Powering a RPi (was Re: Cellular dongles)

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

Powering a RPi (was Re: Cellular dongles)

Joel Maxuel
I may as well add this to the greater groups at this point.

Considering what I plan to attach to the RPi, I will want to ensure all the power I can get:

- Two Phidget InterfaceKits (an 8/8/8 and a 0/16/16)
- The E303 cell dongle

What complicates the situation is that the RaZberry sits on top of P1 (covers the top ten pins) although there is a spot on the RaZberry to solder on another header to use those ten pins covered:


 As long as enough power can pass through back down to the RPi, it should be fine.

Additionally, I would want this battery-backed as I will be using it to send a notice in the event of a power loss.  I could use an existing UPS, but avoiding additional power conversion (battery DC to wall wort AC back to DC for the RPi) would make more sense here.

I will eventually (in some years[1]) have a 12VDC lead-acid bank to work from, so my interim solution should be setting up something that will be easy to adapt to that as well.  (Or, since I need to add a marine battery[2] to the basement next month anyway, maybe I should run a couple wires and adapt from that?)

[1] As part of a future-planned solar-electric system.

[2] Whenever the power goes out during or shortly after a rain storm, my basement starts to flood at the 20 minute mark - so getting a battery-backed emergency sump to complement the existing one.

--
Cheers,
Joel Maxuel

"One should strive to achieve, not sit in bitter regret."
 - Ronan Harris / Mark Jackson


On Sun, May 26, 2019 at 2:11 PM Dave Flogeras <[hidden email]> wrote:

On Sun, May 26, 2019 at 8:06 AM Joel Maxuel <[hidden email]> wrote:

Interestingly, when I looked up the E173, the bands did not match my provider.  However the E303 does and appears to tick the other boxes.

So I took a risk (or two since it is a Huawei being shipped from the USA).

Cool, let me (and the group(s)) know how this works out.   Another data point is always better.

At the same time, I may need to borrow a RPi once I get to a specific phase of prototyping.


Unfortunately I'm probably not a great resource there, mine are generally fully provisioned running bits of my network.  However, I will mention this:  Most cell modems are _quite_ power hungry.  I know on older raspberry pi hardware versions, they had issues when you powered via the micro usb port.  Even if you had a beefy wall-wart milliamp wise, you were still putting everything through the little polyfuse which didn't like heavier drawing devices.  I'd recommend reading if they did any corrective measures on newer hardware (if you have access to say a 3 or 3+).  If not, you may need to power it via the GPIO header P1, which supplies the voltage after the polyfuse.

I can happily help you out making any cabling necessary, or explaining better in the future.  Just ask

Dave

_______________________________________________
nSLUG mailing list
[hidden email]
http://nslug.ns.ca/mailman/listinfo/nslug
Reply | Threaded
Open this post in threaded view
|

Re: Powering a RPi (was Re: Cellular dongles)

Dave Flogeras

On Mon, May 27, 2019 at 9:06 AM Joel Maxuel <[hidden email]> wrote:
I may as well add this to the greater groups at this point.

Considering what I plan to attach to the RPi, I will want to ensure all the power I can get:

- Two Phidget InterfaceKits (an 8/8/8 and a 0/16/16)
- The E303 cell dongle

What complicates the situation is that the RaZberry sits on top of P1 (covers the top ten pins) although there is a spot on the RaZberry to solder on another header to use those ten pins covered:

 
 As long as enough power can pass through back down to the RPi, it should be fine.


Another possibility if you have other requirements, would be to lay out a simple pass through PCB.  Adafruit sells stacking headers:

And this way you could position your board between the pi and Razberry2.  You could break out the supply pins, and any other peripherals for any customization you need (UART, I2C for realtime clock....).  It's pretty cheap to get PCBs made up at OSHPark and if you have a couple random circuits, it can greatly clean up your project.  When you're getting closer, I'd be happy to show you enough KiCad to make you dangerous.
 
Additionally, I would want this battery-backed as I will be using it to send a notice in the event of a power loss.  I could use an existing UPS, but avoiding additional power conversion (battery DC to wall wort AC back to DC for the RPi) would make more sense here.

