Re: [DIYbio] Re: Does Anyone Know Of A Submersible Arduino/RaspberryPI Compatible Spectrometer?

In terms of power I'm not really a fan of inductive technologies, the coupling just isn't good enough for an air core system if you're using anything short of Tesla coil voltages (it would negate the benefits of switching to a lower wattage chip.)  Wireless is something I like to avoid always (I was a software hacker well before a hardware hacker and have spent about a decade focused mostly on computer security, I don't think I could sleep at night if anything I cared about depended on a wireless connection.)  A wire just isn't enough of a pain in my scenario to warrant any form of wireless technology (I'm looking at sticking 20 of these things in different tanks in the same room not more than 50ft from a server room, and after that a bunch more of them outdoors not more than 5,000 ft away in the worst case scenario, I'd rather build some repeater shacks every 1,000 ft than use wireless.)  I know this isn't a common concern with the modern IoT philosophies, but this stuff will probably be sitting on an air-gapped network (or at least behind a firewall and managed switch blocking routes even as far as the firewall) because it just doesn't need to ever be accessed remotely for any reason.  With all the horror stories of foreign hackers going after IP cameras and routing equipment and governments targeting the infrastructure of eachother to have an economic impact there's a very real threat which there's just no reason to expose it to (it's only for an aquaculture research lab, but that's still enough.)

On Wednesday, September 6, 2017 at 1:22:41 PM UTC-4, Jonathan Cline wrote:

It's 2017, those arguments against python are over 10 years old and
python won. There's no need to rehash that stuff.  "Python is like
someone crossed c with whitespace" is simply blatantly incorrect, even
grossly incorrect.

The UBW32 (aka PIC32) does, or did, have an ethernet daughtercard
available and unix networking is supported in the BSD operating system
I believe, but double check that.  The UBW32 has been around for
years.  I found it really neat to write unix code and run it on the
board, without all the baggage that something like the Raspberry Pi
would bring with it.  PoE adapters to 12V or 5V are typical off the
shelf, I'm not sure if there's a dedicated daughtercard available.
True PoE compliance has a certain level of complexity (especially in
Q-A testing), so an external unit is probably best anyway.

If you choose suitable hardware, there's no need for PoE or ethernet.
Why run wire when you don't have to. PoE also seems like a great
solution theoretically but it just is not that simple to add onto a
board because it's intended for strict industrial grade use, it could
be far easier to just run power separately.  Or since there's liquids
involved you could also look at wireless power (or inductive, aka less
efficient cordless toothbrush style) which would be great for user
simplicity and safety.  Some embedded water sensor modules use
inductive power themselves anyway (with optical data comms).

The emerging platform that I see as a good way forward for these types
of sensor projects is, this might sound surprising, the microbit,
which has a long roadmap ahead of it and a huge built-in consumer
base, so it's longevity is assured.  It also runs micropython (or C or
ARM9 assembly, if you want), it has suitable storage, the SoC has very
good low power modes and peripherals, and the board has a built-in
Nordic 2.4 Ghz radio which sounds like a good fit for your
requirement, of periodic data chunks (~32 byte payload etc).  With
minor optimizations (i.e. turn off unnecessary peripherals and use
sleep mode with wakeup timers), the hardware could run from a few
rechargeable batteries and a small solar panel indefinitely, or
minimal wired power if you really want.    I haven't used the microbit
but I have used that SoC a few times for commercial projects.

It used to be, for low power applications i.e. sensing or control or
DSP stuff, I'd instantly cross off any SoC that had hardware floating
point, because that used to indicate a certain level of power
inefficiency or lack of support for low power CPU modes (low power
designs compromised by using integer math to be optimal).  The Cortex
is in that design space of modern SoC's where it finally doesn't
really matter anymore.


On 9/6/17, Cory Geesaman <co...@geesaman.com> wrote:
> I'd agree Arduino is a hacky solution for the reasons you listed, but
> coding speed tends to be determined by familiarity with a language.  I can
> code just as fast in c as c++ or c# or JavaScript, on the other hand Python
>
> is like someone crossed c with whitespace (because that's what it is.)
>  Having indentation as a syntactical requirement only appeals to people who
>
> agree with the indentation chosen while making it a serious pain to port
> code from other languages because the normal stuff (keyword and library
> differences) are all very obvious while the whitespace requirements don't
> stick out as well.
>
> The UBW32 seems interesting, are there any PoE+ethernet communication
> boards which are pretty straightforward to add to it?
>

--
## Jonathan Cline
## jcl...@ieee.org
## Mobile: +1-805-617-0223
########################