout to “motor” but comments can equally apply to power supply inverters.
1/18
SA4828
ABSOLUTE MAXIMUM RATINGS
Supply voltage, VDD
Voltage on any pin
Storage temperature
Operating temperature (industrial)
7V
VSS –0.5V to VDD +0.5V
–55°C to 150°C
–40°C to 85°C
The temperature ranges quoted apply to all package types.
Many package types are available. Further information is available
on request.
Stresses above those listed in the Absolute Maximum Ratings may cause permanent damage to the device. These are
stress ratings only and functional operation of the devices at
these conditions, or at any other condition above those indicated in the operations section of this specification, is not implied. Exposure to Absolute Maximum Ratings conditions for
extended periods may affect device reliability.
ELECTRICAL CHARACTERISTICS
All characteristics are quoted for VDD = 5V ± 10% and Tamb = 25˚C unless otherwise stated.
Characteristic
Symbol
Min.
Typ.
Max.
Units
Input high voltage
VIH
2
-
-
V
-
Input low voltage
VIL
-
-
0.8
V
-
Input low current
IIL
VIN = VSS, VDD = 5.5V
SET TRIP inputs
–1.0
-
1.0
µA
RS, MUX inputs
–25
-
–66
µA
All other inputs
-
-
10
µA
Input high current
Conditions
IIH
VIN = VDD= 5.5V
SET TRIP inputs
22
-
66
µA
RS, MUX inputs
–1.0
-
1.0
µA
All other inputs
-
-
10
µA
Output high voltage
VOH
4.0
4.5
-
V
IOH = –12mA. Phase outputs and TRIP
Output low voltage
VOL
-
0.2
0.4
V
IOH = 12mA. Phase outputs and TRIP
Output high voltage
VOH
4.0
4.5
-
V
IOH = –2mA. All other outputs
Output low voltage
VOL
-
0.2
0.4
V
IOH = 2mA. All other outputs
IDD (static)
-
-
500
µA
All outputs open circuit
IDD (dynamic)
-
(TBD)
(TBD)
µA
fCLK = 25MHz
Supply voltage
VDD
4.5
5.0
5.5
V
-
Clock frequency
fCLK
-
-
25
MHz
-
Clock duty cycle
DCLK
40
-
60
%
-
SET TRIP = 0 → outputs tripped
tTRIP
3/fCLK
-
4/fCLK
µs
fCLK = in MHz
3/fCLK
-
4/fCLK
µs
fCLK = in MHz
Supply current (static)
Supply current (dynamic)
→ TRIP = 0
Note 1: For microprocessor interface timings, see figs. 4 - 7.
2/18
SA4828
PIN DESCRIPTIONS
Pin No.
Name
Type
Pin No. Name Type
1
AD3
I
Address/Data
15
2
AD4
I
Address/Data
16
3
AD5
I
Address/Data
17
4
AD6
I
Address/Data
18
5
AD7
I
Address/Data (MSB)
19
6
WR (R/W)
I
Write Strobe (Read/Write)
7
RD (DS)
I
Read Stobe (Data Strobe)
20
RHPT
O
Red Phase (Top Switch)
8
ALE (AS)
I
Address Latch Enable (Address Strobe)
21
WSS
O
Waveform Sampling Sync.
9
RESET
I
Hardware Reset; active low
22
ZPPR
O
Zero Phase Pulse (Red Phase)
10
CLK
I
Clock Input
23
RS
I
Register Select; internal pull up
11
CS
I
Chip Select; active low
24
MUX
I
Bus Select; internal pull up
12
TRIP
O
Trip Status; low = tripped
25
VDD
P
Positive Power Supply
13
RPHB
O
Red Phase (Bottom Switch)
26
AD0
I
Address/Data (LSB)
14
YPHB
O
Yellow Phase (Bottom Switch)
27
AD1
I
Address/Data
28
AD2
I
Address/Data
Function
Function
O
Blue Phase (Bottom Switch)
VSS
P
0V Power Supply
BPHT
O
Blue Phase (Top Switch)
YPHT
O
Yellow Phase (Top Switch)
SET TRIP
I
Set Output Trip;
BPHB
active high; internal pull down
ZPPR WSS
RESET
SYSTEM
BUS
AD0-AD7
8
BUS
DEMULTIPLEXER
INITIALISATION
REGISTER
PHASING
AND
CONTROL
LOGIC
REGISTER
BANK
CONTROL
REGISTER
CS
WR
RD
PULSE
WIDTH
DELETION
PULSE
DELAY
CIRCUIT
PULSE
WIDTH
DELETION
PULSE
DELAY
CIRCUIT
PULSE
WIDTH
DELETION
PULSE
DELAY
CIRCUIT
UPPER
SWITCHING
OUTPUT
RED PHASE
LOWER
SWITCHING
OUTPUT
UPPER
SWITCHING
OUTPUT
YELLOW PHASE
LOWER
SWITCHING
OUTPUT
BUS
CONTROL
ALE
MUX
UPPER
SWITCHING
OUTPUT
BLUE PHASE
LOWER
SWITCHING
OUTPUT
RS
CLOCK
CLOCK
DIVIDER
TRIP
LATCH
ADDRESS
GENERATOR
WATCHDOG
TIMER
TRIP
SET
TRIP
WAVEFORM
SELECT
WAVEFORM
ROMS
RESET
WATCHDOG
Fig. 2 SA4828 internal block diagram
3/18
SA4828
PWM SWITCHING
INSTANTS
TRIANGLE WAVE AT
CARRIER FREQUENCY,
SAMPLING ON +VE AND –VE PEAKS
+1
0
–1
+1
POWER WAVEFORM
AS READ FROM
INTERNAL ROM
EQUIVALENT
CONTINUOUS
POWER WAVEFORM
RESULTING
PWM
WAVEFORM
0
–1
Fig.3 Asynchronous PWM generation with uniform or ‘double-edged’ regular sampling as used on the SA4828
FUNCTIONAL DESCRIPTION
An asynchronous method of PWM generation is used with
uniform or ‘double-edged’ regular sampling of the stored power
waveform, as illustrated in Fig. 3. Three standard waveforms
exist so that the device may be adapted to particular applications.
In general, a pulse width modulation signal is derived by
comparing a signal waveform (in this case the power waveform)
with a saw-tooth or triangular carrier waveform of significantly
higher frequency. The intersections between the two waveforms
in the time domain define the location of transitions in the digital
output train and hence the width of the output pulses. The width
of the pulses are directly proportional to the magnitude of the
power waveform, thus the larger the magnitude, the longer the
‘ON’ pulse.
The SA4828 uses a digital implementation of this technique
which avoids drift problems associated with the use of analog
circuitry. A triangular waveform is synthesised using an up/down
counter and a digital comparator used to compare this with the
power waveform. The power waveform is sampled regularly at
every peak and every trough of the carrier waveform allowing
both edges of the PWM output pulse to move in time, hence the
term ‘double-edged regular sampling.’ (A saw-tooth carrier waveform would result in one fixed edge and one moving edge for
each PWM pulse).
