AD1938 mis-operation with SPI port COUT logic changes.
In my application we are using a AD1938 or the differential output AD1939 CODEC.
As we need to support a number of bit rates we intend to control the AD1938/9 with the SPI port.
On our proto type we see strange operation of the SPI port.
In my lab tests of a AD1938 CODEC I note that any time the state of COUT is changed from low to high the audio stream is disrupted EVEN when CLATCH is high at all times.
I note app note "
Stand-alone Modes on the AD193x audio CODECs
This note states the SPI port pins are sampled at a 4mS rate and any time a match of the two allow standalone pin patterns is found standalone mode is entered.
Worth noting and of interest the above app note is contradicted by the AD1937 data sheet that says
"There is also a standalone mode available for operation without serial control, configured at reset using the serial control pins." This just add interest to the misery as to how the SPI port pins really work on the AD193x family of parts.
By the above app note definition of the SPI port pins, holding CLATCH high should never allow standalone mode to be entered or a SPI transfer or occur and so the logic state of COUT should never affect CODEC operation in any way.
Yet from my lab tests on a AD1938 CODEC SPI port the above app note does not seem to be the whole story.
The app note above suggests that if CLATCH is held high there should be no effect on codec operation by any change in state of COUT, CIN or CCLK.
In my lab tests of a AD1938 CODEC I note that any time the state of COUT is changed from low to high the audio stream is disrupted EVEN when CLATCH is held high at all times.
This is non-standard operation for a SPI interface and makes the normal sharing of one master SPI MOSI port with several slave SPI MISO ports impossible as any activity on the shared MISO line disrupts the audio stream from the AD1938 CODEC.
How does Analog Devices envision using the AD1938/9 in a system that has multiparty SPI slaves sharing a single SPI master.
I also refereed to app note AN-1365 but this note does not address a shared slave system. Nor does it does cover the reading back of data from the CODEC on the COUT port in question or any noise issues that may occur with data activity on the COUT pin.
So my questions are.
1) Why does the logic state of the COUT pin affect codec function when the SPI CLATCH pin is held high?
2) Because CLATCH being high does not seem to isolate the CODEC from the COUT pin logic state how can one use this part in a shared single master, multi slave SPI port system with the AD1938 being one of many slaves?