DAC63004: Daisy-chain operation without NVM

David Watling

Part Number: DAC63004

Hi,

I'm trying to configure daisy-chain operation, as detailed in the datasheet.

First thing is to set SDO-EN in the INTERFACE-CONFIG. This should cause SDO (the GPIO pin) to output the end of the shift-register.

But I think there is a bug because although SDO does reflect the output of the shift-register for bits [22:0], bit 23 is always asserted.

This is the 'read/write' bit, now set to 'read', so you CANNOT write data to the next devices in a daisy-chain.

Has anyone else come across this?

Have I found an un-fixable bug?

9 months ago

0 Katlynne Jones 9 months ago

TI__Mastermind 30158 points

Hi David,

Unfortunately this is a "bug". The SDO pin of every device in the chain would need to be enabled and saved in NVM before daisy chain can be used. Another user came across this issue a few months ago in the forums as well: https://e2e.ti.com/support/data-converters-group/data-converters/f/data-converters-forum/1243464/dac53204-problem-configuring-2nd-dac-in-daisy-chain-when-dacs-power-up-in-3-wire-spi-mode/4741869?tisearch=e2e-sitesearch&keymatch=dac%252525252525252A3204%2525252525252520daisy%2525252525252520chain#4741869

Bit 23 being set is expected when the first DAC is echoing a command, but because the SDO pin of the second device is disabled, it accepts the first 24-bits of data that is sees rather than the last 24-bits. This is not a valid command for the second device due to bit 23 being set like you mentioned. If SDO were to be enabled first the device would be accepting the last 24-bits and daisy chain would work as expected.

Best,

Katlynne Jones

0 David Watling 9 months ago in reply to Katlynne Jones

Prodigy 10 points

Thanks for your confirmation that this is a bug Katlynne. There appear to be 2 reasons:

1. When SDO is enabled (for device#1), whatever is input to SDI gets output, but the first bit D23 (the read/write bit) is overwritten with a '1' so device#2 always sees a 'read' request.
2. Before SDO is enabled for device #2, it only reads the first 24 clock-cycles, so it looks impossible to program anything into device #2.
These two 'features' conflict, causing the issue.

Programming SDO in NVM is probably possible but very awkward as my devices are already daisy-chained in the hardware.

Can you find out what causes the first SDO bit D23 to be overwritten? And might there be a register setting which would fix that?

David Watling

0 Katlynne Jones 9 months ago in reply to David Watling

TI__Mastermind 30158 points

Hi David,

Bit 23 is hard coded in the design to '1' when the SDO is echoing the previous command. I asked the designer to check this when the other user pointed put this issue in the last thread. Because of this, there is no way around it via a register setting.

Best,

Katlynne Jones

0 David Watling 9 months ago in reply to Katlynne Jones

Prodigy 10 points

Hi Katlynne,
With a great deal of trouble I managed to program SDO_EN into the NVM whilst the devices were on the board. I cannot over-state how much trouble this bug has caused me, as we designed our PCB specifically for the daisy-chain feature.
Please can you ensure the datasheet is updated to emphasise that the daisy-chain feature will only work if the NVM is programmed before assembling these devices onto a PCB?
Best Regards

David Watling

Data converters

Data converters forum

DAC63004: Daisy-chain operation without NVM