The power waveform is stored digitally in on-chip ROM (1536
samples per 360˚). The power frequency is controlled by the
rate at which the ROM is addressed - a rate which is not related
to the carrier frequency on the SA4828, hence the term ‘asynchronous method of PWM generation’. The waveform values
obtained from the ROM may also be scaled to produce variable
voltage amplitude.
Fig. 3 shows the triangular carrier waveform together with
the stepped waveform which results from sampling the output
of the ROM at the peaks and troughs of the carrier. (A continuous power waveform is also shown for reference). It can be
seen that the PWM edges of the waveform (Fig. 3) are obtained
at the points where the carrier and the sampled power waveform intersect.
The carrier (switching) frequency is selectable up to 24kHz
(assuming a 24.576MHz clock input) enabling ultrasonic switch4/18
ing in noise critical applications.
Power frequency ranges of up to 4kHz are possible with the
actual output frequency resolved to 16-bits within the chosen
range. This allows for precise motor speed control and smooth
frequency changing. Because the SA4828 implementation is
entirely digital, operation at (or around) 0Hz is possible without
stability or temperature problems. The output phase sequence
may be changed to allow both forward and reverse motor operation.
PWM output pulses can be tailored to the inverter characteristics by defining minimum allowable pulse width and the pulse
delay (underlap) time without the need for external circuitry. This
gives cost savings in terms of the power semiconductor ratings
and the size of heatsink required.
Power amplitude control is also provided via three 8-bit registers. By default one register is used to define the amplitude of
all three phases. However, by setting the Amplitude Control
(AC) bit in the Initialisation Register, the amplitude of each phase
may be controlled individually. This allows unbalanced load
compensation which is of particular use in static inverter and
uninterruptible power supplies and the generation of 2 phase
supplies for two winding single phase motors.
To ensure a robust solution, a trip input allows the PWM
outputs to be shut down immediately, overriding microprocessor control in the event of an emergency. This circumvents the
delay which a microprocessor interrupt service cycle would inevitably introduce.
A Watchdog Timer function is included. This is intended as
a safeguard to trip the device outputs in the event that communications with a controlling processor are lost.
MICROPROCESSOR INTERFACE
The SA4828 has a parallel write-only interface which is connected to the microprocessor/microcontroller by means of a
configurable data bus. This ensures compatibility with virtually
all microprocessors/ microcontrollers irrespective of bus width
and format and with little or no additional logic.
Most data bus formats may be categorised as either having
a multiplexed address/ data bus or separate address and data
buses. Similarly, most microprocessors have either a WR/RD
or a R/W structure. The SA4828 has been designed to operate
SA4828
t4
ALE
RD
Minimum bus timings:
t1
t2
t3
t4
t5
t6
t7
t8
WR
CS
AD0-7
t2
CS setup time
Address hold time
Address setup time
ALE pulse width
WR pulse width
Data setup time
Data hold time
CS hold time
10ns
10ns
10ns
30ns
60ns
20ns
10ns
0ns
t5
t1
t8
t3
t6
t7
WR mode
Fig.4 Bus timing diagram - multiplexed RD
RD/WR
t4
AS
Minimum bus timings:
DS
R/W
CS
AD0-7
t2
t7
t8
t1
t9
t3
t5
t6
t1
t2
t3
t4
t5
t6
t7
t8
t9
t10
t11
CS setup time
Address hold time
Address setup time
AS pulse width
DS pulse width
R/W pulse width
R/W setup time
CS hold time
Data setup time
Data hold time
R/W hold time
10ns
10ns
10ns
30ns
60ns
60ns
20ns
0ns
20ns
10ns
10ns
t10
t11
W mode
Fig.5 Bus timing diagram - multiplexed R/W
5/18
SA4828
Add
RS
CS
D0-7
RD
WR
Minimum bus timings:
RS/CS setup time
Data setup time
RS/CS hold time
Data hold time
WR pulse width
t1
t2
t3
t4
t5
10ns
30ns
0ns
10ns
30ns
t1
t4
t2
t3
t5
Address write cycle
Data write cycle
N.B. Timings for both cycles are the same
WR mode
Fig.6 Bus timing diagram - non-multiplexed RD
RD/WR
Add
RS
CS
D0-7
DS
R/W
Minimum bus timings:
t1
t2
t3
t4
t5
RS/CS/R/W setup time
DS pulse width
Data setup time
RS/CS/R/W hold time
Data hold time
30ns
30ns
30ns
0ns
10ns
t2
t1
t3
t5
t4
Address write cycle
N.B. Timings for both cycles are the same
W mode
Fig.7 Bus timing diagram - non-multiplexed R/W
6/18
Data write cycle
SA4828
with any combination of these four categories.
A configuration pin, MUX, is used to differentiate between
multiplexed and non-multiplexed buses. This pin is held at a
low logic level for non-multiplexed operation and is either pulled
to a high logic level or left floating for multiplexed operation (by
virtue of an internal pull-up resistor).
A further signal, named Register Select (RS) is used while
in non-multiplexed bus mode to define whether the incoming
byte is address (0) or data (1). It is pulled high via an internal
resistor so that it need not be connected in multiplexed mode. It
is ordinarily connected to an address line in non-multiplexed
mode.
In addition to these signals, the SA4828 automatically adjusts itself to the format of the bus control signals. During each
write cycle, an internal detection circuit determines whether the
RD/WR or R/W format is being used and automatically adapts
the SA4828 to suit. A chip select (CS) pin is also provided to
allow the device to share the same bus as other peripherals.
The interface operates as follows:
(i) Multiplexed mode
Connect MUX to VDD or leave open-circuit. RS is don’t care
so leave open-circuit or connect to either rail.
The Initialisation Register sets up the basic operating parameters associated with the motor and inverter. This would
normally be loaded prior to motor operation (i.e. before the PWM
outputs are activated) and is not normally updated during motor
operation.
The Control Register is used to control the PWM outputs
(and hence the motor) during operation. For example, speed,
forward/reverse, start and stop. Typically, this register will be
written to many times during motor operation in order to achieve
motor acceleration etc.
The addressing of the SA4828 registers is shown in table 1.
Comment
AD3 AD2 AD1 AD0 Register
Temporary register
R0
0
0
0
0
Temporary register
R1
1
0
0
0
Temporary register
R2
0
0
0
1
Temporary register
R3
1
0
0
1
Temporary register
R4
0
0
1
0
Temporary register
R5
1
0
1
0
Transfers initialisation data
R14
0
1
1
1
Transfers control data
R15
1
1
1
1
Table 1 SA4828 register addressing
The incoming address is latched into the address register
during each write cycle on the falling edge of ALE (provided CS
set-up and hold times are adhered to). Within the same cycle,
the data is latched into the temporary registers on the rising
edge of WR (RD/WR format) or the falling edge of DS (R/W
format). This is shown diagramatically in Figs. 4 and 5.
