Hello,

I have worked out the sloution.

Using the status register return values you can detect if an addressed 
device is absent.

Using a Read for 1 byte:

In IDLE the status register is 0x000d

On issuing a Start|Stop, the status reg. reads 0x000e

If No device matching the I2C address is Present, then the status goes 
back to IDLE 0x000d without any data or NACK flags being set.

If the device is present then the status register returns 0x000e 
indicating the TWI_RHR is full, when the TWI_RHR is read the status 
returns to IDLE.

I have only tested this for single byte reads. I found that when I had 
not set the Stop flag in the control reg. I recived a byte back of 0xff, 
and also had instances of the Overflow flag being set. This would 
sudgest that it won't work for the multi byte transfer mode. This would 
need further testing to prove the behaviour.

For now the single byte read responce will do what I need. If the 
request goes back to IDLE before reciving the data byte I can interpret 
this as a NACK in the address cycle.

Regards Peter
	

Lee,

I re-read the post and I can see where you were coming from.  I 
presumed he had a device that was not responding with what he epected 
it to. 

Al

--- In AT91SAM@AT91..., "budfrog99" <lee_h_theusch@h...> 
wrote:
>
> --- In AT91SAM@AT91..., "Al Welch" <alwelch@a...> wrote:
> >
> > I would think the answer to this may be in the data sheet for the 
> external
> > device you are talking to. ...
> 
> You lost me on that one, Al.  I don't know the answer to Peter's 
> question, but he is trying to determine the response from a device--
> that is not there.  I don't see how the datasheet of a non-
existent/non-
> present device has any bearing on the issue.  I
wouldn't expect to 
see 
> a notation:  "When this device has not been
purchased and is not 
> attached to your application, it will not respond to requests."
> 
> Or maybe I'm misreading the issue.
> 
> Lee
>
	

Lee,

I must have missed something in his original post as I thought he was trying
to get an external device to respond.  That being what I thought he was
doing prompted me to suggest what I did post.  I do not think you needed to
make any remarks about a non existant device responding as clearly that was
not what I  was talking about.  Maybe my post made him think about needing a
device in order to get a response if he did not have one there.

I guess I will read his post closer.

Al 

-----Original Message-----
From: AT91SAM@AT91... [mailto:AT91SAM@AT91...] On Behalf Of
budfrog99
Sent: Saturday, January 21, 2006 7:43 PM
To: AT91SAM@AT91...
Subject: [AT91SAM] Re: Detecting absence of I2C bus devices

--- In AT91SAM@AT91..., "Al Welch" <alwelch@a...> wrote:
>
> I would think the answer to this may be in the data sheet for the
external
> device you are talking to. ...

You lost me on that one, Al.  I don't know the answer to Peter's
question,
but he is trying to determine the response from a device-- that is not
there.  I don't see how the datasheet of a non-existent/non- present device
has any bearing on the issue.  I wouldn't expect to see a notation: 
"When
this device has not been purchased and is not attached to your application,
it will not respond to requests."

Or maybe I'm misreading the issue.

Lee
	Yahoo! Groups Links
	

--- In AT91SAM@AT91..., "Al Welch" <alwelch@a...> wrote:
>
> I would think the answer to this may be in the data sheet for the 
external
> device you are talking to. ...

You lost me on that one, Al.  I don't know the answer to Peter's 
question, but he is trying to determine the response from a device--
that is not there.  I don't see how the datasheet of a non-existent/non-
present device has any bearing on the issue.  I wouldn't expect to see 
a notation:  "When this device has not been purchased and is not 
attached to your application, it will not respond to requests."

Or maybe I'm misreading the issue.

Lee
	

I would think the answer to this may be in the data sheet for the external
device you are talking to. Does that device data sheet indicate it sends the
NACK?  Further I think it is a job for a logic analyser where you can
capture the bit pattern going and coming.  A picture is worth a thousand
words.  You can buy low cost usb analysers... Look for ANT-16, etc.

Al Welch 

-----Original Message-----
From: AT91SAM@AT91... [mailto:AT91SAM@AT91...] On Behalf Of
Peter Wintulich
Sent: Wednesday, January 18, 2006 7:46 AM
To: AT91SAM@AT91...
Subject: [AT91SAM] Detecting absence of I2C bus devices

Hello,

I have an interesting problem with the TWI in the at91sam7s64.

I am trying to detect the absence of a device on the TWI bus.
I had expected to receive a NACK in the status register upon sending the
device address.
Insted I received 0xFF data back and no error in the twi_sr.

1) is there something I am missing.

2) is there a correct procedure for detecting the absence of a device on the
bus.

I was planing to use this feature (normaly available on I2C) to implement
module detection.

Regards Peter
-- 

Peter Wintulich
VoiceTronix
Powering Open Telephony
Ph . +61 8 8232 9112
Fax. +61 8 8232 9113
www.voicetronix.com.au
	Yahoo! Groups Links
	

Hello,

I have an interesting problem with the TWI in the at91sam7s64.

I am trying to detect the absence of a device on the TWI bus.
I had expected to receive a NACK in the status register upon sending the 
device address.
Insted I received 0xFF data back and no error in the twi_sr.

1) is there something I am missing.

2) is there a correct procedure for detecting the absence of a device on 
the bus.

I was planing to use this feature (normaly available on I2C) to 
implement module detection.

Regards Peter
-- 

Peter Wintulich
VoiceTronix
Powering Open Telephony
Ph . +61 8 8232 9112
Fax. +61 8 8232 9113
www.voicetronix.com.au