Differences

This shows you the differences between two versions of the page.

--- en:multiasm:exercisesbook:avr:sut [2026/04/29 17:52] – [Hardware reference] pczekalski
+++ en:multiasm:exercisesbook:avr:sut [2026/05/04 14:50] (current) – [Visualising Instruction Execution Time Using an Oscilloscope] pczekalski
@@ Line 11: / Line 11: @@
 <figure sutavrlabimage1>
-{{ :en:multiasm:exercisesbook:avr:sut_avr_lab_node.png?600 |}}
+{{:en:multiasm:exercisesbook:avr:sut_avr_lab_node.png?400|}}
 <caption>AVR (Arduino Uno) SUT Node</caption>
 </figure>
 <figure sutavrlabimage1_2>
-{{ :en:multiasm:exercisesbook:avr:untitled.png?600 |}}
+{{:en:multiasm:exercisesbook:avr:vrelasm_schematic.png?400|}}
 <caption>SUT node's visual interface components schematic</caption>
 </figure>
@@ Line 30: / Line 30: @@
 </table>
+==== Handling of the buffered 4-digit, 7-segment display ====
 To display a digit in the 4x7seg. display, there are two definitions needed: the shape of a digit (or other symbol), and its position (1,2,3,4: a binary mask).
@@ Line 40: / Line 42: @@
     .byte      0xC0, 0xF9, 0xA4, 0xB0, 0x99, 0x92, 0x82, 0xF8, 0x80, 0x90
 </code>
+In a common-anode configuration, the active signal to turn on a segment is LOW (0), and to turn it off, it is HIGH (1). The state of a single digit is represented by an 8-bit mask: 7 segments to build the symbol and a DP (decimal point). For example, a digit 7 is represented by bits corresponding to segments "a", "b", and "c" set to 0 (to turn segments "a", "b", and "c" on) and the remaining bits set to 1 (to turn them off), so the corresponding binary value looks as follows: 11111000b, hence the hexadecimal value is 0xF8 (as in the code above). The MSB bit represents DP, and the LSB segment "a". This definition affects how one loads data into the shift register: starting from MSB towards LSB, because of the way the register is built and connected to the segments - refer to the function ''display_digit'' below.
 <note tip>Naturally, it is possible to expand those definitions to display other symbols, e.g., hexadecimal digits such as A,b,C,d,E,F.</note>
 The way the display works is similar to a typical matrix dot display: instead of having to control 32 independent digital lines to control each LED composing the display independently (8 per digit, 4 digits), we use a digit selector (lines 0,1,2,3) and common symbol lines (lines DP,g,f,e,d,c,b,a).\\
 This way, the display "flashes" because, to display more than one digit, you need to iterate over the lines instantly and set the appropriate symbol definitions. However, the human eye is slow enough not to notice it, and thus we see all 4 digits in parallel, not being displayed one by one that in fact is a real scenario.\\
-The schematic in Figure {{ref>sutavrlabimage1}} shows an idea of how to control a single digit over a serial port pin (SER_PIN): you need to inject bit by bit, starting from the least significant bit of the symbol representing a digit, then 8 bits of the digit number - selected by lines 1,2,3,4, so only 0001b, 0010b, 0100b and 1000b combinations are used. A 0->1->0 pulse on the clock (SER_CLK) writes the data to the left registers and shifts the contents right (including passing from the left register to the right one). This way, after 16 cycles (8+8) the left register holds the line to select the digit (1,2,3,4) and the right register holds the combination representing a symbol at this postion.\\
+The schematic in Figure {{ref>sutavrlabimage1}} shows an idea of how to control a single digit over a serial port pin (SER_PIN): you need to inject bit by bit, starting from the least significant bit of the symbol representing a digit, then 8 bits of the digit number - selected by lines 1,2,3,4, so only 0001b, 0010b, 0100b and 1000b combinations are used. A 0->1->0 pulse on the clock (SER_CLK) writes the data to the left registers and shifts the contents right (including passing from the left register to the right one). This way, after 16 cycles (8+8), the left register holds the line that selects the digit (1,2,3,4), and the right register holds the combination representing the symbol at this position.\\
 When binary combinations in both registers (line and symbol) are ready to be represented, a LAT_PORT  0->1->0 pulse rewrites register counters to the internal buffer, and it instantly causes displays to light according to the symbol definition loaded into the right register (only current digit, others are off at this time).
-Note: those registers store data for only ONE digit. Iterating over digits and displaying them allows it to represent a full, multi-digit number.
-To display, e.g. 1023, it is necessary to handle each digit separately.
