Mitsubishi M66332FP Datasheet

MITSUBISHI 〈DIGITAL ASSP〉

DESCRIPTION

The M66332 is a facsimile image processing controller that converts analog signals that are photoelectrically converted by an image sensor into bi-level signals. It has image processing functions such as peak detection, uniformity correction, γ correction, MTF compensation, detector of background and object levels, dither control, separation of image data area, scale down, and area specification. This controller has a built-in 5-bit flash type A-D converter and interface circuits to image sensor, analog signal processing circuit, and CODEC (Coder & Decoder) to simplify control of the readout mechanism.

FEATURES

0High Speed Scan (MAX. 2 ms/line, TYP. 5 ms/line) 0A3 (8 pixels/mm) Line Sensor Attachment 0Image sensor (CCD,CIS) control signal generation

CCD: SH, CK1, CK2, RS Contact sensor (CIS): SH, CK1 (or CK2)

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

M66332FP

0Analog signal processing circuit control signal generation

CLAMP, S/H, AGC, DSCH

0Built-in 5-bit Flash Type A-D Converter 0Bi-level data external input/output interface Serial output

(→M66330) 8-bit MPU bus output with external DMA control signal

0Image data processing

γ correction Uniformity correction (block correction in units of 8 pixels) MTF compensation (1 dimension) Detector of background and object level (programmable) Dithering control

• Dither method (16 levels using 4 × 4 matrix) Separation of image data area (1 dimension) Scale down A3 → B4, A3 → A4, B4 → A4

05V Single Power Supply

APPLICATION

Facsimiles

PIN CONFIGURATION (TOP VIEW)

GND

47 48

MPU Interface

DMA Interface

Sensor Interface

RESET

GND GND DRQ

DAK

49 50 51 52 53 54 55 56

MPU

Interface

(DATA)

D7D6D5D4D3D2D1

444342414039383736353433323130

MPU

Interface

(ADDRESS)

GND

VCCA3A2A1A0DGND

M66332FP

123456789

CK2

CK1

Sensor

S/H

CLAMP

Interface

Prosessing

Analog Signal

GND

TEST

Test pin

Interface

10111213141516

SVID

TEST

SYSCK

Test pin

System clock

STIM

SCLK

CODEC

Interface

SRDY

Outline 56P6N-A

TEST

ASIG

VWL

VML3

ML2

GND

VML1

AGND

DSCH

AGC

PTIM

Interface

Processing

Analog Signal

Sensor Interface

Test pin Analog Signal Processing Interface White Basic Supply Voltage Middle Basic Supply Voltage 3 Middle Basic Supply Voltage 2

Middle Basic Supply Voltage 1 Black Basic Supply Voltage

Analog Signal Processing Interface

NC: No Connection

BLOCK DIAGRAM

MITSUBISHI 〈DIGITAL ASSP〉

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

Sensor

Interface

Analog

Signal

Interface

ASIG

System Clock ADC Analog

SYSCK

15PTIM 56SH

2CK1

Sensor Control

1CK2 3RS

4CLAMP 5S/H

Analog Control

16AGC 17DSCH

Correction memory (304 words × 5bits)

Uniformity Correction

5bit A-D converter

(flash type)

ML2

ML1

ADC Reference Voltage

ML3

To each block

Vcc

Image Processing Sequence Control Signal

Detection

of Image

Data Area

Simple Bi-level

Conversion/Background

and object Level Detection

SRAM 16

words

× 4bits

MTF compensation

Separation

of Image

Data Area

Collective

Dithering

(16 levels)

AGND

ADC

Analog GND

ADC Logic

DGND

ADC

Logic GND

Vcc

Convert

bi-level

10 35 46 55

DMA Control

MPU Bus Interface

GND

Cur out/

Scale down

525145362263019

14 SRDY 13 STIM 12 SCLK 11 SVID

54 DAK 53 DRQ

50 RESET 47 CS 48 RD 49 WR

MPU

31 A0

Interface

34 A3

(Address)

37 D0

MPU Interface

44 D7

(Data)

CODEC Interface

DMA Interface

MPU Interface

Table 1 Image Processing Functions

Image Processing Function Read Width Resolution

Read speed

Uniformity Correction

MTF Compensation

Simple Bi-level Conversion

Pseudo half-tone Separation of Image Data

Area Scale down

γ Correction

Image Sensor Control Signal

Analog Signal Processing

• A4, B4, A3

• 8 pixels/mm (primary scanning direction)

