SPI bus with ADUM1401

Hi lkdgironi, 

You have correctly identified a very commonly overlooked issue. When multiple isolated SPI devices share the same MISO line on the controller side, we need to worry about contention between the various outputs. The standard data isolators are not SPI devices, they are protocol agnostic. They don't know a CS or /CS when they see it, so a designer needs to make use of the VE1 pin to make the isolator act like an SPI interface.

Its important to set the CHD output to high-Z mode when the SPI device on the other side is not being addressed to keep the uC side MISO signal free for traffic from other devices. And like you pointed out, an SPI device with active low (/CS) signal will need that signal inverted when its sent to the VE1 pin.  

There is AN-1478 that goes into the details on SPI isolation, and that may be useful for general guidance. 

I'll also recommend the ADuM341E. This would be the latest/greatest in isolation technology (and is pin-compatible with the ADuM1401).

Regards,

Jason

Thank you jnaso. The AN-1478 is really interesting. Most interesting part for my purpose is page 9. On "Figure 13. MISO Tristate for Slaves Isolated from One Another", they connect the isolators with VE1 in parallel with VIA (which is CS from controller uC to CS of target device). But from datasheet of ADuM1401, page 23: "VE1 - Output Enable 1. Active high logic input. VOD output is enabled when VE1 is high or disconnected. VOD is disabled when VE1 is low. In noisy environments, connecting VE1 to an external logic high or low is recommended.". So if VE1 is high VOD is enabled, but in the AN-1478 wiring CS on device side will be high, that's the reason of my inversion.
Maybe the AN-1478 is for other purpose?, for some "active high" CS devices?

I've check the ADuM341E, that's sounds interesting, above all for the pin compatibility.

Thank you!

It probably would have been better to include an active low (/CS) SPI device in AN-1478, Figure13. That is probably the more common SPI interface. I'll keep it in mind if I do a revision in the future. 

Here's what it would look like with inverters (with Schmitt trigger) added in.

Thank you!!!
I've just noticed the name on top of the AD-1478. It's a pleasure to speak with one author of that paper!

I'll add a smith trigger. Like below, which should work better.

By the way this mux is going to be used with the AD2S1210 resolver, that is something I'm asking here in another post about SPI communication ez.analog.com/.../ad2s1210-schematic-check-in-spi-mode