(ii) Non-Multiplexed mode
Connect MUX to VSS. Connect RS to any suitable address
bit (usually A0). Connect ALE to VSS.
This means that any write operation which clears A0 will write
the address into the SA4828 and any which sets it will write the
corresponding data.
To write to the device in this mode, first perform a write operation to the device such that RS=0 and the data is equal to
the required address. As WR returns high (RD/WR format) or
DS returns low (R/W format) the information is written into the
SA4828 address register. Next write the data itself with RS=1.
As WR returns high (or DS returns low) the data is written into
the temporary register dictated by the SA4828 address register.
When RS is connected to the least significant address line,
this gives an equivalent memory map of just two locations, the
upper one being ‘Data’ and the lower being ‘Address’.
Operation in this mode is shown diagramatically in Figs. 6
and 7.
Industry standard microprocessors such as the 8051, 6805,
68000 and TMS320 may be used with little or no additional circuitry. Since the microprocessor interface is fully static, bus operation may be emulated using port pins- allowing operation
with microcontrollers which have no data bus such as the PIC,
ST6 and Z86C series.
CONTROLLING THE SA4828
The SA4828 is controlled by loading data into two 48-bit
registers via the microprocessor interface. These are the Initialisation Register and the Control Register. Data is initially
loaded into a series of Temporary Registers which are then transferred simultaneously to either of the two registers by means of
a dummy write operation.
Since the width of the microprocessor interface is restricted
to 8 bits, data is initially loaded into a series of six Temporary
Registers, named R0, R1....R5. This data is then transferred
simultaneously to either the Initialisation or Control Register. New
data is only acted upon once it has been transferred.
Transfer of data takes place by performing a write operation
to a ‘dummy’ register. Writing to Temporary Register 14 transfers the data to the Initialisation Register. Similarly, writing to
Temporary Register 15 transfers the data to the Control Register. The data written to Registers 14 and 15 is irrelevant since
these are not physically implemented registers, but the write
cycle must be completed in order to transfer the data.
INITIALISATION REGISTER FUNCTIONS
3
4
5
6
7
2
1
X
CFS2 CFS1
R0 FRS2 FRS1 FRS0 X
X PDT6 PDT5 PDT4 PDT3 PDT2 PDT1
R1
X PDY5 PDY4 PDY3 PDY2 PDY1
X
R2
O
O
AC
X
X
X
WS1
R3
R4 WD15 WD14 WD13 WD12 WD11 WD10 WD9
R5 WD7 WD6 WD5 WD4 WD3 WD2 WD1
Table 2 Initialisation register allocation
0
CFS0
PDT0
PDY0
WS0
WD8
WD0
The Initialisation Register data is loaded in 8-bit bytes into
Temporary Registers R0-R5. It may be transferred to the Initialisation Register by writing ‘dummy’ data to Temporary Register R14.
Note that don’t care (X) bits should always be written to ensure code compatibility in the event of future augmentations.
Under normal operation, the contents of this register will be
defined during the power up sequence. These parameters are
particular to the inverter circuitry used and therefore changing
these parameters during motor operation is not recommended.
Modifications should be made by first inhibiting the PWM outputs using the INH bit in the Control Register, or by asserting
the hardware reset.
7/18
SA4828
Carrier Frequency (CFS)
R0
FRS2 FRS1 FRS0
The power frequency range, fRANGE, is then given by:
X
X
CFS2 CFS1 CFS0
CARRIER FREQUENCY WORD (3-BIT)
CFS2 = MSB
CFS0 = LSB
111
7
110
6
101 100
5
4
011
3
010
2
384
where fCARR = carrier frequency
Pulse Delay Time (Underlap) (PDY)
Defines the frequency of the triangular waveform to which
the power waveform is compared - see Fig. 3. Choice of carrier
frequency is largely dependent upon the power semiconductors used in the inverter. Slower devices such as bipolar transistors require a lower carrier frequency whilst fast devices such
as MOSFETs and IGBTs may be used at much higher carrier
frequencies.
High carrier frequencies are advantageous since they increase waveform resolution, but with increased switching losses
in the power semiconductors. Carrier frequencies in excess of
about 18kHz are beyond the range of human hearing which helps
to eliminate audible noise.
The carrier frequency is a function of the externally applied
clock frequency and a division ratio n, determined by the 3-bit
CFS word set during initialisation. The values of n are selected
as shown in table 3.
CFS word
Value of n
fCARR x 2m
fRANGE =
001 000
1
0
Table 3 Values of clock division ratio n
R2
X
X
PDY5 PDY4 PDY3 PDY2 PDY1 PDY0
PULSE DELAY WORD (6-BIT)
PDY5 = MSB
PDY0 = LSB
For each output phase there are two PWM control signalscontrolling the upper and lower switches in the inverter. In theory,
these two control signals are always complementary. However,
due to the finite and non-equal turn-on and turn-off times of power
semiconductors, it is necessary to provide a short time during
which both outputs are off in order to avoid a transient short
circuit through the two devices.
The length of this ‘underlap’ period is dependent upon the
technology of the power semiconductors. Again, slow devices
such as bipolar transistors require a longer underlap time whilst
faster devices require a shorter time.
The pulse delay time affects all six PWM outputs by delaying the rising edges of each of the outputs by an amount. The
pulse delay time is a function of the carrier waveform frequency
and PDY, defined by the 6-bit pulse delay time select word. The
value of pdy is selected as shown in table 5.
The carrier frequency, fCARR, is then given by:
fCARR =
PDY word
Value of PDY
fCLK
512 x 2
n+1
Power Frequency Range (FRS)
FRS2 FRS1 FRS0
X
X
101
5
100
4
011
3
010
2
001
1
Table 4 Values of clock division ratio m
8/18
tpdy =
CFS2 CFS1 CFS0
000000
0
63 - PDY
fCARR x 512
where fCARR = carrier frequency.
In order to optimise the frequency resolution of the SA4828,
the required range of power frequencies may be selected using
this parameter. Within the selected range, the frequency may
be set with 16-bit resolution. It is recommended that the next
higher power frequency range than the maximum required motor frequency is used.
The power frequency range defines the maximum limit of
the power frequency. The actual operating power frequency is
controlled by the 16-bit Power Frequency Select (PFS) word in
the Control Register but it may not exceed the value set here.
The power frequency range is a function of the carrier waveform frequency (fCARR) and a multiplication factor m, determined
by the 3-bit FRS word. The value of m is determined as shown
in Table 4.
110
6
...etc...
...etc...
The pulse delay time, tpdy, is then given by:
FREQUENCY RANGE WORD (3-BIT)
FRS2 = MSB
FRS0 = LSB
FRS word
Value of m
111110
62
Table 5 Values of PDY
where fCLK = clock input frequency.
