diff --git a/Makefile b/Makefile
index 8fe4940..4f69a7f 100644
--- a/Makefile
+++ b/Makefile
@@ -1,7 +1,7 @@
-PRG            = 16ch_pwm
+PRG            = rgb16mpm
 OBJ            = main.o
 MCU_TARGET     = atmega32
-OPTIMIZE       = -O2
+OPTIMIZE       = -Os
 
 DEFS           =
 LIBS           =
diff --git a/eagle/rgb16mpm-bot.png b/eagle/rgb16mpm-bot.png
new file mode 100644
index 0000000..56f3a6c
Binary files /dev/null and b/eagle/rgb16mpm-bot.png differ
diff --git a/eagle/rgb16mpm-sch.png b/eagle/rgb16mpm-sch.png
new file mode 100644
index 0000000..aa94497
Binary files /dev/null and b/eagle/rgb16mpm-sch.png differ
diff --git a/eagle/rgb16mpm-top.png b/eagle/rgb16mpm-top.png
new file mode 100644
index 0000000..ba5449d
Binary files /dev/null and b/eagle/rgb16mpm-top.png differ
diff --git a/eagle/rgb16mpm.brd b/eagle/rgb16mpm.brd
new file mode 100644
index 0000000..2b740cd
Binary files /dev/null and b/eagle/rgb16mpm.brd differ
diff --git a/eagle/rgb16mpm.sch b/eagle/rgb16mpm.sch
new file mode 100644
index 0000000..a40905a
Binary files /dev/null and b/eagle/rgb16mpm.sch differ
diff --git a/main.c b/main.c
index d920c2d..7e53a99 100644
--- a/main.c
+++ b/main.c
@@ -1,7 +1,7 @@
 /***************************************************************************
  *   16ch RGB 8bit PWM controller                                          *
  *                                                                         *
- *   Copyright (C) 2006 - 20011 by Olaf Rempel                             *
+ *   Copyright (C) 2011 - 2012 by Olaf Rempel                              *
  *   razzor AT kopf MINUS tisch DOT de                                     *
  *                                                                         *
  *   This program is free software; you can redistribute it and/or modify  *
@@ -20,6 +20,7 @@
  ***************************************************************************/
 #include <avr/io.h>
 #include <avr/interrupt.h>
+#include <avr/wdt.h>
 
 #include <stdio.h>
 
@@ -30,10 +31,10 @@
  *
  * PA0..7          -> COL1..8
  * PC0..7          -> COL9..16
- * PB0 / PD7(OC2)  -> ROW4
- * PB1 / PD5(OC1A) -> ROW3
- * PB2 / PD4(OC1B) -> ROW2
- * PB3(OC0) / PD6  -> ROW1
+ * PB0 / PD7(OC2)  -> ROW4 (OC2 not used)
+ * PB1 / PD5(OC1A) -> ROW3 (OC1A not used)
+ * PB2 / PD4(OC1B) -> ROW2 (OC1B not used)
+ * PB3(OC0) / PD6  -> ROW1 (OC0 not used)
  * PD0             -> RXD
  * PD1             -> TXD
  * PD2             -> /RX_TX
@@ -41,7 +42,6 @@
  */
 
 #define F_CPU       8000000
-
 #include <util/delay.h>
 
 #define ROW1        PORTB1      /* RED */
@@ -49,43 +49,47 @@
 #define ROW3        PORTB3      /* BLUE */
 #define ROW4        PORTB2      /* not used */
 
-#define RXTX        PORTD2
+#define RXTX        PORTD2      /* RS485 TX enable */
 #define LED         PORTD3
 
-//#define BAUDRATE    115200
+/* running without mpmboot? */
+#define STANDALONE  0
+
+#if (STANDALONE)
+#define OSCCAL      0xAA
+#define BAUDRATE    115200
+#define MPM_ADDRESS 0x11
 
-#if (BAUDRATE)
 #define UART_CALC_BAUDRATE(baudRate) (((uint32_t)F_CPU) / (((uint32_t)baudRate)*16) -1)
+#endif /* STANDALONE */
 
-static int uart_putchar(char c, FILE *stream)
-{
-    if (c == '\n') {
-        loop_until_bit_is_set(UCSRA, UDRE);
-        UDR = '\r';
-    }
+const static uint8_t versioninfo[16] = "rgb16mpm v0.99";
 