I will eventually (in some years[1]) have a 12VDC lead-acid bank to work from, so my interim solution should be setting up something that will be easy to adapt to that as well.  (Or, since I need to add a marine battery[2] to the basement next month anyway, maybe I should run a couple wires and adapt from that?)


I've seen several boards go by which are a UPS integrated right with the Pi.  It avoids the inverter making AC just to turn it back into DC problem.  Also, most inexpensive AC UPS' I've seen put out really dirty power that you might not want to subject your delicate electronics to.



I have no direct experience when any of these, but might work for you.

_______________________________________________
nSLUG mailing list
[hidden email]
http://nslug.ns.ca/mailman/listinfo/nslug
Reply | Threaded
Open this post in threaded view
|

Re: Powering a RPi (was Re: Cellular dongles)

Joel Maxuel
Considering what I plan to attach to the RPi, I will want to ensure all the power I can get:

- Two Phidget InterfaceKits (an 8/8/8 and a 0/16/16)
- The E303 cell dongle

What complicates the situation is that the RaZberry sits on top of P1 (covers the top ten pins) although there is a spot on the RaZberry to solder on another header to use those ten pins covered:

 
 As long as enough power can pass through back down to the RPi, it should be fine.


Another possibility if you have other requirements, would be to lay out a simple pass through PCB.  Adafruit sells stacking headers:

 Interestingly enough, I discovered one of those conclusions earlier today with the MoPi2 (paired with the stacking header):

...which will allow up to 2.5A (given if each peripheral peaks at the same time - that can still leave 960mA) for the Pi.
 
And this way you could position your board between the pi and Razberry2.  You could break out the supply pins, and any other peripherals for any customization you need (UART, I2C for realtime clock....).  It's pretty cheap to get PCBs made up at OSHPark and if you have a couple random circuits, it can greatly clean up your project.  When you're getting closer, I'd be happy to show you enough KiCad to make you dangerous.

Not sure how important that detail will be, as the finished product will be in my basement (likely mounted to the wall in a panel box).  You mentioned breakout for an RTC - is internal time-keeping that bad, or just plain non-existent?

Additionally, I would want this battery-backed as I will be using it to send a notice in the event of a power loss.  I could use an existing UPS, but avoiding additional power conversion (battery DC to wall wort AC back to DC for the RPi) would make more sense here.

I will eventually (in some years[1]) have a 12VDC lead-acid bank to work from, so my interim solution should be setting up something that will be easy to adapt to that as well.  (Or, since I need to add a marine battery[2] to the basement next month anyway, maybe I should run a couple wires and adapt from that?)


I've seen several boards go by which are a UPS integrated right with the Pi.  It avoids the inverter making AC just to turn it back into DC problem.  Also, most inexpensive AC UPS' I've seen put out really dirty power that you might not want to subject your delicate electronics to.



I have no direct experience when any of these, but might work for you.

That would probably be a more practical solution, as my current backup power solution includes a subpanel for "life-support" circuits, one of which feeds the core networking (and the RPi will be a part of that).  Although there can be more flexibility for never having to shut down to power shortage, my subpanel (generator-fed currently, option for solar/reserve some later time) limits that time anyway.

_______________________________________________
nSLUG mailing list
[hidden email]
http://nslug.ns.ca/mailman/listinfo/nslug
Reply | Threaded
Open this post in threaded view
|

Re: Powering a RPi (was Re: Cellular dongles)

Dave Flogeras
On Mon, May 27, 2019 at 1:44 PM Joel Maxuel <[hidden email]> wrote:
 Interestingly enough, I discovered one of those conclusions earlier today with the MoPi2 (paired with the stacking header):

...which will allow up to 2.5A (given if each peripheral peaks at the same time - that can still leave 960mA) for the Pi.

That's the one I couldn't remember earlier!  Make sure you read up on any candidates, there is always a varying degree of quality/support.  Some of these products are just pet projects, with no intention of doing proper software/kernel level support after it is released.  YMMV.  IIRC I evaluated a Pimoroni E-Paper board, and while it did do what it said it would, it wasn't integrated in any way to the operating system; just some hacks and python scripts to communicate with it.  Same went for the early Wolfson audio boards, unsupported bash nonsense and user daemons in lieu of a proper driver (which the community has since created).


