:: Re: [DNG] question on programming (…
Pàgina inicial
Delete this message
Reply to this message
Autor: o1bigtenor
Data:  
CC: Devuan ML
Assumpte: Re: [DNG] question on programming (maybe OT???)
On Thu, Mar 21, 2024 at 10:15 AM Vince Mulhollon <vince@???>
wrote:

> On Thu, Mar 21, 2024 at 8:42 AM o1bigtenor via Dng <dng@???>
> wrote:
> > Some would consider what I want to set up as IoT, some might suggest
> using SCADA some would suggest a web based system - - - but I'm wondering
> if I could be served by something else.
>
> Sometimes it helps to look at similar working systems.
>
> You are reinventing the PLC. Codesys is a commercial soft-PLC that
> runs on Debian, quite cheap and affordable on a raspi, I've messed
> around with this stuff and it works well. Even if you end up not
> using codesys if you're reinventing the concept of a PLC this would
> give you some insight into what you're trying to invent.
>
> One system-level suggestion I'd make is the cost of a pi or three is
> pretty low compared to the cost of the entire system's capex and opex
> so you're probably in territory where you have like three separate
> implementations. One's a literal electronic humidistat and fan,
> another third of the sensors and fans are on a "real PLC" like codesys
> and another third of the sensors and fans are on I donno some homemade
> Node-RED thing or homeassistant automation or a big ole Python script.
>


Fan(s) in question run some over 6k cfm - - - this is not quite regular
household stuff.
Controlls are available - - - I don't like because not a one is rated to
work below
0C (32F) and as my facility is unheated - - - - well this winter like was
on the warm
side but we still got to -20C (if not slightly colder).
This lack in commercial controls is one reason why I want to roll my own!

>
> Also, you can have the systems monitor each other with available GPIO.
> Nothing wrong with the PLC "keeping an eye on" the behavior of the
> other two systems.
>
> You might even want to break the sensors/fans into quarters and have
> one system be as simple as a timer, no sensors at all.
>
> "All hands on deck" everything has to row in the same direction
> probably statistically never happens but complete system failure might
> be very bad, so having multiple control systems would help overall
> reliability.
>
> Another systems-level suggestion is you have four separate problems,
> acting, logging what it did, and monitoring/reporting unusual short
> term conditions and long term trends. It's likely you need four
> separate "systems" not one monolith that tries to do it all. So far
> only talking about the "acting" PLC solution.
>
>

Interesting ideas - - - thanks!