+** Display single digit: function definition **\\
 To handle display, a sample function that displays a digit in a selected position is presented below. Note that it does not check parameters and thus assumes that the digit position is a number between 0 and 3, and that a digit to display is 0..9. Going beyond these limits causes unpredictable behaviour and usually an MCU program crash.
 <code asm>
@@ Line 146: / Line 147: @@
 </code>
+<note important>Note: registers in this schema store data for only ONE digit. Iterating over digits and displaying them allows it to represent a full, multi-digit number. To display, e.g., 1023, it is necessary to handle each digit separately: "1", "0", "2", and "3", and to repeat this process continuously. If you stop, only the last digit will be visible.</note>
+<note tip>Changing the definitions of the symbols stored in ''segment_masks'' enables you to easily present characters other than numbers. Think about ''segment_masks'' as a font definition that defines how a symbol looks.</note>
+** Display single digit: how to use it to display a number? **\\
+Sample code that uses the function declared above and displays 1975 is presented below. Note, the MCU runs here at full speed, constantly updating the display. While it is not necessary to (a minimum, comfortable LED display refresh rate should be around 10Hz), we do not present such a solution here for the sake of simplicity. It is common to address timers for this job to periodically refresh the screen.
+<code asm>
+.equ SREG,     0x3F     ; Status Register
+.equ SPH,      0x3E     ; Stack Pointer High
+.equ SPL,      0x3D     ; Stack Pointer Low
+.equ SER_PORT, 0x05     ; PORTB I/O address
+.equ PINB,     0x03     ; Input Pins Port B (Toggle Shortcut)
+.equ SER_PIN,  0        ; GPIO8
+.equ DDRD,     0x0A     ; Data Direction Port D
+.equ DDRB,     0x04     ; Data Direction Port B
+.equ CLK_PORT, 0x0B     ; PORTD I/O address
+.equ CLK_PIN,  7        ; GPIO7
+.equ LAT_PORT, 0x0B     ; PORTD I/O address
+.equ LAT_PIN,  4        ; GPIO4
+.equ RAMEND,   0x08FF
+.global display_digit
+; ---------------------------------------------------------
+; Data stored in Program Memory (Flash)
+; ---------------------------------------------------------
+.section .text
+.org 0x0000
+    rjmp RESET
+RESET:
+    ; Prepare stack
+    ldi r16, hi8(RAMEND)
+    out SPH, r16
+    ldi r16, lo8(RAMEND)
+    out SPL, r16
+    ; Initialise display control outputs
+    sbi DDRB, SER_PIN  ; Set PB0 as output
+    sbi DDRD, CLK_PIN  ; Set PD7 as output
+    sbi DDRD, LAT_PIN  ; Set PD4 as output
+    clr r25
+    clr r23
+    ; --- Main Loop, displays in sequence 1->9->7->5 ---
+LOOP:
+    ldi r24,0
+    ldi r22,1
+    call display_digit ; Display 1
+    ldi r24,1
+    ldi r22,9
+    call display_digit ; Display 9
+    ldi r24,2
+    ldi r22,7
+    call display_digit ; Display 7
+    ldi r24,3
+    ldi r22,5
+    call display_digit ; Display 5
+    rjmp LOOP
+; void display_digit(uint8_t pos, uint8_t number);
+; r24 = position (0 to 3)
+; r22 = number (0 to 9)
+.... here comes the body of the display_digit function
+</code>
+In the function above, we used fixed (constant) digits to display. A common scenario, however, is when the number is stored in some register or in a memory variable.
+** Convert number to digits: function definition **\\
+To display a number on this kind of display, you need to convert it into an array of bytes, each representing a digit. A function below does the trick.
