ATmega SPI question

Probably not -- why can't you configure the SPI port as a slave to get your job done?

--
Tim Wescott
Wescott Design Services

Le Mon, 14 Mar 2005 09:07:59 -0800, Tim Wescott a écrit :

Hi Tim,

Configuring SPI bus in slave mode may require one more signal to drive SS. AVR received and send at the same time DOUT and DIN to and from CoDeC. Master SPI does not require SS driven, IHMO Master mode should be more simple if fortunenately SCK would be a real clock input.

Habib, betula.fr

But, by being master, the SPI interface knows exactly when a byte starts, because the transmission is started when a byte is written in the data register. In slave mode, you only know the beginning of a byte by detecting the assertion of SS. In short: what you want is impossible.

Meindert

After reviewing:

This can be done with a simple CPLD. ( did it once already, albet 2 years ago )

Make the AVR an SPI master, with the 7705 a slave. The CPLD will translate the RSYNC into a "start". Count 24 BCLK pulses into 3 SPI bytes. After the first SPI interrupt, wait for the other 2 bytes.

No Problem.

Me>

Le Mon, 14 Mar 2005 20:21:22 +0100, Meindert Sprang a écrit :

Hi Meindert,

Take a look at this :

AVR is configured as a slave. SS is grounded, therefore AVR is not able to update SPI shift reg for transmit after a byte is completely received from CodeC. That's why i'm seeking about Master SPI mode with AVR. Believe me i'm not a flasher !

Habib betula.fr

So your problem is SS stays low, you cannot access the shift register. Can't you connect SS to the same sync signal that you feed to RSYNC?

Meindert

Le Tue, 15 Mar 2005 09:42:52 +0100, Meindert Sprang a écrit :

Rrrrrriiiiiiiight ! you cannot access the shift register. Can't

I cannot do that regarding FSYNC waveform. It seems that a little more complex signals generation circuitry must be implemented to drive SS properly.

Many thanks, Habib betula.fr

Le Tue, 15 Mar 2005 10:11:06 +0100, Meindert Sprang a écrit :

SS controlled by soft, AVR as a Master ... it could work, but i will try to find out more about AVR Slave mode. In fact i'm not very inclined for the moment to cut PCB traces and so ... for implement AVR Master mode.

For sure anyway it will work this very day.

thanx, Habib betula.fr

Mmm, you would have to cut SS anyway if you want to run the AVR in slave mode, to be able to read and write to the SPI shift register.

Meindert

[...]

Hello Habib,

I use the Mega16 as a slave on an SPI line. And there is no problem to transmit more than 8 bit when /SS stays pemanently low. When the 8th bit is transmitted, MCU will receive a SPI interrupt (if this interrupt is enabled), no matter what logic state /SS has. The only problem is that you have only about half a SPI-clock cycle of time for writing the next transmit data byte into the transmitter register.

Hope that helps.

--
****----*-------*-------*-------*-------*-------*-------*-------*----****
*\ Viele     --     	Volker Bosch                                 \***

Le Tue, 15 Mar 2005 19:45:56 +0100, Volker Bosch a écrit :

Hi Volker,

I'm quite puzzled. SPI may be reloaded by AVR as Slave device regardless ~SS level ?

here a code snippet i tried :

#include #include #include #include

#include "init.h" #include "types.h" #include "uart.h" #include "maya.h" #include "pcm.h"

u08 pcm_in[256]; u08 pcm_out[256]; u08 spi_release = 0;

/************************ * SPI Interrupt handler ************************/ SIGNAL (SIG_SPI) { static u08 i = 0;

/* Read before Write SPDR !!! */ pcm_out[i] = SPDR; SPDR = pcm_in[i++]; if (i == 254) spi_release = 1; if (i == 255) i = 0; }

AVR as a slave, SPDR never been updated in that code. In fact, the same byte received by the slave (MOSI) is transmitted the next shift process into MISO pin.

Habib betula.fr

Yes. But you have to write the first data byte into the transmit register `by hand' before you enable the SPI interface. When /SS is pulled low, the data byte in the transmit register is shifted out (Beware: no SPI interrupt is triggered when /SS is pulled low!). When the last bit of the 1st byte is shifted out, the first SPI interrupt is executed by the MCU. The irpt-handler has to read the 1st received byte from the receive register and write the 2nd byte to be transmitted into the transmit register.

--
****----*-------*-------*-------*-------*-------*-------*-------*----****
*\ Viele     --     	Volker Bosch                                 \***

Le Wed, 16 Mar 2005 20:33:01 +0100, Volker Bosch a écrit :

Hi Volker !

Many many thanks for your blessed relief ! SPI comm between AVR and MSM7705 CoDec now works with AVR as a slave. You 're right for timming shift reg updating (1/2 MCU clock phase)

Thankx, Habib betula.fr