-    loop_until_bit_is_set(UCSRA, UDRE);
-    UDR = c;
-    return 0;
-}
-
-static FILE uart = FDEV_SETUP_STREAM(uart_putchar, NULL, _FDEV_SETUP_WRITE);
-#endif /* (BAUDRATE) */
+#define CMD_WAIT                0x00
+#define CMD_SWITCH_MODE         0x01
+#define CMD_GET_VERSION         0x02
+// #define CMD_GET_CHIPINFO        0x03
+// #define CMD_READ_MEMORY         0x11
+// #define CMD_WRITE_MEMORY        0x12
+#define CMD_WRITE_COLOR         0x81
+#define CMD_WRITE_RAW_COLOR     0x82
+#define CMD_READ_RAW_COLOR      0x83
+#define CMD_WRITE_CONFIG        0x81
 
 /* 16 values per color */
 static uint8_t chan_value[3][16];
 
-/* 16 +1 * 4 values (portA, portC, OCR0, flags) per color */
+/* (16 +1) * 4 values (portA, portC, OCR0, flags) per color */
 static uint8_t chan_rawdata[3][17 * 4];
 
-/* used only in ISR */
+/* used only in softpwm ISRs */
 register uint8_t * pCurrentStep asm("r2"); /* r3:r2 */
 
-/* used to sync ISR and color_update() */
+/* used to sync softpwm ISRs and color_update() */
 register uint8_t nextColor asm("r4");      /* r4 */
 
 /* worst case: 9+2+12+8+7+13 = 61 clks -> 7.625us @8MHz */
-void __attribute__ ((naked)) SIG_OVERFLOW0 (void)
+ISR(TIMER0_OVF_vect, ISR_NAKED)
 {
     asm volatile(
         /* save registers 2+1+2+2+2 = 9 */
@@ -106,30 +110,29 @@ void __attribute__ ((naked)) SIG_OVERFLOW0 (void)
     );
 
     asm volatile(
-        /* switch color and assign pCurrentStep */
+        /* switch color and assign pCurrentStep */                     //  R   G   B
         "mov r24, %0       \n\t"                                       //  1   1   1
         "inc %0            \n\t" /* nextColor++ */                     //  2   2   2
         "cpi r24, 1        \n\t"                                       //  3   3   3
         "brlo L_red%=      \n\t" /* if (nextColor < 1) -> red */       //  5   4   4
         "breq L_green%=    \n\t" /* if (nextColor == 1) -> green */    //  -   6   5
         "clr %0            \n\t" /* nextColor    = 0;           */     //  -   -   6
-        "ldi r24, %8       \n\t" /* PORTB        = (1<<ROW4);   */     //  -   -   7
-        "ldi r30, lo8(%4)  \n\t" /* pCurrentStep = &rawdata[2]; */     //  -   -   8
-        "ldi r31, hi8(%4)  \n\t"                                       //  -   -   9
+        "ldi r24, %7       \n\t" /* PORTB        = (1<<ROW4);   */     //  -   -   7
+        "ldi r30, lo8(%3)  \n\t" /* pCurrentStep = &rawdata[2]; */     //  -   -   8
+        "ldi r31, hi8(%3)  \n\t"                                       //  -   -   9
         "rjmp L_end%=      \n\t"                                       //  -   -  11
         "L_red%=:          \n\t" /* red:                        */
-        "ldi r24, %6       \n\t" /* PORTB        = (1<<ROW1);   */     //  6   -   -
-        "ldi r30, lo8(%2)  \n\t" /* pCurrentStep = &rawdata[0]; */     //  7   -   -
-        "ldi r31, hi8(%2)  \n\t"                                       //  8   -   -
+        "ldi r24, %5       \n\t" /* PORTB        = (1<<ROW1);   */     //  6   -   -
+        "ldi r30, lo8(%1)  \n\t" /* pCurrentStep = &rawdata[0]; */     //  7   -   -
+        "ldi r31, hi8(%1)  \n\t"                                       //  8   -   -
         "rjmp L_end%=      \n\t"                                       // 10   -   -
         "L_green%=:        \n\t" /* green:                      */
-        "ldi r24, %7       \n\t" /* PORTB        = (1<<ROW2);   */     //  -   7   -
-        "ldi r30, lo8(%3)  \n\t" /* pCurrentStep = &rawdata[1]; */     //  -   8   -
-        "ldi r31, hi8(%3)  \n\t"                                       //  -   9   -
+        "ldi r24, %6       \n\t" /* PORTB        = (1<<ROW2);   */     //  -   7   -
+        "ldi r30, lo8(%2)  \n\t" /* pCurrentStep = &rawdata[1]; */     //  -   8   -
+        "ldi r31, hi8(%2)  \n\t"                                       //  -   9   -
         "L_end%=:          \n\t"
-        "out %5, r24       \n\t" /* set PORTB */                       // 11  10  12
+        "out %4, r24       \n\t" /* set PORTB */                       // 11  10  12
         :: "r" (nextColor),
-           "r" (pCurrentStep),
            "i" (&chan_rawdata[0]), /* RED    */
            "i" (&chan_rawdata[1]), /* GREEN  */
            "i" (&chan_rawdata[2]), /* BLUE   */
@@ -184,7 +187,7 @@ void __attribute__ ((naked)) SIG_OVERFLOW0 (void)
 }
 