+<code asm>
+; void convert_to_digits(uint16_t value, uint8_t* array);
+; Inputs:
+; r25:r24 = Value to convert (up to 9999)
+; r23:r22 = Pointer to SRAM array (4 bytes long)
+convert_to_digits:
+    ; Save registers we are about to use
+    push r26
+    push r27
+    push r18
+    push r19
+    push r20
+    ; Move the SRAM pointer from r23:r22 into the X pointer (r27:r26)
+    movw r26, r22
+    ; ---------------------------------------------------
+    ; 1. Thousands Digit (Subtract 1000 = 0x03E8)
+    ; ---------------------------------------------------
+    clr r18             ; Clear digit counter
+    ldi r19, 0x03       ; High byte of 1000
+    ldi r20, 0xE8       ; Low byte of 1000
+loop_1000:
+    cp r24, r20         ; Compare value low byte with 1000 low byte
+    cpc r25, r19        ; Compare value high byte with 1000 high byte
+    brlo done_1000      ; If value < 1000, branch out
+    sub r24, r20        ; Subtract 1000 low byte
+    sbc r25, r19        ; Subtract 1000 high byte (with carry)
+    inc r18             ; Increment thousands digit
+    rjmp loop_1000
+done_1000:
+    st X+, r18          ; Store thousands digit in array[0] and increment X
+    ; ---------------------------------------------------
+    ; 2. Hundreds Digit (Subtract 100 = 0x0064)
+    ; ---------------------------------------------------
+    clr r18             ; Reset digit counter
+    ldi r19, 0x00       ; High byte of 100
+    ldi r20, 0x64       ; Low byte of 100
+loop_100:
+    cp r24, r20
+    cpc r25, r19
+    brlo done_100
+    sub r24, r20
+    sbc r25, r19
+    inc r18
+    rjmp loop_100
+done_100:
+    st X+, r18          ; Store hundreds digit in array[1] and increment X
+    ; ---------------------------------------------------
+    ; 3. Tens Digit (Subtract 10 = 0x000A)
+    ; ---------------------------------------------------
+    clr r18             ; Reset digit counter
+    ldi r19, 0x00       ; High byte of 10
+    ldi r20, 0x0A       ; Low byte of 10
+loop_10:
+    cp r24, r20
+    cpc r25, r19
+    brlo done_10
+    sub r24, r20
+    sbc r25, r19
+    inc r18
+    rjmp loop_10
+done_10:
+    st X+, r18          ; Store tens digit in array[2] and increment X
+    ; ---------------------------------------------------
+    ; 4. Ones Digit (The Remainder)
+    ; ---------------------------------------------------
+    ; Whatever is left in r24 is the ones digit (0-9)
+    st X, r24           ; Store ones digit in array[3] (no need to increment X)
+    ; Restore registers and return
+    pop r20
+    pop r19
+    pop r18
+    pop r27
+    pop r26
+    ret
+</code>
+Note, this function operates on a buffer located in the memory, which can be declared, e.g. as follows:
+<code asm>
+.section .bss  ; .bss is for uninitialized variables in SRAM
+; Reserve 4 bytes in SRAM to hold the 4 converted digits
+display_array:
+    .space 4
+</code>
 ===== Communication =====
 Devices (laboratory nodes) are interconnected in pairs, so it is possible to work in groups and implement scenarios involving more than one device:
@@ Line 184: / Line 340: @@
 <note tip>Nodes are interconnected in pairs: 1-2, 3-4, 5-6, 7-8, 9-10. Scenarios for data transmission between MCUs require booking and the use of correct nodes for sending and receiving messages.</note>
+==== Visualising Instruction Execution Time Using an Oscilloscope ====
+Let's try to visualise how code operates the GPIO. Naturally, in the remote lab, it is not possible to do it remotely, so here we present some desk-based experiments.\\
+The ''LAT_PIN'' is GPIO4, and an oscilloscope is connected to it.
+In the function that displays a single digit, there is a section that loads a binary mask into the internal registers, enabling the LED segments that constitute the digit to be turned on and off. It is:
+<code asm>
+...
+    sbi LAT_PORT, LAT_PIN
+    cbi LAT_PORT, LAT_PIN
+...
+</code>
+The figures {{ref>arduinounodigitoscilloscope1}} and {{ref>arduinounodigitoscilloscope2}} present the ''LAT_PIN'' signal, called periodically during the display of the consecutive digits (they represent the same signal but differ by the oscilloscope time base for better observation).\\
+''SBI'' causes the signal to rise, while ''CBI'' to fall. Thus, the HIGH time is the exact time during which the CBI instruction executes. It takes about 120-130ns.\\
+The Arduino Uno operates at 16 MHz, so each cycle is 1/16000000 s, which is about 63 ns. According to the documentation, ''CBI'' takes 2 cycles, which is ~126ns.
+<figure arduinounodigitoscilloscope1>
+{{:en:multiasm:exercisesbook:avr:pic_1054.png?300|}}
+<caption>''LAT_PIN'' signal (50ns time base)</caption>
+</figure>
+<figure arduinounodigitoscilloscope2>
+{{:en:multiasm:exercisesbook:avr:pic_1055.png?300|}}
+<caption>''LAT_PIN'' signal (25ns time base)</caption>
+</figure>

en/multiasm/exercisesbook/avr/sut.1777474368.txt.gz · Last modified: 2026/04/29 17:52 by pczekalski