R0
111111
63
000
0
Fig.8 shows the effect of pulse delay on a pure PWM waveform.
It should be noted that as the pulse delay circuit follows the
pulse deletion circuit (see Fig.2), the minimum pulse width seen
at the PWM outputs will be shorter than the pulse deletion time
set in the initialisation register. The actual shortest pulse generated is given by tpd – tpdy.
Pulse Deletion Time (PDT)
R1
X
PDT6 PDT5 PDT4 PDT3 PDT2 PDT1 PDT0
PULSE DELETION WORD (7-BIT)
PDT6 = MSB
PDT0 = LSB
Pure PWM pulse trains contain pulses which vary in duty
cycle from 0% to 100%. Therefore pulse widths may become
very small indeed. In practice short pulses have no useful purpose since the power semiconductors cannot fully turn on/off
within the active period of the pulse. Such pulses only increase
the power dissipation in the power devices.
SA4828
induction motors.
For three phase induction motor control a Triplen waveform
is included which provides maximum utilisation of the inverter
DC link voltage using a harmonic injection technique. Also for
motor control, a Deadbanded Triplen waveform may be selected
which in addition to providing DC link voltage boost, acts to reduce the number of switching events in the power semiconductors in order to reduce the switching losses. A symmetrical technique is used to ensure that each power semiconductor benefits to the same degree.
PWM SIGNAL
REQUIRED AT
INVERTER OUTPUT
tpdy
tpdy
OUTPUT SIGNAL TO
DRIVE TOP SWITCH
INVERTER ARM
tpdy
tpdy
R3
OUTPUT SIGNAL TO
DRIVE BOTTOM SWITCH
INVERTER ARM
X
X
AC
0
0
X
WS1 WS0
WAVEFORM SELECTION BITS
tpdy = PULSE DELAY TIME
Two bits, WS0 and WS1, in Temporary Register 3 are used
to define the power waveform, according to Table 7:
Fig.8 Effect of pulse delay on PWM pulse train
Therefore, a minimum pulse width may be defined. All pulses
shorter in duration than this are eliminated from the PWM train,
whether they are low-going or high-going.
To eliminate short pulses the true PWM train is passed
through a pulse deletion circuit. The pulse deletion circuit compares pulse widths with the pulse deletion time set in the Initialisation Register. If a pulse (either positive or negative) is greater
in duration than the pulse deletion time, it is passed through
unaltered, otherwise the pulse is deleted.
The pulse deletion time, tpd , is a function of the carrier wave
frequency and PDT, defined by the 7-bit pulse deletion time word.
The value of PDT is selected as shown in Table 6.
PDT word
Value of PDT
1111111
127
1111110
126
...etc...
...etc...
0000000
0
Table 6 Values of PDT
WS1 WS0
Waveform
0
0
Sinusoid (default)
0
1
Triplen (harmonic injection)
1
0
Deadbanded Triplen (switching loss reduction)
1
1
Reserved
Table 7 Waveform selection
The waveforms may be described by the following mathematical relationships and are shown graphically in Fig. 10:
Sinusoid:
f(t) = A sin (ωt) where A = amplitude,
ω = angular displacement
Triplen:
f(t)
f(t) = A(2. sin (ωt + 30˚) –1
f(t) = A
f(t) = A(2. sin (ωt – 30˚) –1
f(t) = A(2. sin (ωt + 30˚) +1
f(t) = –A
f(t) = A(2. sin (ωt –30˚) +1
The pulse deletion time, tpd, is then given by:
tpd =
127 - PDT
fCARR x 512
where fCARR = carrier frequency.
Valid
0˚ ≤ ωt < 60˚
60˚ ≤ ωt ≤ 120˚
120˚ ≤ ωt < 180˚
180˚ ≤ ωt < 240˚
240˚ ≤ ωt ≤ 300˚
300˚ ≤ ωt < 360˚
Fig. 9 shows the effect of pulse deletion on a pure PWM waveform.
Waveform Selection
Three waveforms are included as standard with the SA4828.
A pure sine wave is available for applications where waveform
purity is important such as static inverter power supplies,
uninterruptible power supplies and for driving single or two phase
PWM SIGNAL
BEFORE
PULSE DELETION
>tpd
tpd
>tpd
>tpd
>tpd
>tpd
>tpd
<tpd
>tpd
tpd
>tpd
<tpd
PWM SIGNAL
AFTER
PULSE DELETION
PULSE DELETED
PULSE DELETED
Fig.9 The effect of the pulse delay deletion circuit
9/18
SA4828
f(t)
Line output voltages appearing across the load are:
Vfg = f(t) – g(t)
Vgh = g(t) – h(t)
Vhf = h(t) – f(t)
SINUSOID
A
The line voltage waveforms are sinusoidal.
0°
180°
360°
Amplitude Control (AC)
R3
X
AC
0
0
X
WS1 WS0
AMPLITUDE CONTROL BIT
f(t)
TRIPLEN
A
0°
60°
120°
180°
240°
300°
360°
This is a single bit, AC in Temporary register 3, which controls the amplitude control mode for each of the three phases.
When the AC bit is cleared the red Amplitude byte in the Control
Register is used to define the amplitude of all three phases.
When the AC bit is set, the individual Amplitude bytes, Red,
Yellow and Blue in the Control Register are used to define the
amplitudes of the respective phases.
Watchdog Timer (WD)
f(t)
DEADBANDED TRIPLEN
Example with A = 0.75
0°
60°
120°
180°
240°
300°
360°
Fig.10 Waveforms implemented in SA4828
Deadbanding:
Below are the modulating functions for the Deadbanded Triplen
waveform. These have been normalised and scaled to give a
peak line (phase to phase) voltage of 2A. All the 3 phases are
shown for clarity, f(t), g(t) and h(t).
Function
f(t) = 2A. sin(ωt + 30˚) –1
g(t) = –1
h(t) = 2A. sin(ωt + 90˚) –1
Valid
0˚ < ωt ≤ 60˚
f(t) = 1
g(t) = 1 + 2A. sin (ωt – 150˚)
h(t) = 1 + 2A. sin(ωt + 150˚)
0˚ < ωt ≤ 120˚
f(t) = 2A. sin(ωt – 30˚) –1
g(t) = 2A. sin(ωt – 90˚) –1
h(t) = –1
120˚ < ωt ≤ 180˚
f(t) = 1 + 2A. sin(ωt + 30˚)
g(t) = 1
h(t) = 1 + 2A. sin(ωt + 90˚)
180˚ < ωt ≤ 240˚
f(t) = –1
g(t) = 2A. sin(ωt – 150˚) –1
h(t) = 2A. sin(ωt + 150˚) –1
240˚ < ωt ≤ 300˚
f(t) = 1 + 2A. sin(ωt – 30˚) –1
g(t) = 1 + 2A. sin(ωt – 90˚) –1
h(t) = 1
300˚ < ωt ≤ 360˚
10/18
X
R4 WD15 WD14 WD13 WD12 WD11 WD10 WD9 WD8
R5 WD7 WD6 WD5 WD4 WD3 WD2 WD1 WD0
WATCHDOG COUNTER WORD (16-BIT)
WD15 = MSB
WD0 = LSB
The Watchdog Timer consists of a 16-bit programmable
counter, which is decremented at a sub-multiple of the fCLK frequency. If the counter is allowed to reach its terminal condition
(time-out), then the PWM outputs are set to the off (low) state
and the TRIP output asserted low.
