Re: SPI-ADC

[H Hartley Sweeten <hartleys-3FF4nKcrg1dE2c76skzGb0EOCMrvLtNR@public.gmane.org>]

On Thursday, April 22, 2010 10:09 PM, arjun rath wrote:
> Hi ,

Please do a reply-all to keep the thread on the lists.  Also, please
do not top-post.  I fixed both here...

> On Thu, Apr 22, 2010 at 9:53 PM, H Hartley Sweeten <hartleys-3FF4nKcrg1dE2c76skzGb0EOCMrvLtNR@public.gmane.org> wrote:
>> On Wednesday, April 21, 2010 11:20 PM, arjun rath wrote:
>>> Hi Everybody,
>>>
>>> I am using a SPI based 12 bit adc(max1241) from maxim.I am trying to read
>>> the value from adc  using spi_read() in linux kernel.But every time i used
>>> to read i am getting *0xff* only.(i have to read two consecutive bytes).To
>>> the adc chip i am probing with CRO which always shows DOUT pin of ADC as
>>> high provided the input to ADC is always 1.1 volt.
>>> And the clock is also coming as seen in the CRO with Chip select line going
>>> low.
>>>
>>> Is it a delay problem since for adc's we have to wait for sometime for
>>> conversion of data....????
>>>
>>> Can anybody give some suggestion regarding this....
>>	
>> According to the datasheet, to start a conversion with the max124[0|1] you
>> need to pull the chip select low.  At the falling edge the T/H enters its
>> Hold mode and a conversion is initiated.  During this time the ADC will keep
>> the DOUT pin pulled low and the SCLK must be kept low during the conversion.
>> The end of conversion is signaled by DOUT going high.  At this point the
>> data can be shifted out with the SCLK signal.
>>	
>> It appears the SPI API is supposed to support this with the SPI_READY mode
>> flag.  But, the only SPI master driver in mainline that supports this flag
>> is the davinci_spi driver.
	
> Thanks for the reply.
>
> You mean to say with atmel spi master it is not possible(at91sam9263)...
> and to start the conversation i am pulling the chip select line low...

Looking at the atmel_spi driver it appears that it currently does not
support the SPI_READY mode.  So yes, with the driver as it is now it will
not work correctly with the max124[0|1] parts.

> Is it possible that i can write the serial protocol procedure what ever
> given in the datasheet instead of spi.

The max124[0|1] is still a spi device so using the atmel_spi driver is
the correct route.  You just need to add the necessary support to the
driver to handle the SPI_READY mode.

I added the author, Haavard Skinnemoen, to the Cc list to see if he has any
ideas to help.

Regards,
Hartley



------------------------------------------------------------------------------