 /* worst case: 9+9+5+13 = 36 clks -> 4.5us @ 8MHz */
-void __attribute__ ((naked)) SIG_OUTPUT_COMPARE0 (void)
+ISR(TIMER0_COMP_vect, ISR_NAKED)
 {
     asm volatile(
         /* save registers 2+1+2+2+2 = 9 */
@@ -237,7 +240,7 @@ void __attribute__ ((naked)) SIG_OUTPUT_COMPARE0 (void)
     );
 }
 
-
+/* calc chan_valueX => chan_rawdataX */
 static void calculate_timer_values(uint8_t *value, uint8_t *pDataStart)
 {
     uint8_t  *pData    = pDataStart +4; /* skip first entry (init) */
@@ -285,7 +288,7 @@ static void calculate_timer_values(uint8_t *value, uint8_t *pDataStart)
             *pData++        = ((chan_used >> 8) & 0xFF);    /* PORTC */
 
             /* previous step needs timervalue and enable IRQ */
-            *(pData++ -4)   = min_value -0;                 /* OCR0  */ /* FIXME: -1 ? */
+            *(pData++ -4)   = min_value;                    /* OCR0  */
             *(pData++ -4)   = 0x01;                         /* flags */
         }
 
@@ -311,55 +314,149 @@ static void calculate_timer_values(uint8_t *value, uint8_t *pDataStart)
     *pData++ = ((chan_init >> 8) & 0xFF);  /* PORTC */
 }
 
-#if (BAUDRATE)
-void print_values(uint8_t *data)
+static uint8_t rx_cmd;
+static uint16_t rx_bcnt = 0xFF;
+static uint16_t rx_length;
+
+static uint8_t tx_cmd;
+static uint8_t tx_cause;
+static uint16_t tx_length;
+static uint16_t tx_bcnt;
+
+ISR(USART_RXC_vect)
 {
-    uint8_t i;
+    uint8_t data = UDR;
+    sei();
 
-    for (i = 0; i < 17; i++) {
-        fprintf(&uart, "%2d: %02X %02X %02X %02X\n", i,
-                data[(i<<2)], data[(i<<2) +1],
-                data[(i<<2) +2], data[(i<<2) +3]);
+    if (rx_bcnt == 0xFF) {
+        /* MPM address stored in TWI address register by bootloader */
+        if (data == TWAR) {
+            UCSRA &= ~(1<<MPCM);
+            rx_bcnt = 0;
+        }
 
-        if (data[(i<<2) +3] == 0x00)
-            break;
+    } else {
+        if (rx_bcnt == 0) {
+            rx_cmd = data;
+
+        } else if ((rx_bcnt == 1) || (rx_bcnt == 2)) {
+            rx_length = (rx_length << 8) | data;
+
+        } else if ((rx_bcnt -3) < rx_length) {
+            // TODO: get data
+        }
+
+        if ((rx_bcnt -2) == rx_length) {
+            /* enable RS485 TX */
+            PORTD |= (1<<RXTX);
+
+            /* first byte */
+            tx_cmd = rx_cmd;
+            UDR    = rx_cmd;
+
+            /* prepare header */
+            tx_bcnt   = 1;
+            tx_cause  = 0;
+            tx_length = 0;
+
+            if (tx_cmd == CMD_GET_VERSION) {
+                tx_length = sizeof(versioninfo);
+            }
+
+            /* enable interrupt */
+            UCSRB |= (1<<UDRIE);
+        }
+        rx_bcnt++;
     }
 }
-#else
-#define print_values(data)
-#endif /* (BAUDRATE) */
 