 
And this way you could position your board between the pi and Razberry2.  You could break out the supply pins, and any other peripherals for any customization you need (UART, I2C for realtime clock....).  It's pretty cheap to get PCBs made up at OSHPark and if you have a couple random circuits, it can greatly clean up your project.  When you're getting closer, I'd be happy to show you enough KiCad to make you dangerous.

Not sure how important that detail will be, as the finished product will be in my basement (likely mounted to the wall in a panel box).  You mentioned breakout for an RTC - is internal time-keeping that bad, or just plain non-existent?


The Pis don't include an RTC, probably to save a bit of BOM cost/real estate of having a coin cell on board.  It's easy enough to add one; I've spit rolled my own but I would imagine kits are readily available by now.  Most people don't care, since typically they are always connected to get NTPD to update the clock.  In my experience, the internal clock doesn't skew more than your average PC.

Dave

_______________________________________________
nSLUG mailing list
[hidden email]
http://nslug.ns.ca/mailman/listinfo/nslug
Reply | Threaded
Open this post in threaded view
|

Re: Powering a RPi (was Re: Cellular dongles)

Joel Maxuel
On Mon, 2019-05-27 at 14:30 -0300, Dave Flogeras wrote:

> On Mon, May 27, 2019 at 1:44 PM Joel Maxuel <[hidden email]>
> wrote:
> >  Interestingly enough, I discovered one of those conclusions
> > earlier today with the MoPi2 (paired with the stacking header):
> > https://pi.gate.ac.uk/pages/mopi.html  
> >
> > ...which will allow up to 2.5A (given if each peripheral peaks at
> > the same time - that can still leave 960mA) for the Pi.
> >
>
> That's the one I couldn't remember earlier!  Make sure you read up on
> any candidates, there is always a varying degree of quality/support. 
> Some of these products are just pet projects, with no intention of
> doing proper software/kernel level support after it is released. 
> YMMV.  IIRC I evaluated a Pimoroni E-Paper board, and while it did do
> what it said it would, it wasn't integrated in any way to the
> operating system; just some hacks and python scripts to communicate
> with it.  Same went for the early Wolfson audio boards, unsupported
> bash nonsense and user daemons in lieu of a proper driver (which the
> community has since created).
>

I hear that.  The MoPi2 documentation seems quite complete except ...
the device comes with a charge gate terminal pair if you want to
control charging of one of the voltage inputs (say 8 x 1.2V NiMH AA
cells - picking on this scenario as it is their default).  I am having
difficulty in finding a corresponding charging circuit aside from a
couple circuit sketch-ups for other scenarios (for use in a vehicle;
for connecting a Li-Po battery) and seemingly impossible for an off-
the-shelf companion for their software default.

An annoying outcome since this was featured in the shortlist of an
Instructables article, and most in that list (how to power a RPi) were
not floating something third party.  

>
> >  
> > > And this way you could position your board between the pi and
> > > Razberry2.  You could break out the supply pins, and any other
> > > peripherals for any customization you need (UART, I2C for
> > > realtime clock....).  It's pretty cheap to get PCBs made up at
> > > OSHPark and if you have a couple random circuits, it can greatly
> > > clean up your project.  When you're getting closer, I'd be happy
> > > to show you enough KiCad to make you dangerous.
> > >
> >
> > Not sure how important that detail will be, as the finished product
> > will be in my basement (likely mounted to the wall in a panel
> > box).  You mentioned breakout for an RTC - is internal time-keeping
> > that bad, or just plain non-existent?
> >
> >
>
> The Pis don't include an RTC, probably to save a bit of BOM cost/real
> estate of having a coin cell on board.  It's easy enough to add one;
> I've spit rolled my own but I would imagine kits are readily
> available by now.  Most people don't care, since typically they are
> always connected to get NTPD to update the clock.  In my experience,
> the internal clock doesn't skew more than your average PC.
>

Oh,that makes more sense.  Yeah, that won't bother me.  I
misinterpreted it as it would not be unable to tell time if it didn't
have constant network access.
_______________________________________________
nSLUG mailing list
[hidden email]
http://nslug.ns.ca/mailman/listinfo/nslug