• 5ms/line Typ. 2ms/line maximum

• White correction only

• Block correction in units of 8 pixels

• 50% Correction range

• Laplacian filter circuit for 3 × 1 pixels in current line (1 dimension)

• Floating threshold method using background and object level detection circuit

• Dither method: 16 levels (4 × 4matrix)

• Detection by brightness difference in 5 × 1 pixels area in current line

• Selection method • Scale down: A3 → B4 set to 13/15; B4 → A4, 9/11; A3 → A4, 12/17

• Logarithmic correction

• Control signal generation for contact sensor (CIS) and scale down CCD

• Generate control signals for external CLAMP circuit, sample/hold circuit, and AGC circuit

Specifications

Remarks

• Operated with system clock and PRE_DATA (registers 2, 3)

• Built-in SRAM as correction memory (304 words × 5bits) (read/write allowed from MPU)

• No need for compensation memory

• Built-in SRAM as dither memory (16 words × 4bits) (read/write allowed from MPU)

• No need for processing memory

• Apply external voltage (resistor connection is also allowed) to A-D converter middle basic supply voltage pins.

• Built-in 5-bit flash A-D converter

PIN DESCRIPTIONS

Block

Sensor Interface

Analog Circuit Interface

CODEC Interface

DMA Interface

Clock MPU Interface

Pin Names

CK1

CK2 RS PTIM CLAMP S/H ASIG AGC DSCH SRDY STIM SCLK SVID DRQ

DAK SYSCK RESET

RD WR A0~A3 D0~D7

MITSUBISHI 〈DIGITAL ASSP〉

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

I/O

CCD: Shift pulse signal to transmit photo charges from the sensor to the transfer unit.

CIS: Start signal for the sensor read circuit. CCD: Clock pulse signal for sequentially transmitting the transfer unit signal charge of the

sensor. CIS: Clock pulse signal for the sensor read circuit shift register.

Reverse of CK1.

Pulse to reset the voltage of the CCD sensor floating capacitor to initial status.

Read roller pulse motor control signal.

CLAMP pulse to set the dark level of the sensor to reference voltage of the digital circuit.

Sample-hold signal to smooth out sensor image signal waveform.

Analog signals.

External AGC circuit gain down signal.

External AGC circuit gain up signal.

Data transmission ready signal from CODEC.

Data transmission bound signal for CODEC.

Clock signal for transmitting image data to CODEC.

Serial output of image data to CODEC. “H”: Black; “L”: White.

DMA request signal to external DMA controller for parallel output of image data through MPU

bus. DMA acknowledge signal from external DMA controller for the above DRQ signal.

System clock input pin.

System reset signal. Resets counter, register, F/F, and latch, sets internal memory in standby

mode, and halts clock generation circuit. Chip select signal used by MPU to access M66332. Set to “H” in operating mode (AGC, UNIF,

SCAN). Control signal used by MPU to read data from M66332.

Control signal used by MPU to write data to M66332.

Address signals used to access M66332 internal registers.

8-bit bidirectional buffer.

I/O

Description

PIN DESCRIPTIONS (CONTINUED)

Block

Others

Pin Names VCC AVCC DVCC GND AGND DGND VWL VBL VML1 VML2 VML3 TEST(IN) TEST(OUT)

I/O

–– –– –– –– –– –– –– –– –– –– –– –– ––

MITSUBISHI 〈DIGITAL ASSP〉

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

Description Plus supply voltage. Plus supply voltage for A-D converter analog units. Plus supply voltage for A-D converter logic units. GND pin. Ground for A-D converter analog units. Ground for A-D converter digital units. A-D converter white basic supply voltage pin. A-D converter black basic supply voltage pin. Middle basic supply voltage pin. VML1 =(VWL – VBL) /4 Middle basic supply voltage pin. VML2 =2 ·(VWL – VBL) /4 Middle basic supply voltage pin. VML3 =3 ·(VWL – VBL) /4 Test input pin. Fix to “L”. Test output pin. Keep open.

FUNCTIONAL DESCRIPTION

The following items which are necessary to use the image processing functions of the M66332 are described. (1)Operating mode (2)Line period and read sequence (3)Image processing function (4)Sensor unit/analog signal processing unit interface (5)CODEC interface (6)Read/write to dither memory and uniformity correction

memory (7) Reset (8)Image quality control using registers

MITSUBISHI 〈DIGITAL ASSP〉

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

(1)Operating mode

The M66332 performs three basic operations.

