Can anyone advise on the cheapest way to do PoE for powering the Bifferboard?
The setup: 4 Bifferboards approximately 20M from the hub. Assume each requires 5v @1.2A just like the standard PSU supplies.
So do I get 4x48v PSUs, 4x injectors/splitters, 4x 48v->5v buck regulators?
Or would I be better off doing this at 24v over the wire which may make the PSUs slightly cheaper. I believe the ethernet cable can take about 0.4 amps so I'm assuming I need at least 24v over the wire for this to be practical, but if it results in a substantially cheaper solution I might be prepared to risk a lower voltage. I do want to be able to power a 500mA USB device connected to the Bifferboard though.
Not sure about cheap sources for 24v/48v PSUs, if it's worth obtaining a multi-rail PSU for this, or, (god forbid) investing in a proper PoE hub, although they seem to be way over-priced even bearing in mind the hassle of setting up the necessary converters, PSUs etc..
Love to hear from anyone who got something like this working and where you sourced your kit.
No recommendations, as I've never done this myself, but a question
that might be relevant:
Are any of the bifferboards in the same location, i.e. could 2
bifferboards share the same PoE line, or are they all in disparate
locations?
Lurch
On 3 August 2012 20:39, biffe...@yahoo.co.uk <biffe...@yahoo.co.uk> wrote:
> Can anyone advise on the cheapest way to do PoE for powering the
> Bifferboard?
> The setup: 4 Bifferboards approximately 20M from the hub. Assume each
> requires 5v @1.2A just like the standard PSU supplies.
> So do I get 4x48v PSUs, 4x injectors/splitters, 4x 48v->5v buck regulators?
> Or would I be better off doing this at 24v over the wire which may make the
> PSUs slightly cheaper. I believe the ethernet cable can take about 0.4 amps
> so I'm assuming I need at least 24v over the wire for this to be practical,
> but if it results in a substantially cheaper solution I might be prepared to
> risk a lower voltage. I do want to be able to power a 500mA USB device
> connected to the Bifferboard though.
> Not sure about cheap sources for 24v/48v PSUs, if it's worth obtaining a
> multi-rail PSU for this, or, (god forbid) investing in a proper PoE hub,
> although they seem to be way over-priced even bearing in mind the hassle of
> setting up the necessary converters, PSUs etc..
> Love to hear from anyone who got something like this working and where you
> sourced your kit.
> thanks,
> Biff.
> --
> To unsubscribe send email to bifferboard+unsubscribe@googlegroups.com
> No recommendations, as I've never done this myself, but a question
> that might be relevant:
> Are any of the bifferboards in the same location, i.e. could 2
> bifferboards share the same PoE line, or are they all in disparate
> locations?
using 4 spare wire to carry DC directly for load 500-600mA at 5V
distance 20m resistance about 3.8 ohm cause voltage drop to 3.6V
that may not enough for USB device.
Increase supply voltage to 6V help voltage at target increase to 4V
make LDO 3.3V give stable supply to bifferboard CPU.
When using low power USB device or no USB device
assume 100mA load (about 50 ohm load),
the voltage at end = 50 ohm * 6V/(3.8 ohm + 50 ohm) = 5.6 V
still safe for bifferboard?
> Sorry, what I meant to say was, I want the flexibility to site them
> anywhere.
> cheers,
> Biff.
> On Aug 3, 9:28 pm, "biffe...@yahoo.co.uk" <biffe...@yahoo.co.uk>
> wrote:
> > It's a temporary setup, so you can assume that they are all in
> > independent locations.
> --
> To unsubscribe send email to bifferboard+unsubscribe@googlegroups.com
On Aug 4, 6:45 am, EMBES_Nathanan Srisa-ard <natha...@embes.com>
wrote:
> Hello,
> using 4 spare wire to carry DC directly for load 500-600mA at 5V
> distance 20m resistance about 3.8 ohm cause voltage drop to 3.6V
> that may not enough for USB device.
Hi,
Thanks for the reply, but I was hoping to setup something a little
more robust, so would like to introduce a non-linear step-down (buck)
regulator at the Bifferboard end. If I estimate based on cable
voltage drop I will end up with a system that only works for certain
cable lengths.
Estimating voltage drop on a cable is only
useful as a means of determining the worst
case scenario.
For example, it is interesting to note that
some load of 500mA or 600mA may cause
a voltage drop of 3.6V in the cable, but note
that if there is no load then there is no
voltage drop.
For variable current (which is common for
things with CPU's in them) the instantaneous
current may be, for example, 500mA for a
moment and then maybe zero on the next
moment. And in these "next moments" the
device will be subjected to all the voltage,
without any drop, which may easily be
enough to fry it.
So I guess you must use a higher voltage
at the source and then regulate it to a lower
one at the destination.
Here is a simple step-down buck converter
which is a drop-in replacement for a
conventional 7805:
It is a little bit pricey though. I have a few
such switched 7805 drop-in replacements
for sale, but the work the other way around
(lower voltage up to 5V), but they provide
only 100mA of current.
