Autor: o1bigtenor Datum: To: Simon CC: Devuan ML Betreff: Re: [DNG] Information request (maybe OT - - - dunno)
On Tue, May 16, 2023 at 10:29 AM Simon <linux@???> wrote: >
> o1bigtenor <o1bigtenor@???> wrote:
>
> >>>> PoE over CAT-5 can deliver 5V or 12V to multiple remote locations. Other options include solar and battery
> >>>
> >>> Hmmmmm - - - just did a little playing with ampacity charts
> >>> seems at 5V with 2.5 A (not that big a draw imo) well I need to run 10
> >>> ga if I'm running 5 m.
> >>
> >> I wouldn’t run 5V over any distance - you’ve little tolerance for volt drop, and it’s hard to provide any form of protection against being connected to the wrong voltage.
> >> Using passive PoE (basically just stick power on two of the wires) you can use 12V or 24V - with 24V you’d be down to just under 1/5A which is a lot more manageable. Not only that, but (assuming you a switching down-converter) you get to compensate for any volt drop in the conversion.
> >> Active PoE (e.g. 803.11AF) uses 48V. Also, the power sourcing equipment does not apply power unless it sees the right signal (basically a resistor), and provides protection against connecting stuff that wasn’t expecting power coming down the wires.
> >
> > Hmmmmmm - - - - sounds like 12V for short runs, 24V for runs up to
> > say a couple m and
> > 48V for at least up to 5 m - - - - that's off the cuff and will need checking.
> > Then I will need a buck transformer at each location. (Oops - - -
> > costs just went up
> > some more - - - grin!!! - - - - as the 'interesting factor goes up the
> > $$$$ factor is
> > escalating even faster!)
>
> I think you’ll find that packaged buck converters (a.k.a. switch mode regulators) are available quite cheaply - but as someone else has said, if you have local power for the rest of the rig (outstation) then just use that. Where PoE (or more generally, "power over same cable as data which needn’t necessarily be ethernet") really wins is where you have a low power requirement and can dispense with the need for a local power supply (hence AC outlet, cable & PSU or plug in PSU, etc.) which can really clean up the system. For CCTV systems PoE can mean not needing a power outlet at what may be a remote location, and for VoIP phones it means having a single cable to the phone which keeps desks tidy. But it’s not automatically the best option for every application.
I hadn't realized how important it was going to be to have power
available easily at every station - - - was more concerned about
capabilities
so your pointers are very useful - - - - likely have to develop a
specific system for this - - - - there might easily be more than 3 or
4 micro controller
or SoCs at each station so it seems to make sense to have solid strong
24V (useful for other things that even electronics) available at
every station.
(An interesting point!!- - - Thanks!!) >
> Whether passive or active (e.g. 802.11AF, or 802.11AT which does higher power over ethernet) the electronics for Po<Something> is now commodity item status with integrated devices available from a choice of suppliers.
>
> Another consideration is your communication.
> Centralised power with power feed to the outstations via the data cable is suited in a star arrangement.
> However, if you use a multi-drop bus arrangement (RS485, CANbus, etc.) then logic starts suggesting the units are daisy chained - data cable loops along the devices (but depends a lot on your physical layout of equipment). In this situation, centralised power likely doesn’t make sense as you’d still be adding star wiring for the power separate to the bus wiring for data. Of course, there’s nothing to stop you using star wiring for a bus based system, just paralleling all the connections in the hub unit (and observing any termination requirements) - and then centralised power makes sense again.
Have lately been considering using something like smallish unmanaged
switches liberally to minimize long cable runs.
Its enough to have one cable from each function center imo. Will be
checking with the amount of data that might
be passed to make sure that I'm not running past say 70% of capacity. >
>
> I would suggest you consider communication method first, and power option may then seem an obvious choice.
>
>
> To give an idea of how my thought processes work in this sort of thing ...
>
> I’m planning a DIY heating control setup for the house. My key priority is reliability so don’t plan to be using a full OS (e.g. Linux on ARM) or needing a control system (e.g. OpenHAB) to be running - and absolutely not reliant on working internet - for the heating to work.
>
> So I’ve more or less settled on Arduino with a central controller that takes care of the boiler, pump, valves, and monitoring the thermal store temperatures - and a separate controller for each room also using Arduino. By analogy, if the most important thing for you project is that each weighing cell keeps working even if data logging etc. is down then you might want to concentrate on a self contained reliable controller in each unit.
>
> For simplicity, I intend using ethernet (just pop bytes in a buffer, call a routine to send a packet, bytes appear in buffer at other end), and PoE to power the room controllers so they don’t need local power (the thermal valve actuators only take about 3W each) - each controller can just periodically send a status packet to the central controller so it can decide what needs to be running and what flow temperature to set. So far, autonomous and doesn’t require a server to be running.
>
> As I want logging, and a GUI, and such like, there will be a server. Each Arduino can ping packets to it and the server can run software to extract information and log it - but if the server falls over, the system keeps running. Using ethernet means the comms is simple with the (commodity PoE switch) taking care of routing packets breweenc multiple devices.
> And then I can run some sort of "control system” with web interface etc. from which the Arduino’s can poll for program changes (i.e. how warm to have each room vs. time of day). Again, if the server is down they’ll carry on with what they last pulled down. Worst case would be the server down and the Arduino’s get power cycled in which case they’d not have a time reference and would need to revert to a basic program until they could set the internal (software) clock.
>
> So I’m planning :
> Arduino for the stuff that needs to be reliable and run unattended for years - the WAF for a heating system that needs contact upgrades, patches, reboots, or anything like that would be “rather low”.
> A Linux box (virtual machine) to do the heavy lifting stuff like data logging and remote GUI.
> PoE for the comms to make the comms simple and eliminate the need for local power supplies.
> As it’s modular, it’s possible to do it in stages - so no need for a “big bang” go live with the whole system and I can iterate over time as I figure out each piece.
>
> The main thing I’m missing is the time to do any of this. The WAF for me spending such tinkering time is also low :-(
>
>
> Anyway, that’s how my thought processes go. Others will have a different process, and may well end up at a very different solution - for example, I know some people use something like OpenHAB to run everything which is fine if you are happy enough with needing a home server to be running for the heating to work.
> By analogy, a different topology you could consider is having weighing cells that have no smarts at all - and have the central box do all the I/O as well. That means lots of I/O in one box, and the wiring gets more complicated, but it means no electronics in the multiple cells.
>
> Thank you for your example - - - - that helps.
One of my problems is that each working node is likely generating data
that I do want to keep.
Some of the data is generated every say 0.5 sec and the data is needed
for longer term analysis
and at the same time when a preset flow rate is reached then there is
a another part of the system
that is halted. Then there will be specific functions that have to
happen and there is a limited amount
of time for this all to happen. Following this the process starts all
over again. Eventually up to 15 times
3 times per day - - - all to reduce the amount of human employee time
needed to get the job done.
Thank you for your assistance and ideas - - - - - muchas gracias!!!