Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.sys.raspberry-pi > #9034 > unrolled thread

UPS recommendations?

Back to article view | Back to comp.sys.raspberry-pi

Contents

  UPS recommendations? Adam Funk <a24061@ducksburg.com> - 2015-07-03 12:25 +0100
    Re: UPS recommendations? Gordon Henderson <gordon+usenet@drogon.net> - 2015-07-03 14:08 +0000
      Re: UPS recommendations? "Dave Liquorice" <allsortsnotthisbit@howhill.com> - 2015-07-05 17:17 +0100
        Re: UPS recommendations? Martin Gregorie <martin@address-in-sig.invalid> - 2015-07-05 17:03 +0000
          Re: UPS recommendations? "Dave Liquorice" <allsortsnotthisbit@howhill.com> - 2015-07-05 20:58 +0100
      Re: UPS recommendations? Adam Funk <a24061@ducksburg.com> - 2015-07-07 17:00 +0100
    Re: UPS recommendations? "Dave Liquorice" <allsortsnotthisbit@howhill.com> - 2015-07-05 17:40 +0100
      Re: UPS recommendations? Martin Gregorie <martin@address-in-sig.invalid> - 2015-07-05 17:19 +0000
        Re: UPS recommendations? "Dave Liquorice" <allsortsnotthisbit@howhill.com> - 2015-07-05 20:39 +0100
          Re: UPS recommendations? Folderol <general@musically.me.uk> - 2015-07-05 21:24 +0100
    Re: UPS recommendations? Adam Funk <a24061@ducksburg.com> - 2015-07-07 16:57 +0100
    Re: UPS recommendations? Adam Funk <a24061@ducksburg.com> - 2015-07-10 21:18 +0100
      Re: UPS recommendations? Martin Gregorie <martin@address-in-sig.invalid> - 2015-07-10 22:42 +0000
    Re: UPS recommendations? David Taylor <david-taylor@blueyonder.co.uk.invalid> - 2015-07-11 16:19 +0100

#9034 — UPS recommendations?

Hi,

I've used one of those plug-through energy monitors to measure the Pi
and some other equipment that I'd like to protect with a UPS.

networking stuff: 
   26 W (average over 6 hours)
   27 W (instantaneous reading when I checked)
   49 VA (" ")
   PF 0.54 (" ")

Pi + external hard drive
   6 W (average over 14 hours, including some intensive file transfers)
   5 W (instantaneous reading when I checked)
   13 VA (" ")
   PF 0.41 (" ")

So I figure 65 VA + safety margin.  I'd prefer a small, quiet UPS (to
go in a cupboard) with surge protection.  

It would be a bonus if it can be configured not to shut the Pi down
immediately when the power goes out (in case of brief interruptions) &
to wait a few minutes after the power comes back on before turning the
Pi back on (in case the power flickers before staying on).

I'd be grateful for any recommendations.

Thanks.

[toc] | [next] | [standalone]

#9035

In article <u18i6cxiv7.ln2@news.ducksburg.com>,
Adam Funk  <a24061@ducksburg.com> wrote:
>Hi,
>
>I've used one of those plug-through energy monitors to measure the Pi
>and some other equipment that I'd like to protect with a UPS.
>
>networking stuff: 
>   26 W (average over 6 hours)
>   27 W (instantaneous reading when I checked)
>   49 VA (" ")
>   PF 0.54 (" ")
>
>Pi + external hard drive
>   6 W (average over 14 hours, including some intensive file transfers)
>   5 W (instantaneous reading when I checked)
>   13 VA (" ")
>   PF 0.41 (" ")
>
>So I figure 65 VA + safety margin.  I'd prefer a small, quiet UPS (to
>go in a cupboard) with surge protection.  
>
>It would be a bonus if it can be configured not to shut the Pi down
>immediately when the power goes out (in case of brief interruptions) &
>to wait a few minutes after the power comes back on before turning the
>Pi back on (in case the power flickers before staying on).
>
>I'd be grateful for any recommendations.

APC Back-UPS 500 (500VA)

It has serial and USB interfaces and I think it is compatible with upsd.

The Back-UPS 300 is a little smaller/cheaper but only has the serial
output.

I have a few of the 300's.

Most small UPSs have IEC (kettle) sockets, so you might need to make
up a standard 4-way 13A extension with an IEC plug on it...

Gordon

[toc] | [prev] | [next] | [standalone]

#9043

On Fri, 3 Jul 2015 14:08:34 +0000 (UTC), Gordon Henderson wrote:

>> I've used one of those plug-through energy monitors to measure the
Pi
>> and some other equipment that I'd like to protect with a UPS.

Be aware that at low power levels some plugin monitors can be way off
and non unity PF's confuses 'em even more.

>> networking stuff: 
>>   26 W (average over 6 hours)
>>   27 W (instantaneous reading when I checked)
>>   49 VA (" ")

Are thos figures really W (power) or Whr (energy). For sizing a UPS
you really need to know the energy consumption of all the kit for the
period of time you wish it to be maintained. You could have a 300 W
UPS but a tiddly battery, yes it'll deliver the 300 W but maybe only
for a couple of minutes.

> APC Back-UPS 500 (500VA)

Again power not energy is that 500W for 5 mins or 5 hours? Most UPS's
of that size have 2 x 12 V, 7 Ahr batteries. Say 65 W load at 80%
effciency give 80 W from the batteries. 80 W @ 24 V = 3.3 A. 7 / 3.3
= *about* 2 hours, in practice you don't want to complely drain the
batteries, so real maintained time will be shorter.

Also be aware that APC have a reputation for cooking their batteries.
The charge voltage is to high for float charging @ 20 C, let alone
the 40 C the batteries end up at in the case with the high charge
voltage. The high voltage is to shorten the recycle time when the
power returns and the UPS has shut down on low battery. Trouble is
the charger is "dumb" and doesn't drop back to the lower float
voltage when the battery is recharged.

I'm in the middle of a long term trial of my UPS (APC Smart UPS 750)
after getting at it to reduce the charge voltage and fit a fan to
keep the batteries cool.

-- 
Cheers
Dave.

[toc] | [prev] | [next] | [standalone]

#9045

On Sun, 05 Jul 2015 17:17:40 +0100, Dave Liquorice wrote:

> Again power not energy is that 500W for 5 mins or 5 hours? Most UPS's of
> that size have 2 x 12 V, 7 Ahr batteries. Say 65 W load at 80% effciency
> give 80 W from the batteries. 80 W @ 24 V = 3.3 A. 7 / 3.3 = *about* 2
> hours, in practice you don't want to complely drain the batteries, so
> real maintained time will be shorter.
>
FWIW my Riello UPS has, AFAIK, a pair of 7Ah 21v SLAs (Sealed Leadacid 
Accumulators) installed. and I'm using it to run a Dual Athlon desktop 
PC, for which I don't have any good power consumption figures. The Riello 
says the batteries are reading 27v, that the UPS is supplying 0.3A at 
230v and that the available run-time from the batteries is 52 minutes. 
That seems to show that its usable capacity is about 60 W.hours, which is 
in rough agreement with your battery capacity estimate.

I'm currently running it with the batteries permanently in circuit - it 
can be run two selectable modes (a) with the batteries continuously in 
circuit or (b) with the batteries kept charged on standby, ready to be 
switched on of the mains drops.

> Also be aware that APC have a reputation for cooking their batteries.
> The charge voltage is to high for float charging @ 20 C, let alone the
> 40 C the batteries end up at in the case with the high charge voltage.
> The high voltage is to shorten the recycle time when the power returns
> and the UPS has shut down on low battery. Trouble is the charger is
> "dumb" and doesn't drop back to the lower float voltage when the battery
> is recharged.
> 
> I'm in the middle of a long term trial of my UPS (APC Smart UPS 750)
> after getting at it to reduce the charge voltage and fit a fan to keep
> the batteries cool.
>
Interesting.

At least the Riello has a fan in it. Its currently reporting an internal 
temp of 35C for an internal temp in my house of 26C.



-- 
martin@   | Martin Gregorie
gregorie. | Essex, UK
org       |

[toc] | [prev] | [next] | [standalone]

#9049

On Sun, 5 Jul 2015 17:03:26 +0000 (UTC), Martin Gregorie wrote:

> FWIW my Riello UPS has, AFAIK, a pair of 7Ah 21v SLAs (Sealed Leadacid 
> Accumulators) 

21 V @ a nominal 2 V/cell?  B-)

SLA = Sealed Lead Acid 

> I'm currently running it with the batteries permanently in circuit - it 
> can be run two selectable modes (a) with the batteries continuously in 
> circuit or (b) with the batteries kept charged on standby, ready to be 
> switched on of the mains drops.

(a) = on line
(b) = off line

(a) means that the invertor etc is running all the time. Possibly not
the most effcient mode with mains available but the supply from the
UPS will be clean.

(b) means that the mains power is routed through the UPS, it might go
via the invertors transformer with automatic tap switching to reduce
or increase the mains voltage back to where it should be for the UPS
output. This does mean that glitches on the mains may get through and
there will be a glitch when the mains fails and the invertor has to
start up. Probably more effcient and if your kit isn't bothered by
the switching glitch less stress on the elecronics.

> At least the Riello has a fan in it. Its currently reporting an internal 
> temp of 35C for an internal temp in my house of 26C.

26 C *inside*. Glad I'm not there! *Far* too hot. 13 C outside and 22
inside ATM. Inside rarely gets above 23 C.

My APC UPS 750 VA is saying 45% load and 35 C internal temp. That's
for two PC's and the VOIP/Router/Switch etc

-- 
Cheers
Dave.

[toc] | [prev] | [next] | [standalone]

#9061

On 2015-07-03, Gordon Henderson wrote:

> APC Back-UPS 500 (500VA)
>
> It has serial and USB interfaces and I think it is compatible with upsd.
>
> The Back-UPS 300 is a little smaller/cheaper but only has the serial
> output.
>
> I have a few of the 300's.

OK, obviously I'd need USB output for a Pi.  500 VA sounds way over
what I need, & I'm a bit concerned about the 40 dB noise --- that's
supposedly the same as a fridge, & I'm aiming to get it in the back of
a wardrobe (an irregularly shaped one with plenty of air space around
the equipment, not with clothes pressed up).


> Most small UPSs have IEC (kettle) sockets, so you might need to make
> up a standard 4-way 13A extension with an IEC plug on it...

Not a problem!


-- 
They do (play, that is), and nobody gets killed, but Metallic K.O. is
the only rock album I know where you can actually hear hurled beer
bottles breaking against guitar strings.             --- Lester Bangs

[toc] | [prev] | [next] | [standalone]

#9044

On Fri, 3 Jul 2015 21:06:21 +0000 (UTC), Martin Gregorie wrote:

> I think you'll find the cheapies just run their loads until the 
> batteries are almost empty and then dump the load. 

Ultimately that is what happens but if the UPS can report it's status
to the Pi. The Pi can decide what to do, just use a time out and "on
battery" status rather than "low battery" as the shutdown trigger.
Will NUT (Network UPS Tools) run on a Pi? That can certainly do that
but may be a bit heavy weight.

> I think any PC with a mechanical power switch and that doesn't use disk 
> encryption will simply restart by default when the power comes back 

PC's haven't had a "proper" power switch since the ATX power supply
came in what 15 years ago...

"Soft switched" PCs are a bit of PITA. You can configure them to
shutdown and power off and have them power up on power return. Seems
OK but if the UPS doesn't drop the power to the PC it'll stay
shutdown. This is a minor niggle here. I have the server switch
itself off if the UPS is on battery for more than two mins, this is
to extend the maintained period for the VOIP phones as much as
possible. This can be the best part of an hour, if the mains comes
back in that hour the UPS doesn't drop the power to the server and it
doesn't restart.

There is a hardware work around involving the +5 V SB rail the normal
+5 V rail and a relay. The result of me asking how to automagically
resolve the problem years ago, probably over in uk.d-i-y, I've yet to
impliment it as unexpected power outages are very rare so the tuit
takes even longer to arrive. 

-- 
Cheers
Dave.

[toc] | [prev] | [next] | [standalone]

#9046

On Sun, 05 Jul 2015 17:40:32 +0100, Dave Liquorice wrote:

> There is a hardware work around involving the +5 V SB rail the normal +5
> V rail and a relay. The result of me asking how to automagically resolve
> the problem years ago, probably over in uk.d-i-y, I've yet to impliment
> it as unexpected power outages are very rare so the tuit takes even
> longer to arrive.
>
To me the obvious mod would be to buy a mechanical switch (rocker or push 
on/push off) from Maplins, Farnell or wherever that fits in the hole left 
when you remove the soft switch. Use a bit of acrylic or epoxy-glass 
sheet to block the hole/form a mount for the switch and Bob's your auntie.
 
Get the plastic or epoxy-board from your local friendly model shop (if 
you're lucky enough to have one in these degenerate times) or off eBay.


-- 
martin@   | Martin Gregorie
gregorie. | Essex, UK
org       |

[toc] | [prev] | [next] | [standalone]

#9048

On Sun, 5 Jul 2015 17:19:45 +0000 (UTC), Martin Gregorie wrote:

>> There is a hardware work around involving the +5 V SB rail the
normal 
>> +5 V rail and a relay. The result of me asking how to
automagically 
>> resolve the problem years ago, probably over in uk.d-i-y, I've yet
to 
>> impliment it as unexpected power outages are very rare so the tuit

>> takes even longer to arrive.
>
> To me the obvious mod would be to buy a mechanical switch ...

But that requires some one to be present to operate said switch. This
is no different to the UPS not toggling the power if mains comes back
before the UPS shuts down on low battery. Without the mains toggle
the server stays shutdown rather than restarting automagically.

-- 
Cheers
Dave.

[toc] | [prev] | [next] | [standalone]

#9050

On Sun, 05 Jul 2015 20:39:03 +0100 (BST)
"Dave Liquorice" <allsortsnotthisbit@howhill.com> wrote:

> On Sun, 5 Jul 2015 17:19:45 +0000 (UTC), Martin Gregorie wrote:
> 
> >> There is a hardware work around involving the +5 V SB rail the
> normal 
> >> +5 V rail and a relay. The result of me asking how to
> automagically 
> >> resolve the problem years ago, probably over in uk.d-i-y, I've yet
> to 
> >> impliment it as unexpected power outages are very rare so the tuit
> 
> >> takes even longer to arrive.
> >
> > To me the obvious mod would be to buy a mechanical switch ...
> 
> But that requires some one to be present to operate said switch. This
> is no different to the UPS not toggling the power if mains comes back
> before the UPS shuts down on low battery. Without the mains toggle
> the server stays shutdown rather than restarting automagically.

Said switch could just as well be the relay referred to above :)
Besides, it would still need the availability of the tuit.

-- 
W J G

[toc] | [prev] | [next] | [standalone]

#9060

On 2015-07-03, Martin Gregorie wrote:

> On Fri, 03 Jul 2015 12:25:18 +0100, Adam Funk wrote:
>
>> So I figure 65 VA + safety margin.  I'd prefer a small, quiet UPS (to go
>> in a cupboard) with surge protection.
>>
> That, the UPS capacity, pretty much depends on what you want the UPS to 
> do.
>
> I installed a 1000 VA Riello Sentinel Pro, but then again its there to 
> tide my Dual Athlon house server over power glitches and to power it 
> during a controlled shut-down if the power is out for more than 5 minutes 
> (yes, its overkill as things stand - it can run the server for 55 mins 
> (it says), but a basic UPS didn't provide some of the features I wanted - 
> controlled shut-down being one of them and, because it has 4 output 
> sockets, I may yet make it power a few other items, such as the router, 
> if I decide to increase the time it will power the computer to , say 30 
> mins.

Sounds good, but way more power than I need.  I'm also a bit concerned
about noise, i.e., being allowed to put it in the wardrobe.


>> It would be a bonus if it can be configured not to shut the Pi down
>> immediately when the power goes out (in case of brief interruptions) &
>> to wait a few minutes after the power comes back on before turning the
>> Pi back on (in case the power flickers before staying on).

On further reflection, that's pointless; the UPS ought to keep the
devices powered thorugh the flickers.


> If you just want the UPS to power an RPi plus a few low power add-ons, 
> you might want to look at something like this:
>
> http://piups.net/

I've seen that, but AFAICT it won't power the external hard drive too.


-- 
Nam Sibbyllam quidem Cumis ego ipse oculis meis vidi in ampulla 
pendere, et cum illi pueri dicerent: beable beable beable; respondebat 
illa: doidy doidy doidy.                                   [plorkwort]

[toc] | [prev] | [next] | [standalone]

#9078

On 2015-07-03, Adam Funk wrote:

> Hi,
>
> I've used one of those plug-through energy monitors to measure the Pi
> and some other equipment that I'd like to protect with a UPS.
>
> networking stuff: 
>    26 W (average over 6 hours)
>    27 W (instantaneous reading when I checked)
>    49 VA (" ")
>    PF 0.54 (" ")
>
> Pi + external hard drive
>    6 W (average over 14 hours, including some intensive file transfers)
>    5 W (instantaneous reading when I checked)
>    13 VA (" ")
>    PF 0.41 (" ")

I guess the thing that bugs me the most about UPSs is that they all
(AFAICT) assume you want 230 V AC, so they have an inverter.  All the
devices I'm interested in protecting are powered by wall warts with DC
output.  (I haven't checked that they all have the same output, but I
suspect that all of them are 5 V DC.)  Some power & a lot of
components & cost are wasted in running an inverter to supply a bunch
of transformers & rectifiers.

Why hasn't anyone responded to the market for this sort of set-up?


-- 
All crime is due to incorrect breathing.
                 --- Sir Henry Rawlinson

[toc] | [prev] | [next] | [standalone]

#9081

On Fri, 10 Jul 2015 21:18:18 +0100, Adam Funk wrote:

> 
> I guess the thing that bugs me the most about UPSs is that they all
> (AFAICT) assume you want 230 V AC, so they have an inverter.  All the
> devices I'm interested in protecting are powered by wall warts with DC
> output.  (I haven't checked that they all have the same output, but I
> suspect that all of them are 5 V DC.)  Some power & a lot of components
> & cost are wasted in running an inverter to supply a bunch of
> transformers & rectifiers.
> 
> Why hasn't anyone responded to the market for this sort of set-up?
>
They have: you'll find them on the specialist RPi vendor websites, e.g. 
ModMyPi. 

I suspect the reason you won't find what you want anywhere else is that 
UPS aren't common anyway (where do you find 240v ones outside data 
centres and places running specialist fault-tolerant kit. 

Now consider that the RPi and other small systems such as the Beagleboard 
are a tiny market compared with mains powered kit. Of these, the RPi has 
the lion's share of the market, so it follows that about the only place 
you'll find a UPS suitable for an RPi-based system is in an RPi accessory 
shop.


-- 
martin@   | Martin Gregorie
gregorie. | Essex, UK
org       |

[toc] | [prev] | [next] | [standalone]

#9092

.. and were it just for the RPi (which yours isn't, I appreciate), you 
might like a large battery which can be float-charged at the same time 
as powering the RPi.  For example, the 10400 mAh unit here, stock number 
775-7517:

 
http://uk.rs-online.com/web/c/batteries/rechargeable-batteries/power-banks/?searchTerm=power+bank&h=s&sra=oss&redirect-relevancy

-- 
Cheers,
David
Web: http://www.satsignal.eu

[toc] | [prev] | [standalone]

Back to top | Article view | comp.sys.raspberry-pi

csiph-web