Initialisation temporary registers R4 and R5 are used to transfer the initial data for the 16-bit Watchdog programmable counter. Register R4 is the most significant byte (MSB). The programmable counter is reset to its initial value every time a write
is made to the Control register.
The time-out period twd, is given by the formula:
twd =
TIM x 1024
fCLK
Where TIM is the equivalent decimal value of the 16-bit binary number loaded to registers R4 and R5, 0 < TIM < 65535.
This gives a time-out range of 41µs < twd < 2.68s at 25MHz
master clock.
When the Watchdog times out a TRIP condition is activated,
(see SET TRIP). This state can only be cleared by application
of a Reset signal, (either hardware or software).
The Watchdog Timer function is always disabled following a
hardware or software reset. It is enabled by setting the WTE bit
in the Control Register to an active logic 1 level. If the Watchdog
function is not required the WTE bit should be set to logic 0.
SA4828
CONTROL REGISTER FUNCTIONS
R0
R1
R2
R3
R4
R5
7
6
5
4
3
2
1
0
PFS7 PFS6 PFS5 PFS4 PFS3 PFS2 PFS1 PFS0
PFS15 PFS14 PFS13 PFS12 PFS11 PFS10 PFS9 PFS8
RST
X
X
X
WTE
CR
INH
F/R
RAMP7 RAMP6 RAMP5 RAMP4 RAMP3 RAMP2 RAMP1 RAMP0
BAMP7 BAMP6 BAMP5 BAMP4 BAMP3 BAMP2 BAMP1 BAMP0
YAMP7 YAMP6 YAMP5 YAMP4 YAMP3 YAMP2 YAMP1 YAMP0
where A = integer value of 8-bit amplitude select word.
Note that the Amplitude Control (AC) bit in the Initialisation
Register determines whether the Red Amplitude register value
is used for all three phases, or if they are controlled independently.
R2
RST
X
X
X
WTE
CR
INH
F/R
FORWARD/REVERSE BIT
Table 8 Control register allocation
The Control Register data is loaded in 8-bit bytes into Temporary Registers R0-R5. It may be transferred to the Control
Register by writing ‘dummy’ data to Temporary Register R15.
Note that don’t care (X) bits should always be written to logic
0 to ensure code compatibility in the event of future product
augmentations.
Power Frequency (PFS)
R0 PFS7 PFS6 PFS5 PFS4 PFS3 PFS2 PFS1 PFS0
R1 PFS15 PFS14 PFS13 PFS12 PFS11 PFS10 PFS9 PFS8
POWER FREQUENCY WORD (16-BIT)
PFS15 = MSB
PFS0 = LSB
The power frequency is selected as a proportion of the power
frequency range (defined in the Initialisation Register) by the
16-bit power frequency select word, PFS, allowing the power
frequency to be defined in 65536 equal steps. As the PFS word
spans the two temporary registers R0 and R1 it is essential,
when changing the power frequency, that both these registers
are updated before writing to R15.
The power frequency (fPOWER) is given by:
fRANGE
fPOWER =
x PFS
65536
F/R)
Forward/Reverse (F
The phase sequence of the three-phase PWM output waveforms is controlled by the Forward/Reverse bit F/R.
The actual effect of changing this bit from 0 (forward) to 1
(reverse) is to reverse the power frequency phase counter from
incrementing the phase angle to decrementing it. The required
output waveforms are all continuous with time during a forward/
reverse change.
In the forward mode the output phase sequence is red-yellow-blue and in the reverse mode the sequence is blue-yellowred.
R2
RST
X
X
X
WTE
CR
INH
F/R
OUTPUT INHIBIT BIT
INH
Output Inhibit (INH
INH)
When active (i.e., low) the output inhibit bit INH sets all the
PWM outputs to the off (low) state. No other internal operation
of the device is affected. When the inhibit is released, the phase
bottom outputs are driven high for a whole carrier cycle before
the phase top outputs are enabled. This allows time for the capacitors in bootstrap drive circuits to be charged.
R2
RST
X
X
X
WTE
CR
INH
F/R
COUNTER RESET BIT
where PFS = decimal value of the 16-bit PFS word and fRANGE =
power frequency range set in the Initialisation Register.
CR
Counter Reset (CR
CR)
R3 RAMP7 RAMP6 RAMP5 RAMP4 RAMP3 RAMP2 RAMP1 RAMP0
R4 BAMP7 BAMP6 BAMP5 BAMP4 BAMP3 BAMP2 BAMP1 BAMP0
R5 YAMP7 YAMP6 YAMP5 YAMP4 YAMP3 YAMP2 YAMP1 YAMP0
AMPLITUDE WORDS (3x 8-BIT)
RAMP7, BAMP7, YAMP7 = MSB
RAMP0, BAMP0, YAMP0 = LSB
This facility allows the internal power frequency phase counter to be set to 0˚ (red phase) while Counter Reset (CR) is low.
Normal frequency control is suspended; the red phase outputs
have 50% duty cycle and yellow and blue phases have duty
cycles corresponding to +120˚ and –120˚ respectively.
R2
RST
X
X
X
WTE
CR
INH
F/R
WATCHDOG ENABLE BIT
Power Frequency Amplitude (RAMP, YAMP, BAMP)
This defines the effective rms amplitude of the output PWM
signals. In a motor application this allows the voltage/frequency
(flux) relationship to be defined.
The power waveform amplitude is determined by scaling the
amplitude of the waveform samples stored in the internal ROM
by the value of the 8-bit amplitude select words (RAMP, YAMP
and BAMP).
The percentage amplitude control for each
A phase is given
by:
255 x 100
Power Amplitude (%), APOWER =
Watchdog Timer Enable (WTE)
The Watchdog Timer Enable bit WTE enables the watchdog
function when set to an active high level. When WTE is set to
11/18
SA4828
a low level the Watchdog function is disabled. The 16-bit Watch-
R2
RST
X
X
X
WTE
CR
INH
F/R
SOFTWARE RESET BIT
dog counter forms part of the Initialisation Register.
isters remain unchanged.
Notes:
1. After a reset operation, in order to enable the PWM outputs,
the INH bit-1 and the CR bit-2 in Temporary Register R2
must be set to a logic “1” state, and a Write operation to
register address R15 performed.
