|
|
Relay Switched
Router UPS
 This
simple Uninterruptible Power Supply (UPS) was built as an
addition to my 12V solar powered battery backup system to give
continued internet access in the event of a mains power cut
(conveniently my router runs on 12V). I thought about using MOSFET
transistors to do the switching but decided instead to use a relay because there will always be isolation between the
main and backup power supply due to the air gap between the contacts.
The
relay chosen draws less than 13mA at 12V when energised so hardly
adds any extra
load to the router's mains adapter. One problem with relays is that
there is a moment when the
contacts
are in mid air as they move from one position to the other, effectively
disconnecting the load for a short period of time. This won't
matter
too much if the load being
switched is, for example, a light bulb where all you might notice is
the
light dim for a fraction of a second, but if the relay is switching a
router, a brief power interruption will cause it
to quickly turn off
and on again taking your network down for several minutes
while it reboots. The purpose of this circuit is to provide power
during the changeover period so that the router doesn't drop out  Schematic
Under
normal
operating conditions
the relay is energised, supplying 12V to the router from its own mains
adapter via the normally open (NO) contacts which are now closed
because this same
voltage is also used to power the relay. If there is a mains failure
the relay will de-energise, the contacts will return to their normally
closed (NC) position and the router will now be supplied from the
12V backup battery. If the relay actually switched over at 12V it would
be fine but in practice it doesn't, because another problem with
relays is that they don't energise at their specified voltage. The
relay used in this circuit has a
12V
coil but will pull the contacts at approximately 8V and, even worse, only
release them when the coil voltage drops to approximately 4V. So without this circuit, the relay would de-energise only when its supply voltage has fallen below 4V which will cause a
router designed to work at 12V to reboot. Conversely, when
the router is running from the battery and
mains power is restored, the relay would energise at approximately 8V which
still isn't enough for the router to operate and again it will reboot.
This occurs because the mains adapter that powers the router
has an internal smoothing capacitor which takes time to discharge when
power is removed and charge when power is restored. To address the two main issues that relays have (not
cutting in and out at their specified voltage and relatively slow
changeover times) we need a circuit that not only makes the relay
operate at 12V (slightly under is better), but can also supply
power during the contact changeover period
 Veroboard layout
Circuit operation is straight forward.
The relay
is switched by an NPN transistor with its base connected to the 12V
supply via a zener diode and current limiting resistor. With this arrangement the transistor
will only switch on and power the relay when the supply voltage rises
to around 11V because of the voltage dropped across the 10V zener and
the transistor base junction. So the relay now pulls and releases close to 11V
and the router will continue to operate
uninterrupted during the contact changeover period due to the charge stored in the two 2200u reservoir capacitors
which is also close to 11V. The only reason that two 2200u capacitors are used in parallel instead of a
single 4700u is because it's housed in a compact enclosure and two smaller caps are 5mm shorter than a single larger one! It's worth noting that some routers have enough internal capacitance to
allow them to 'ride the storm' during very brief voltage dips, in which
case the capacitors may not be required (not so my TalkTalk router).
The diode across the relay coil is to protect the transistor from
being damaged by the large reverse voltage that occurs when the
magnetic field in the relay core collapses
Usually,
connecting
relay poles in parallel to increase the switching
current is not recommended because one pole could connect and/or
disconnect slightly
before the
other causing it to pass more amps than it's design rating, which could
prematurely burn out the contacts. Here it's OK though, as I'm not
aiming to double
the current capacity but halve it by sharing the load between the two
poles. The design rating is never exceeded
The circuit is constructed in a Hammond/Eddystone diecast enclosure (type 27134PSLA)
with 'chunky' 2.1mm DC power connectors for inputs and
output. Because this type of
socket has the
negative terminal connected
directly to its body, the enclosure can serve as a common ground. Due
to the enclosure not having much headroom, the board is mounted on two
5mm M3 hex pillars which were the longest I could get away with while
still allowing the capacitors to fit with
the lid on (20mm was the lowest height capacitors I could find). The pillars are also used to ground the circuit down to chassis
by simply adding four little 'dobs' of solder, one either side of each
mounting hole (shown on the board layout as 4 black dots) |