> On Aug 4, 6:45 am, EMBES_Nathanan Srisa-ard <natha...@embes.com>
> wrote:
>> Hello,
>> using 4 spare wire to carry DC directly for load 500-600mA at 5V
>> distance 20m resistance about 3.8 ohm cause voltage drop to 3.6V
>> that may not enough for USB device.
> Hi,
> Thanks for the reply, but I was hoping to setup something a little
> more robust, so would like to introduce a non-linear step-down (buck)
> regulator at the Bifferboard end. If I estimate based on cable
> voltage drop I will end up with a system that only works for certain
> cable lengths.
The solution I arrived at is as follows:
At the switch end a 2A 24v PSU can be obtained relatively
inexpensively (£10). 5.5/2.1mm power jacks seem to be standard output
for these, they seem to be used for CCTV, although I'm not sure what
camera needs 24V!
This can be split into 4 ways with one of the standard power splitters
for CCTV use. I could wire this up myself but when people are selling
these for a couple of quid I didn't see any point.
An injector/extractor pair costs about £3. Sure, I could have saved a
small amount of money by using a patch panel and wiring the power to
the relevant terminals myself but I decided why take the risk for this
kind of money?
At the Bifferboard end you can buy 5.5/1.2mm power jacks with a length
of cable, they are about £1 each.
Step-down converters with screw terminals are available for about £3
from various Chinese manufacturers. They have terminals on one side
and a USB (host) connector on the other. They claim 3A rating but I
don't need anything like that.
The remaining part to obtain is USB host to Bifferboard power jack,
still TBD but I'll probably solder my own for that part.
So in short, it seems 24v is a better fit than 48v for powering the
Bifferboard. Assuming a 4v voltage drop on the line, and 1.2A
required at the Bifferboard end and 80% efficiency in the converter
I'll draw about 0.39A down the cat5 cable, which should be fine.
Only snag is I could be waiting a long time for the 24v -> 5v
regulator to arrive from China. Everything else can be sourced
locally.
> The solution I arrived at is as follows:
> At the switch end a 2A 24v PSU can be obtained relatively
> inexpensively (£10). 5.5/2.1mm power jacks seem to be standard output
> for these, they seem to be used for CCTV, although I'm not sure what
> camera needs 24V!
> This can be split into 4 ways with one of the standard power splitters
> for CCTV use. I could wire this up myself but when people are selling
> these for a couple of quid I didn't see any point.
> An injector/extractor pair costs about £3. Sure, I could have saved a
> small amount of money by using a patch panel and wiring the power to
> the relevant terminals myself but I decided why take the risk for this
> kind of money?
> At the Bifferboard end you can buy 5.5/1.2mm power jacks with a length
> of cable, they are about £1 each.
> Step-down converters with screw terminals are available for about £3
> from various Chinese manufacturers. They have terminals on one side
> and a USB (host) connector on the other. They claim 3A rating but I
> don't need anything like that.
> The remaining part to obtain is USB host to Bifferboard power jack,
> still TBD but I'll probably solder my own for that part.
> So in short, it seems 24v is a better fit than 48v for powering the
> Bifferboard. Assuming a 4v voltage drop on the line, and 1.2A
> required at the Bifferboard end and 80% efficiency in the converter
> I'll draw about 0.39A down the cat5 cable, which should be fine.
> Only snag is I could be waiting a long time for the 24v -> 5v
> regulator to arrive from China. Everything else can be sourced
> locally.
I've got 2 cables like that - a black one that came with a 2.5" USB
HDD caddy (it had two USB leads, one for power+data, and an additional
one just for extra power) and a silver one that came with a
USB-powerable mini network switch.
> Still might make my own.
If making your own, I guess it'd be cheaper to just solder the power
lead directly to your step-down regulators, rather than going the
extra step of using a USB connector.
> I've got 2 cables like that - a black one that came with a 2.5" USB
> HDD caddy (it had two USB leads, one for power+data, and an additional
> one just for extra power) and a silver one that came with a
> USB-powerable mini network switch.
Does it fit the Bifferboard jack properly?
> > Still might make my own.
> If making your own, I guess it'd be cheaper to just solder the power
> lead directly to your step-down regulators, rather than going the
> extra step of using a USB connector.
>> I've got 2 cables like that - a black one that came with a 2.5" USB
>> HDD caddy (it had two USB leads, one for power+data, and an additional
>> one just for extra power) and a silver one that came with a
>> USB-powerable mini network switch.
> Does it fit the Bifferboard jack properly?
Yup, it's also the same size jack plug / socket that a couple of
self-powered USB hubs that I have use :)
Which is quite handy as the HDD I put in the caddy very occasionally
suffers brownouts (which causes USB disconnects) when powered from the
twin USB leads (it must peak momentarily at over 1A), so I usually
just power it from a USB hub 2A supply instead.
>> If making your own, I guess it'd be cheaper to just solder the power
>> lead directly to your step-down regulators, rather than going the
>> extra step of using a USB connector.
> And I still need a 3.5mm jack plug for the Bifferboard, unless I want
> to butcher an existing PSU.
I meant you could solder the power-leads (running to the 3.5mm jack
plug) to the underside of the USB connector on your regulator, instead
of also butchering a USB lead as well ;) Depends if you want to avoid
'modifying' your regulators or not.
> >> I've got 2 cables like that - a black one that came with a 2.5" USB
> >> HDD caddy (it had two USB leads, one for power+data, and an additional
> >> one just for extra power) and a silver one that came with a
> >> USB-powerable mini network switch.
> > Does it fit the Bifferboard jack properly?
> Yup, it's also the same size jack plug / socket that a couple of
> self-powered USB hubs that I have use :)
> Which is quite handy as the HDD I put in the caddy very occasionally
> suffers brownouts (which causes USB disconnects) when powered from the
> twin USB leads (it must peak momentarily at over 1A), so I usually
> just power it from a USB hub 2A supply instead.
> >> If making your own, I guess it'd be cheaper to just solder the power
> >> lead directly to your step-down regulators, rather than going the
> >> extra step of using a USB connector.
> > And I still need a 3.5mm jack plug for the Bifferboard, unless I want
> > to butcher an existing PSU.
> I meant you could solder the power-leads (running to the 3.5mm jack
> plug) to the underside of the USB connector on your regulator, instead
> of also butchering a USB lead as well ;) Depends if you want to avoid
> 'modifying' your regulators or not.
Hope that's clearer. There is no soldering required, and no
butchering of anything. In fact the only 'electronics' is to strip
some wire ends with a wire stripper. Cost per Bifferboard is £11.71.
Assuming it works, of course!
>> I meant you could solder the power-leads (running to the 3.5mm jack
>> plug) to the underside of the USB connector on your regulator, instead
>> of also butchering a USB lead as well ;) Depends if you want to avoid
>> 'modifying' your regulators or not.
> I'm not sure you understood what I was describing, anyway I've added
I understood perfectly ;) I was just referring to your "Actually they
do seem to supply a USB host to Bifferboard-
compatible power jack. ... But I've not found a cheap source. Still
might make my own." comment. I was merely suggesting you could
'bypass' the USB connector (by soldering directly to the regulator's
PCB) if making up your own cables. Nevermind.
I think you mean "Ethernet hub" rather than "USB hub"? ;-)
Might be worth putting a warning on that page to be careful when using
PoE injectors / splitters. I'm sure the Bifferboard wouldn't be happy
if you plugged 24V straight into its Ethernet socket!
On Aug 5, 8:05 pm, Andrew Scheller <ya...@loowis.durge.org> wrote:
> I understood perfectly ;) I was just referring to your "Actually they
> do seem to supply a USB host to Bifferboard-
> compatible power jack. ... But I've not found a cheap source. Still
> might make my own." comment. I was merely suggesting you could
> 'bypass' the USB connector (by soldering directly to the regulator's
> PCB) if making up your own cables. Nevermind.
And I'd need to solder directly to the Bifferboard, or else obtain a
3.5mm jack with flying lead, so I was thinking to buy a proper 3.5mm
jack and lop the end off one of the USB 'A' to mini 'B' leads that I
have a pile of. It's the Bifferboard end that presents more of a
problem than the regulator end.
> I think you mean "Ethernet hub" rather than "USB hub"? ;-)
Yeah, fixed.
> Might be worth putting a warning on that page to be careful when using
> PoE injectors / splitters. I'm sure the Bifferboard wouldn't be happy
> if you plugged 24V straight into its Ethernet socket!
Well, it may actually survive it due to the isolator, however I
thought the whole idea of the injector/extractors was that they
protected you against this, by not actually taking the leads that
carry voltage all the way to the device. I must get hold of an
injector to check though.
> And I'd need to solder directly to the Bifferboard, or else obtain a
> 3.5mm jack with flying lead, so I was thinking to buy a proper 3.5mm
> jack and lop the end off one of the USB 'A' to mini 'B' leads that I
> have a pile of. It's the Bifferboard end that presents more of a
> problem than the regulator end.
Right, well I didn't know you had no 3.5mm jack plugs, and 'a pile of'
spare USB leads ;-P
:)
>> Might be worth putting a warning on that page to be careful when using
>> PoE injectors / splitters. I'm sure the Bifferboard wouldn't be happy
>> if you plugged 24V straight into its Ethernet socket!
> Well, it may actually survive it due to the isolator, however I
> thought the whole idea of the injector/extractors was that they
> protected you against this, by not actually taking the leads that
> carry voltage all the way to the device. I must get hold of an
> injector to check though.
Yeah, that *is* what the extractor does (AIUI). But if you don't
realise what you're doing and aren't careful, it may be all too easy
for a 'noob' to plug the 'loaded' ethernet cable directly into the
Bifferboard, forgetting to insert the extractor inbetween them...
The bifferboard might survive, but your ethernet socket/isolator might
not. But I don't have any familiarity with PoE, so I might be talking
out of my arse!! ;-)
I wonder if "(Linear)" on the PCB means it's using a linear (rather
than switching) step-down regulator, which means it's less efficient,
which means more current will be pulled over the ethernet cable? Not
that it probably makes much difference in the grand scheme of
things...