Software Reset
The software reset bit RST provides a facility to reset the
device from the microprocessor interface instead of using the
RESET pin. When active (high), the chip is put into the same
state as that when RESET is asserted except that the RST bit is
not itself forced low. The reset condition may be cleared by writing ‘0’ to the RST bit using the microprocessor interface, or by
toggling the RESET pin.
HARDWARD INPUT/OUTPUT INSTRUCTIONS
Set Output Trip (SET TRIP input)
The SET TRIP input is provided separately from the microprocessor interface in order to allow an external source to override the microprocessor and provide a rapid shutdown facility.
For example, logic signals from overcurrent sensing circuitry
might be used to activate this input.
When the SET TRIP input is taken to a logic high, the output
trip latch is activated. This results in the TRIP output and the six
PWM outputs being latched low immediately. This condition can
only be cleared by applying a reset cycle to the RESET input or
by toggling the RST control bit.
It is essential that when not in use SET TRIP is tied low and
isolated from potential sources of noise; on no account should it
be left floating.
SET TRIP is latched internally at the master clock rate in
order to reduce noise sensitivity.
2. Following a reset sequence after power up, in order to prevent spurious operation, it is essential to write known values
to the Initialisation and Control Registers before the INH bit
is set to a logic “1” state. (See Fig.12 Programming example.)
3. Following any other reset sequence it should be noted that
PFS and the AMP registers are not cleared and it may be
desirable to write to these registers before INH and CR bits
are set to logic 1.
Zero Phase Pulse (ZPPR output)
The zero phase pulse output provides a signal at the same
frequency as the power frequency with a 1 : 1 mark-space ratio.
When in the forward mode of operation the falling edge of ZPPR
corresponds to 0° for the red phase. In the reverse mode, the
rising edge of ZPPR corresponds to 0° for the red phase.
Waveform Sampling Synchronisation (WSS output)
The WSS output is derived from the LSB of the waveform
ROM address generator and toggles once every ROM address.
There are a total of 768 rising and 768 falling edges output over
a complete power frequency cycle. Note that because the ROM
is addressed at a rate controlled by a digital rate generator, the
edges are not equispaced.
TRIP output)
Output Trip Status (TRIP
Clock (CLK input)
The TRIP output indicates the status of the output trip latch
and is active low.
The CLK input provides a timing reference used by the
SA4828 for all timings related to the PWM outputs.
The microprocessor interface, however, derives all its timings from the microprocessor and therefore the microprocessor
and the SA4828 may be run either from the same or from different clocks.
RESET input)
Reset Operation (RESET
Operation of the Hardware Reset, RESET (active low); or
Software Reset, RST (active high); performs the following functions:
1. All PWM outputs are forced low (if not already low) thereby
turning off the drive switches.
2. All internal counters, (including the Watchdog Timer), are
reset. (This corresponds to zero degrees for the red phase
output.)
3. The trailing edge of reset sets the TRIP output high - assuming that the SET TRIP input is inactive (i.e. low).
Default/Reset Conditions
The following default conditions are imposed after a reset:
1. INH = 0. PWM outputs disabled.
2. CR = 0. Internal Counter Reset held active low.
3. WTE = 0. Watchdog Timer is disabled.
4. RST = 0. Software reset is deasserted (assertion of RESET
only).
The status of all other bits in the Initialisation and Control Reg12/18
SA4828 PROGAMMING EXAMPLE
The following example assumes that a master clock of 24.576
MHz is used. This clock frequency will allow a maximum carrier
frequency of 24 kHz and a maximum power frequency of 4 kHz.
Initialisation Register Programming Example
A power waveform range of up to 250Hz is required with a
carrier frequency of 6kHz, a pulse deletion time of 10µs and an
underlap of 5µs. The Triplen waveform is required and common
amplitude control should be used across all three phases.
1. Setting the carrier frequency
The carrier frequency should be set first as the power frequency, pulse deletion time and pulse delay time are all defined
relative to the carrier frequency.
Calculate the valuef of n that will give the required carrier freCLK
quency:
512 x 2n + 1
fCARR =
⇒ 2n+1 =
fCLK
24.576 x 106
512 x fCARR
512 x 6 x 103
=
=8
SA4828
⇒n=2
From Table 3, n = 2 corresponds to a 3-bit CFS word of 010
in temporary register R0.
2. Setting the power frequency range
Calculate the value ofmm that will give the required power
fCARR x 2
frequency:
384
fRANGE =
⇒ 2m =
6 x 10
=
The Triplen waveform is selected by setting WS0 = 1 and
WS1 = 0.
Common amplitude control is required across the three
phases. This is selected by setting AC = 0. Hence Temporary
Register R3 should be set to 00000001.
This function is disabled so temporary registers R4 and R5
can be set to all zero.
3
fCARR
5. Setting waveform select, amplitude control etc.
6. Setting Watchdog Timer Counter.
250 x 384
fRANGE x 384
From Table 6, PDT = 80 corresponds to a value of PDT, the
7-bit word in temporary register R1 of 1010000.
= 16
⇒m=4
From Table 4, m = 4 corresponds to a 3-bit FRS word of 100
in temporary register R0.
3. Setting the pulse delay time
Calculate the
of PDY that will give the required pulse
(63value
- PDY)
delay time:
fCARR x 512
Control Register Programming Example
The control register would normally be updated many times
while the motor is running, but just one example is given here. It
is assumed that the Initialisation Register has already been programmed with the parameters given in the previous example.
A power waveform of 100Hz is required with a PWM waveform amplitude of 80% of that stored in the ROM across all three
phases. The phase sequence should be set to give forward motor
rotation. The outputs should be enabled.
tpdy =
1. Setting the power frequency
⇒ PDY = 63 – (tpdy x fCARR x 512)
The power frequency, fPOWER, can be selected to 16-bit accuracy (i.e 65536 equal steps) from 0Hz to fRANGE as defined in the
Initialisation Register.
In this case, with fRANGE = 250Hz, the power
fRANGE
frequency can be adjusted in increments of 0·0038Hz.
65536
⇒ 63 – ((5 x 10–6) x (6 x 103) x 512) = 47.64
fPOWER =
However, the value of PDY must be an integer. As the purpose of the pulse delay is to prevent ‘shoot-through’ (where both
top and bottom arms of the inverter are on simultaneously), it is
sensible to round the pulse delay time up to a higher, rather
than a lower figure.
Thus, assigning the value 47 to PDY gives a delay time of
5·2µs. From Table 5, PDY = 47 corresponds to a 6-bit PDY word
of 101111 in temporary register R2.
4. Setting the pulse deletion time
In setting pulse deletion time (i.e., the minimum pulse width)
account must be taken of the pulse delay time, as the actual
minimum pulse width seen at the PWM outputs is equal to
tpd –tpdy.