+ISR(USART_UDRE_vect)
+{
+    /* enable IRQs again, but prevent multiple UDRE IRQs */
+    UCSRB &= ~(1<<UDRIE);
+    sei();
+
+    if ((tx_bcnt < 4) || (tx_bcnt -4) < tx_length) {
+        uint16_t pos = (tx_bcnt -4);
+        uint8_t data = 0xFF;
+
+        if (tx_bcnt == 1) {
+            data = tx_cause;
+
+        } else if (tx_bcnt == 2) {
+            data = (tx_length >> 8);
+
+        } else if (tx_bcnt == 3) {
+            data = (tx_length & 0xFF);
+
+        } else if (tx_cmd == CMD_GET_VERSION) {
+            data = versioninfo[pos];
+
+        } else {
+            data = 0xAA;
+        }
+
+        UDR = data;
+
+        /* re-enable for next round */
+        UCSRB |= (1<<UDRIE);
+    }
+
+    tx_bcnt++;
+}
+
+ISR(USART_TXC_vect, ISR_NOBLOCK)
+{
+    /* disable RS485 TX */
+    PORTD &= ~(1<<RXTX);
+
+    /* enable MP mode again */
+    UCSRA |= (1<<MPCM);
+    rx_bcnt = 0xFF;     // FIXME: cli?
+
+    if (tx_cmd == CMD_SWITCH_MODE) { // FIXME: check mode
+        wdt_enable(WDTO_15MS);
+    }
+}
+
+int main(void) __attribute__ ((noreturn));
 int main(void)
 {
     /* 16 PWM Outputs */
     PORTA = 0x00;
     DDRA  = 0xFF;
 
-    PORTB = 0x00;
-    DDRB  = (1<<ROW1) | (1<<ROW2) | (1<<ROW3) | (1<<ROW4);
-
     PORTC = 0x00;
     DDRC  = 0xFF;
 
-    PORTD = (1<<RXTX) | (1<<LED);
+    /* color ROWs */
+    PORTB = 0x00;
+    DDRB  = (1<<ROW1) | (1<<ROW2) | (1<<ROW3) | (1<<ROW4);
+
+    PORTD = (1<<LED);
     DDRD  = (1<<RXTX) | (1<<LED);
 
+#if (STANDALONE)
+    OSCCAL = OSCCAL_VALUE;
+#endif /* (STANDALONE) */
+
     /* timer0, FCPU/64, overflow interrupt */
     TCCR0 = (1<<CS01) | (1<<CS00); /* FCPU/64 */
     TIMSK = (1<<TOIE0);
     TCNT0 = 0x00;
 
-#if (BAUDRATE)
-    /* Set baud rate */
+    /* USART config */
+    /* Multi Drop Mode, 9n1 */
+    UCSRA = (1<<MPCM);
+    UCSRB = (1<<RXEN) | (1<<TXEN) | (1<<RXCIE) | (1<<TXCIE) | (1<<UCSZ2);
+    UCSRC = (1<<URSEL) | (1<<UCSZ1) | (1<<UCSZ0);
+#if (STANDALONE)
     UBRRH = (UART_CALC_BAUDRATE(BAUDRATE)>>8) & 0xFF;
     UBRRL = (UART_CALC_BAUDRATE(BAUDRATE) & 0xFF);
 
-    /* enable usart with 8n1 */
-    UCSRB = (1<<TXEN) | (1<<RXEN);
-    UCSRC = (1<<URSEL) | (1<<UCSZ1) | (1<<UCSZ0);
-
-    fprintf(&uart, "good morning dave\n");
-#endif /* (BAUDRATE) */
+    /* MPM address stored in TWI address register */
+    TWAR  = MPM_ADDRESS;
+#endif /* (STANDALONE) */
 
     sei();
 
@@ -368,29 +465,37 @@ int main(void)
     uint16_t ramp = 0;
     uint8_t step = 0;
 
+#if 0
+    /* create worst case for R+G, relaxed for B */
+    for (x = 0; x < 16; x++) {
+        chan_value[0][x] = 254 - x;
+        chan_value[1][x] = x +1;
+        chan_value[2][x] = x * 16;
+    }
+    x = 0;
+#endif
+
     while (1) {
         uint8_t color_update = 0x07;
         while (color_update) {
             if ((color_update & 0x01) && (nextColor == 2)) {
                 calculate_timer_values(chan_value[0], chan_rawdata[0]);
                 color_update &= ~(0x01);
-                print_values(chan_rawdata[0]);
 
             } else if ((color_update & 0x02) && (nextColor == 0))  {
                 calculate_timer_values(chan_value[1], chan_rawdata[1]);
                 color_update &= ~(0x02);
-                print_values(chan_rawdata[1]);
 
             } else if ((color_update & 0x04) && (nextColor == 1))  {
                 calculate_timer_values(chan_value[2], chan_rawdata[2]);
                 color_update &= ~(0x04);
-                print_values(chan_rawdata[2]);
             }
         }
+#if 1
 
         PORTD ^= (1<<LED);
 
-        _delay_ms(100);
+//         _delay_ms(100);
 
         step++;
         if (step == 16) {
@@ -473,7 +578,6 @@ int main(void)
 #endif
             }
         }
+#endif
     }
-
-    return 0;
 }