[sdiy] nice Digi Pot

[rsdio at audiobanshee.com]

On Jun 27, 2016, at 1:42 PM, Vladimir Pantelic <vladoman at gmail.com> wrote:
> On 27.06.2016 22:17, rsdio at audiobanshee.com wrote:
>> Daisy-chaining is rarely necessary. Usually, it's just a waste of a pin. If
> 
> what pin does it waste? sorry, I do not follow.

For a write-only serial device, there is no need for a data output pin, since you can't read back any values. In that case, the daisy-chain output is basically a wasted pin. Some chips don't have a spare pin to dedicate to an output that's not used for the chip's function.

>> your main processor's MOSI has sufficient fan-out, you can simply connect it
>> directly to the input of every SPI slave. If the fan-out is weak, then an
>> external buffer will suffice.
> 
> and how do I address every individual slave?

Decoder or dedicated I/O pins. Lots of interface chips have spare I/O pins that can be used in a pinch.

>> The only time daisy-chaining is really useful is when you have a
>> serial-to-parallel converter, and you want downstream parallel devices to
> 
> daisy chaining is useful when I want to control a lot of similar devices from a single SPI connection, why would I not want that?

This is fine, so long as there isn't any negative side-effect to the requirement that all chips be updated on every write.

For the example of a group of DPOT chips, your latency will increase if you have to write to all of them every time one of them changes. If the user is only changing the setting on one DPOT, it's faster to simply access one DPOT and leave all the rest at their previous setting until they need an update.


By the way, I was thinking that a daisy-chain output would require extra circuitry inside. But any serial device is going to need a shift register anyway, so it's probably not that much extra to hook that up to a pin, if one is available on the package. But extra drivers are needed for signals that flow out of the chip.

Brian