Therefore, the value of the pulse deletion time must, in this
instance, be set 5.2µs longer than the minimum pulse length
required.
Minimum pulse length required = 10µs
... tpd 127
to be- PDT
set to 10µs + 5.2µs = 15.2µs
Now,
fCARR x 512
tpd =
⇒ PDT = 127 – (tpd x fCARR x 512)
⇒ PFS =
PFS
fPOWER xx65536
100 x 65536
fRANGE
250
=
= 26214.4
PFS must be an integer, so assigning a value of 26214 gives
fPOWER = 99.998 Hz.The 16-bit binary equivalent of this value
8051
SA4828
AD0-7
AD0-7
VDD
RS
MUX
CS DECODE
CS
ALE
ALE
RD
WR
RD
WR
6805
SA4828
AD0-7
AD0-7
VDD
RS
MUX
⇒ 127 – ((15.2 x 10–6) x (6 x 103) x 512) = 80.30
Again, PDT must be an integer and so must be either rounded
up or down – the choice of which will depend on the application.
Choosing in this case the value 80 for PDT, gives a value of tpd
of 15.3µs.
CS DECODE
CS
AS
ALE
DS
RD
R/W
WR
Fig.11 Connection details for industry standard
multiplexed architecture
13/18
SA4828
gives a PFS word of 0110011001100110 in temporary registers
R0 and R1.
2. Setting the powerA waveform amplitude
RESET
(SOFTWARE OR
HARDWARE)
255
APOWER (%)
= x 255x 10080 x 255
APOWER
100
⇒A=
100
=
SET UP TEMP
REGS 0-5 WITH
INIT DATA
= 204
The 8-bit binary equivalent of this value gives a RAMP word
of 11001100 in temporary register R3. YAMP and BAMP contents are don’t care.
WRITE TO
TEMP REG R14
TO TRANSFER
3. Setting forward/reverse, output inhibit etc.
Forward motor control is required (i.e., the phase sequence
of the PWM outputs should be red-yellow-blue) therefore forward/reverse bit, F/R = 0.
Output inhibit should be disabled (i e., the outputs should be
active), therefore output inhibit bit, INH = 1.
Watchdog Timer is disabled so WTE = 0, RST bit = 0 and
CR bit = 1.
Hence Temporary Register R2 should be set to 00000110.
SET UP TEMP
REGS 0-5 WITH
CNTRL DATA
WRITE TO
TEMP REG R15
TO TRANSFER
APPLICATIONS INFORMATION
Microprocessor Interfacing
(i) Multiplexed data buses
SET UP TEMP
REGS 0-5 WITH
CONTRL DATA
CR = 1,INH = 1
Fig. 11 shows the connection details when interfacing industry standard microprocessors to the SA4828. The 8051 and
6805 series are given as examples.
Since the address is written into the SA4828 during each
write cycle, this mode of operation is the simplest in terms of the
required software. Fig. 12 shows a typical programming example when using multiplexed data buses.
WRITE TO
TEMP REG R15
TO TRANSFER
(ii) Non-multiplexed data buses
Fig. 13 shows the connection scheme for some industry
standard microprocessors with non-multiplexed data busses. The
6802, 68000 and TMS320 series are given as examples.
Before each successive write operation with non-multiplexed
data busses, it is essential that the address is specified using
an extra write operation. Therefore, a write to a specific Temporary Register actually consists of two write cycles - the first defines the address and the second defines the data. To differentiate between them, the Register Select (RS) pin is used. This
should be low for the first (address) cycle and high for the second (data) cycle. As shown in Fig. 13, this pin is usually tied to
the least significant microprocessor address line.
Fig. 14 shows the flow diagram for each successive write
operation to a Temporary Register.
(iii) Using microcontroller port pins to emulate a data
bus
Since the SA4828 is implemented in static logic, the bus
timings have no maximum values. This allows the use of
microcontrollers where port pins are used to emulate the function of a databus.
In order to minimise the number of port pins used, it is clearly
better to emulate a multiplexed data bus. The control lines may
be minimised as follows:
If there are no other peripherals connected to the ‘bus’ then
CS may be tied to VSS. Emulating the RD/WR control structure
14/18
N
Y
UPDATE
Fig.12 SA4828 programming example
means that the RD pin would be permanently high. It may therefore be tied to VDD. RS is not relevant to multiplexed mode, so
this pin may be tied to VDD. MUX should also be tied to VDD.
Therefore, the necessary control may be achieved using just
two signals - WR and ALE, giving ten port pins in total i.e. 8
address/data and 2 control. Fig. 15 shows this for a generic
microcontroller and Fig. 16 shows a flow diagram for a single
write cycle.
Circuit and Layout Considerations
The SET TRIP input forms a fast means of shutting down
the inverter bridge by circumventing the latency associated with
a microprocessor interrupt. It is recommended that the TRIP
output is also linked to a microprocessor interrupt pin in order
that the microprocessor is made aware of an incoming trip. It
may then perform the necessary housekeeping following the
inverter shut down.
Although the SET TRIP input has a debounce circuit built in,
SA4828
6802
SA4828
VDD
PORT 9 = LOW
PORT 10 = HIGH
AD0-7
D0-7
ALE
CS DECODE
A0
CS
E
RS
RD
R/W
WR
OUTPUT ADDRESS
ON PORTS 1-8
MUX
VSS
TOGGLE PORT 9
HIGH FOR ≥
1µs
68000
SA4828
VDD
OUTPUT DATA ON
PORTS 1-8
AD0-7
D0-7
ALE
CS DECODE
CS
A0
RS
LDS
RD
WR
R/W
TMS320
MUX
TOGGLE PORT 10
LOW FOR ≥ 1µs
VSS
SA4828
VDD
AD0-7
D0-7
RD
ALE
CS DECODE
CS
A0
RS
WE
WR
MUX
VSS
Fig.13 Connection details for industry-standard
non-multiplexed architectures
WRITE
ADDRESS
WITH RS=0
Fig.16 Flow diagram - interfacing SA4828 to
microcontroller
compromised. It is therefore important to eliminate any possible
sources of noise from the SET TRIP input. It may be advantageous to place decoupling capacitors close to the SET TRIP pin
if nuisance trips are of concern.
The clock input (CLK) has a high impedance CMOS input
and therefore presents very little load. However, when driving
this pin from an unbuffered microprocessor crystal, the tracking
capacitance may cause the oscillator to stop. It is therefore recommended that a CMOS Schmitt buffer is placed between the
crystal and the CLK input of the SA4828 which is physically
located as close to the crystal as possible.
The six PWM outputs each have a ±12mA drive capability,
enabling them to directly drive optocouplers for isolation purposes. Small pulse transformers may also be driven directly,
provided that due consideration is given to back emf generated
at turn-off. In addition, the TRIP pin has a ±12mA drive capability to enable it to drive a status LED directly.