• Peak value detection: The peak value of the analog signal output from the analog signal processing circuits is matched to the white reference voltage (V M66332 internal A-D converter. (See also Figs. 19 to 22 in the M66333FP document.)

• Uniformity correction data creation: White reference data is created for sensor unit uniformity correction and written to the correction memory (SRAM: 304 words × 5bits).

• Read operation: A document is read and the image is processed to output bi-level data as serial or parallel output.

These three basic operations are performed in the following sequence depending on whether the sensor is CCD or CIS. The sensor is selected with register 0 (SENS).

When the sensor is CCD:

UNIF mode Operation is started by setting the

UNIF command in register 0 to “H”. If the sensor is CCD, peak detection (16 line periods) and white uniformity correction data creation (8 line periods) are performed consecutively. To exit this operating mode, wait 30 line periods (at least 24 lines) from the start and set the UNIF command to “L”.

SCAN mode The read operation is started by set-

ting the SCAN command in register 0 to “H”. Set the SCAN command to “L” to exit this operation mode.

WL) of the

When the sensor is CIS: (Creation and transmission of uniformity correction data)

UNIF mode This mode is started when the UNIF

command in register 0 is set to “H”. When the sensor is CIS, if white correction is started with the UNIF command, peak detection (16 line periods) and uniformity white correction data creation (8 line period) are started. To exit this operating mode, wait 30 line periods (at least 24 line periods) from the start and set the UNIF command to “L”.

Data transfer The uniformity correction data per-

taining to white correction created in UNIF mode are transferred to the backup memory.

(Read operation)

AGC mode Peak detection is performed for 16

line periods when the AGC command in register 0 is set to “H”. To exit this operating mode, wait 20 line periods (at least 16 lines) from the start and set the AGC command to “L”.

SCAN mode The read operation is started by set-

ting the SCAN command in register 0 to “H”. Set the SCAN command to “L” to exit this operating mode.

The signal functions and data flow in each mode are shown on pages 4–123 and 4–124. Flowcharts are shown on pages 4-158 to 4–160.

Operation During Peak Detection

MITSUBISHI 〈DIGITAL ASSP〉

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

SYSCK

Image Processing Sequence Control Signal

PTIM SH

Image sensorAnalog signal processing circuit

CK1 CK2 RS

CLAMP S/H AGC DSCH

ASIG

Sensor Control

Analog Control

Correction memory (304 words × 5bits)

Uniformity Correction

5bit A-D converter

(flash type)

VBL

VML2

VML1

VML3

To each block

Resolution compensation

VWL

Data Flow in Creation of Uniformity Correction Data

Detection

of Image

Data Area

Simple Bi-level Conversion/Background and object level detection

SRAM 16

words

× 4bits

Separation

of Image

Data Area

Collective

Dithering

(16 levels)

Convert to

bi-level

DMA Control

MPU Bus Interface

Cut out/

Scale down

STIM SCLK SVID

CODEC

SRDY DAK

DMA

DRQ RESET

CS,RD WR A0~A3

MPU

D0~D7

SYSCK

PTIM SH

Image sensorAnalog signal processing circuit

CK1 CK2 RS

CLAMP S/H AGC DSCH

ASIG

Sensor Control

Analog Control

Correction memory (304 words × 5bits)

Uniformity Correction

5bit A-D converter

(flash type)

VBL

VML2

VML1

VML3

To each block

VWL

Image Processing Sequence Control Signal

Detection

of Image

Data Area

Simple Bi-level Conversion/Background and object level detection

SRAM 16

words

MTF Compensation

× 4bits

Separation

of Image

Data Area

Collective

Dithering

(16 levels)

Convert to

bi-level

DMA Control

MPU Bus Interface

Cut out/

Scale down

STIM SCLK SVID

CODEC

SRDY DAK

DMA

DRQ RESET

CS,RD WR A0~A3

MPU

D0~D7

Date Flow During Read Operation (for serial output)

SYSCK

MITSUBISHI 〈DIGITAL ASSP〉

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

Image Processing Sequence Control Signal

PTIM SH

Image sensorAnalog signal processing circuit

CK1

Sensor Control

To each block

CK2 RS

Analog Control

Correction memory (304 words × 5bits)

CLAMP S/H AGC DSCH

Uniformity Correction

SRAM 16

MTF Compensation

5bit A-D converter

(flash type)

ML2

ML1

ML3

ASIG

Date Flow During Read Operation (for parallel output)

