In my experience, the chip knows chip select which in turn yields the first bit, which in turn yields the last bit.  The application does not depend on -CS.

-----Original Message-----
From: Harold Hallikainen [mailto:harold@HALLIKAINEN.COM]
Sent: Wednesday, March 17, 2004 12:41 PM
To: PICLIST@MITVMA.MIT.EDU
Subject: Re: [ee]: Multiple SPI slaves *without* chip select lines?

As other people have pointed out, a shift register ignores data until a
clock comes along. But, not knowing these parts, I still don't understand
how the device knows where the end of a byte or word is. A typical SPI
device clocks data through a serial in parallel out shift register, then
captures the data in a parallel latch on the trailing edge of -CS. Without
that, it seems that you'd never know where to latch the data.

Harold

> Agilent optical mouse sensors. They use something that can be thought of
> as
> SPI if you tie the Data In and Data Out lines together (only the sensor
> *or* the microprocessor is speaking at a time and this is ridgedly
> controlled) without a slave select.
>
> You *can* have SPI without a slave select, you can only have one slave.
>
> These changes are why I called it "spi-like".


--
FCC Rules Online at http://www.hallikainen.com

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads