Hey guys,
I'm posting to see if anyone would be interested in a pass-through
shield that provided the option of voltage isolation. I've been
hearing from a few adventurous people their stories of blowing up
boards so thought it may be worth considering, some good progress has
been made on this technology front lately and given the increased
industrial usage scenarios of the Bone it sounded like a plan.
If I looked at producing something like this, would anyone be
interested?
If so, what price range would you be looking at, how many channels
would you wish to be isolated at any one time and would you be looking
for any specific isolation qualities?
Cheers guys,
Alistair
If you were to provide isolation, What function would each pin have? Would each pin be bidirectional for example? Would those signals requiring high frequency be unaffected by the isolation? Would all pin mux options be supported?
Gerald
Gerald Coley wrote:
If you were to provide isolation, What function would each pin have? Would each pin be bidirectional for example? Would
those signals requiring high frequency be unaffected by the isolation? Would all pin mux options be supported?
Gerald
one could make a shield-cape with all pins totally unconnected for maximum isolation...
As isolation isn't cheap we'd look to provide less channels than there
are pins but allow for various jumper configurations and attempt to
reach all pins that could benefit from isolation.
We're looking at RF isolators, they're high speed and have great
response times. Unidirectional you'd be able to achieve 150Mbps, we're
still looking at bidirectional and the approaches we can take, we'll
definitely ensure there's a provision for this but standard offerings
don't provide this at high rates.
We'll base our decisions of the feedback provided and so it would be
great to know what everyone is looking for.
Cheers
As isolation isn't cheap we'd look to provide less channels than there
are pins but allow for various jumper configurations and attempt to
reach all pins that could benefit from isolation.
My personal feeling is that a limited number of signals is OK (do try
to pick them wisely), but that having a lot of manual configuration to
try to get to other signals results in a less usable product. Try to
select well and make things programmatic rather than jumper
configurable.
We're looking at RF isolators, they're high speed and have great
response times. Unidirectional you'd be able to achieve 150Mbps, we're
still looking at bidirectional and the approaches we can take, we'll
definitely ensure there's a provision for this but standard offerings
don't provide this at high rates.
Consider a bit of a mix, but try to make sure any unidirectional
drivers you have going into the Bone won't blow it up in case someone
configures the muxes wrong. I'm pretty confident it'll happen.
Theoretically, the EEPROM on the board should contain some mux
information that will setup the pins in a happy way after boot, but I
wouldn't trust the software infrastructure is there that would prevent
pins being configured on the Bone to be in output mode if this is for
hacking. It would destroy the concept if your board meant for
protection became the source of damaging Bones.
Consider isolation and level shifting for the A/D converters. I think
lots of people will have difficulty using the 1.8V inputs.
Consider shifting up to 5V.
My guess is that protection is more important than performance for
such a product.