Fig. 17 shows a typical application of the SA4828 to a three
phase variable speed motor drive.
QUICK REFERENCE GUIDE TO ENHANCEMENTS
The following list gives a brief outline of the differences between the SA4828 and the SA828 series:-
WRITE DATA
WITH RS=1
Fig.14 Flow diagram for non-multiplexed write operation
MICROCONTROLLER
SA4828
VDD
PORTS 1-8
AD0-7
RS
RD
PORT 9
PORT 10
ALE
WR
MUX
CS
VSS
Fig.15 Connection details for microcontroller interface
to SA4828
● Microprocessor interface re-designed for compatibility with
more micros.
● Register memory map changed to accommodate more
registers.
● Frequency resolution extended to 16-bits.
● Optional independent control of amplitude for each phase.
● Software reset provision.
● Device powers up with the inhibit bit (INH) active to force
PWM outputs off.
● ZPPY and ZPPB outputs deleted.
● WSS output re-defined.
● Watchdog Timer added.
● Triplen and Deadbanded Triplen waveforms added.
● All waveform options are user selectable.
● All phase bottom outputs pulse high on deassertion of INH
to charge boot-strap drive capacitors.
15/18
SA4828
INVERTER
–
RECTIFIER
AND
SMOOTHING
R
Y
3-PHASE AC
INDUCTION
MOTOR
DC LINK
B
SINGLE OR
3-PHASE
POWER
SUPPLY
3-PHASE
VARIABLE VOLTAGE,
VARIABLE FREQUENCY
WAVEFORM
+
6
ISOLATOR
TTL LEVEL
PWM
WAVEFORMS
6
FAST
SHUTDOWN
SA4828
DATA/ADDRESS BUS
(AD0-AD7)
8
MICROPROCESSOR
OR
MICROCONTROLLER
WITH ON-CHIP
ROM AND RAM
Fig.17 A typical SA4828 application
16/18
OPTIONAL
EXTERNAL
RAM
OPTIONAL
EXTERNAL
ROM
SA4828
Package Details
Dimensions are shown in : mm (in). For further package information, please contact your local representative or nearest
Customer Service Centre.
0-8°
28
10·00/10·64
7·40/7·60
(0·291/0·299) (0·394/0·419)
SPOT REF.
CHAMFER
REF.
0·25/0·71
(0·010/0·028)
¥45°
0·41/1·27
(0·016/0·050)
1
0·36/0·48
(0·014/0·019)
0·23/0·33
(0·009/0·013)
2·36/2·64
(0·093/0·104)
0·74 (0·029)
MAX.
28 LEADS AT
1·27 (0·050)
NOM. SPACING
NOTES
1. Controlling dimensions are inches.
2. This package outline diagram is for guidance
only. Please contact your Customer Service
Centre for further information.
0·10/0·30
(0·004/0·012)
17·70/18·10
(0·697/0·713)
28-LEAD MINIATURE PLASTIC DIL - MP28
1
PIN 1 REF
NOTCH
14·73 (0·58)
MAX
15.24 (0·6)
NOM CTRS
28
1·14/1·65
(0·045/0·065)
0·23/0·41
(0·009/0·016)
38·10 (1·5)
MAX
3·05 (0·120)
MIN
0·51 (0·02)
MIN
5·08/(0·200)
MAX
SEATING PLANE
0·38/0·61
(0·015/0·024)
NOTES
1. Controlling dimensions are inches.
2. This package outline diagram is for guidance
only. Please contact your Customer Service
Centre for further information.
28 LEADS AT 2·54 (0·10)
NOM. SPACING
28-LEAD PLASTIC DIL – DP28
17/18
SA4828
http://www.dynexsemi.com
e-mail: [email protected]
HEADQUARTERS OPERATIONS
DYNEX SEMICONDUCTOR LTD
Doddington Road, Lincoln.
Lincolnshire. LN6 3LF. United Kingdom.
Tel: 00-44-(0)1522-500500
Fax: 00-44-(0)1522-500550
DYNEX POWER INC.
Unit 7 - 58 Antares Drive,
Nepean, Ontario, Canada K2E 7W6.
Tel: 613.723.7035
Fax: 613.723.1518
Toll Free: 1.888.33.DYNEX (39639)
CUSTOMER SERVICE CENTRES
France, Benelux, Italy and Spain Tel: +33 (0)1 69 18 90 00. Fax: +33 (0)1 64 46 54 50
North America Tel: 011-800-5554-5554. Fax: 011-800-5444-5444
UK, Germany, Scandinavia & Rest Of World Tel: +44 (0)1522 500500. Fax: +44 (0)1522 500020
SALES OFFICES
France, Benelux, Italy and Spain Tel: +33 (0)1 69 18 90 00. Fax: +33 (0)1 64 46 54 50
Germany Tel: 07351 827723
North America Tel: (613) 723-7035. Fax: (613) 723-1518. Toll Free: 1.888.33.DYNEX (39639) /
Tel: (831) 440-1988. Fax: (831) 440-1989 / Tel: (949) 733-3005. Fax: (949) 733-2986.
UK, Germany, Scandinavia & Rest Of World Tel: +44 (0)1522 500500. Fax: +44 (0)1522 500020
These offices are supported by Representatives and Distributors in many countries world-wide.
© Dynex Semiconductor 2000 Publication No. DS4204-4 Issue No. 4.0 January 2000
TECHNICAL DOCUMENTATION – NOT FOR RESALE. PRINTED IN UNITED KINGDOM
Datasheet Annotations:
Dynex Semiconductor annotate datasheets in the top right hard corner of the front page, to indicate product status. The annotations are as follows:Target Information: This is the most tentative form of information and represents a very preliminary specification. No actual design work on the product has been started.
Preliminary Information: The product is in design and development. The datasheet represents the product as it is understood but details may change.
Advance Information: The product design is complete and final characterisation for volume production is well in hand.
No Annotation: The product parameters are fixed and the product is available to datasheet specification.
This publication is issued to provide information only which (unless agreed by the Company in writing) may not be used, applied or reproduced for any purpose nor form part of any order or contract nor to be regarded
as a representation relating to the products or services concerned. No warranty or guarantee express or implied is made regarding the capability, performance or suitability of any product or service. The Company
reserves the right to alter without prior notice the specification, design or price of any product or service. Information concerning possible methods of use is provided as a guide only and does not constitute any guarantee
that such methods of use will be satisfactory in a specific piece of equipment. It is the user's responsibility to fully determine the performance and suitability of any equipment using such information and to ensure
that any publication or data used is up to date and has not been superseded. These products are not suitable for use in any medical products whose failure to perform may result in significant injury
or death to the user. All products and materials are sold and services provided subject to the Company's conditions of sale, which are available on request.
All brand names and product names used in this publication are trademarks, registered trademarks or trade names of their respective owners.
18/18