Detection

of Image

Data Area

Simple Bi-level

Conversion/ Background

and object

level detection

words

× 4bits

Separation

Convert

of Image

Data Area

Collective

Dithering

bi-level

MPU Bus Interface

(16 levels)

: Image date : Correction, Compensation date

Cut out/

Scale down

DMA Control

STIM SCLK SVID

CODEC

SRDY DAK

DMA

DRQ RESET

CS,RD WR A0~A3

MPU

D0~D7

SYSCK

PTIM SH

Image sensorAnalog signal processing circuit

CK1

Sensor Control

CK2 RS

Analog Control

Correction memory (304 words × 5bits)

CLAMP S/H AGC DSCH

Uniformity Correction

5bit A-D converter

(flash type)

ML2

ML1

ML3

ASIG

To each block

Image Processing Sequence Control Signal

Detection

of Image

Data Area

Separation

of Image

Data Area

Simple Bi-level

Conversion/ Background

and object

level detection

SRAM 16

words

Resolution compensation

× 4bits

Collective

Dithering

(16 levels)

: Image date : Correction, compensation date

Convert

bi-level

Cut out/

Scale down

DMA Control

MPU Bus Interface

STIM SCLK SVID

CODEC

SRDY DAK

DMA

DRQ RESET

CS,RD WR A0~A3

MPU

D0~D7

MITSUBISHI 〈DIGITAL ASSP〉

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

(2)Line period and read sequence

Figure 1 shows the relationship between the M66332 line period and the read sequence.

• 1 line period (1/ACCK): Defines the processing time per line for M66332. The line period is determined from the line period counter registers 2 and 3 (PRE_DATA) and pixel transmission clock (ADCK). ADCK is 1/ 16th of SYSCK.

1 line period (1/ACCK) [NS] = line period counter × pixel transmission clock period [NS] = (PRE_DATA + 1) × 1/ADCK [NS] = (PRE_DATA + 1) × 16/SYSCK [NS]

The line period counter is counted down with the pixel transmission clock after loading the PRE_DATA value and generates the following addresses.

• Sensor start pulse (SH): Image sensor start pulse. The position of the start pulse is determined by the value in register 4 (ST_PL) which is the offset from the uniformity correction range (UNIFG). Set ST_PL to the following values according to the type of image sensor. CCD: ST_PL = sensor dummy

pixel + 2

CIS: ST_PL =2

• Uniformity correction range (UNIFG):

• AGC range (AGCG):

• Source document read width:

• Pulse motor control signal (PTIM):

Defines the uniformity correction range. This range corresponds to the sensor width (A3 to A4). Refer to Table 2 for the relationship between sensor width and uniformity correction range. Defines the peak detection range. This range corresponds to the sensor width (A3 to A4). Auto gain control is performed for the entire width (solid line) of the sensor in AGC mode and for the range inside the sensor width (dotted line) in SCAN mode. Refer to Table 2 for the relationship between sensor width and AGC range. Defines the source document read width. If the document width is less than the sensor width, the document should be centered on the sensor because the read range is set from the center of the sensor. Refer to Table 3 for the relationship between sensor width and source document read width. Generates the pulse motor control signals for the read roller.

PRE_DATA load 0

Line period (ACCK)

Sensor start pulse (SH)

Uniformity correction range (UNIFG)

AGC range (AGCG)

Source document read width

Pulse motor control

(PTIM)

Fig. 1 Line period and read sequence

ST_PL

Down count

1 line period

Relationship with registers

Registers 2, 3 (PRE_DATA)

1 line period (ACCK)

Sensor start pulse (SH)

SRDY

(SSCAN)

(BCSCAN)

STIM

SCLK

MITSUBISHI 〈DIGITAL ASSP〉

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

SVID

PTIM

Fig. 2 CODEC Interface and read sequence

Table 2 Sensor width and gate signal range

Sensor width

Gate signal Uniformity correction

range (UNIFG)

AGC range (AGCG)

AGC mode

SCAN mode

2487/55

2370/162

2279/231

2194/306

X : Left end address Y : Right end address

2119/391

1760/740

: Output bound(SSCAN), (BCSCAN) : Internal signals

Table 3 Source document read width according to

sensor width and source document size

Sensor width

Source document size

XYX / Y

2487/55

2278/230

2118/390

––

2278/230

2118/390

––

MITSUBISHI 〈DIGITAL ASSP〉

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

(3)Image processing function

The M66332 converts image signals from the image sensor to bi-level signals. Bi-level conversion can be either simple bilevel conversion or pseudo half-tone conversion which converts image shades into bi-levels. The signal output from the image sensor must be corrected and compensated to reduce distortion and degradation before it can be converted to bi-level signals. Furthermore, for reduction in transmission time, separation of image data area and optimum bi-level conversion must be performed. The functions necessary for image processing are described below.

• Peak detection

• Uniformity correction

• MTF compensation

• Background and object level detection (simple bi-level conversion)

• Pseudo half-tone dither method

• Separation of image data area

• Image scale down/area specification

• Peak detection The A-D converter of the M66332 is used with its reference voltages (V

BL, is set to 0V to keep the dynamic range of the A-D con-

WL, VBL) fixed. Normallly, VWL is set to VCC and

verter wide. Peak detection must be performed for analog signals to match them with the full scale value of the A-D converter before they are input to the A-D converter. Peak detection is performed by reading white data in AGC mode, one of the three M66332 operating modes (AGC, UNIF, SCAN). In AGC mode, 8-line period worth of DSCH signal to raise gain—for gain control—and 16-line period worth of AGC signal to lower gain—for the overflowing of the A-D converter— are generated after AGC command start (register 0: AGC) as shown in Fig. 3. This changes the gain as shown in Fig. 4.

Peak detection

Line period

DSCH

(raise gain)

AGC

A-D converter

overflow

(lower gain)

12345678910111213141516

Peak detection

preprocessing

Peak gain control

Peak detection (preprocessing)

After completing peak detection preprocessing

White data

1 line

Match the output level of the last pixel of a line to V

Fig. 4 Changes in gain during peak detection

: Generated in pixel units only when the A-D converter overflows.

Fig. 3 Peak detection

Peak gain control

1 line

After completing gain control for peak value

Match the peak sensor output within a line toV

• Uniformity correction Uniformity correction corrects the drop in lighting level at both ends of the light source, shading distortion due to drop in lighting level at the rim of the lens, and high frequency distortion caused by the scattering of pixel-unit image sensor characteristic (see Fig. 5). The M66332 creates uniformity correction data in UNIF mode, one of the three operating modes (AGC, UNIF, SCAN), handling 8 pixels as a unit as shown in Fig. 6. The created data is written to the internal correction memory (SRAM: 304 words × 5 bits). In SCAN mode, the correction data is read from the internal correction memory to successively correct the input image data in pixel units.

MITSUBISHI 〈DIGITAL ASSP〉

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

Black level

High frequency distortion

Shading distortion

White level

1 line

Primary scanning direction

189161724

For pixels from 1 to 8, uniformity correction is carried out by correction data 1. For pixels from 9 to 16, uniformity correction is carried out by correction data 2. For pixels from 17 to 24, uniformity correction is carried out by correction data 3.

Fig. 6 Creation of uniformity correction data

Fig. 5 Image sensor white data output waveform

Pixels

Correction data 3Correction data 2Correction data 1

MITSUBISHI 〈DIGITAL ASSP〉

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

• Correction

The M66332 performs entire pixel correction for 50% correction range as shown in Fig. 7.

White Correction

Analog signal input

50%

1 line

If correction range is exceeded (during white correction)

Analog signal input

Correction is not possible if the white correction data exceeds the 50% correction range as shown in Fig. 7. Therefore, be sure to keep the input signal within the correction range.

White data

-1

Fig. 7 Uniformity correction

50%

1 line

-1

White data exceeding correction range

Portion exceeding correction range

MITSUBISHI 〈DIGITAL ASSP〉

M66332FP

FACSIMILE IMAGE DATA PROCESSOR

• MTF compensation As shown in Fig. 8, characters and photos that have been photoelectrically converted by the sensor unit are characterized by a drop in resolution. The MTF compensation per-

Photoelectric conversion

Source document (characters)

formed by the M66332 enhances the high frequency components with a Laplacian filter to maintain the resolution of the image data and creates a perception of increased dynamic range.

Photoelectric conversion

Image signal

MTF compensation

Compensated data

Photoelectric conversion

Source document (photo)

MTF compensation

Image signal

Compensated data

X’ = X + α ((X – A) + (X – B))

α = Primary scanning compensation coefficient, register 5 (MTF)

Secondary scanning direction

Primary scanning direction

Window 3 × 1

AXB

MTF compensation

Fig. 8 MTF Compensation

F–G

Pixel

+ 30 hidden pages