Saturday, 11 August 2018

Tawny Owl NestBox Wifi Hotspot - setup the correct way...

Earlier this year I setup a Raspberry Pi v2 minicomputer as a wifi access point to use it as a 'Wifi Hub' so that two nestbox cams based on its smaller WiFi-only sibling, Rasp Pi Zero W could connect to my local network, described in 'Wireless access point bird box', this sort of worked but had issues.

This version is for the 2019 Tawny Owl nest box:
Owlbox 2019... work in progress

I'm trying the new Pi B+ to use as a remote, wifi access hub that will live in the box and allow other nearby/camera-enabled ZeroW's to connect via wifi. The PiB+ wifi hub will double up to stream a webcam & attached Raspi Camera module too.

My first attempt at this (Pi v2 no cameras + 2x Pi zero Ws, 1 camera module each) is all powered by one Power over ethernet (PoE connection).  The Pi B+ is described as having faster wired network (ethernet) connection, so I theorised that it could better act a video streaming hub for several Pi zero Ws.

Why not buy a commercial wifi access point you ask?
  • This build costs the price of a raspberry pi 3B+, case, SD card & power adapter - so not far off buying a commercial Wifi access point.
  • The access point will 'double up' to host/stream video for its own webcam and a RaspPi camera module - show me a commercial wifi access point that can do that...
  • Its more fun than just buying a commercial one.  Really.
  • You can re-task it as an arcade table (I made this one), Media server, Kodi box, weather station, etc, etc

My first attempt at this worked.  Mostly.  The problem was that the method that I used included adding a DHCP server.  Every network has a DHCP server on it somewhere. Domestic networks typically have one running on their Internet service provider-provided modem/router (BT/Virgin/Sky etc).  A DHCP server ' hands out' network addresses (called IP addresses) to local devices connected to the network (e.g. PC/tablet/smart TV).  Think of it as creating a 'map' so that network traffic can be routed between devices.

The presence of a second DHPC server on my Raspberry Pi v2  caused intermittent problems as there were two operating on my network, beavering away and assigning different IP addresses to the same devices.  Annoyingly the device they fought over the most was the TV, so understandably that was not well received.

So, plan B (+)

An upfront disclaimer, most of this method is not mine, and is reproduced from parts of this 'how to' , and credit to its author jamesh (Paspberry pi Foundation).  I re-jigged it a bit to make it flow better and added in my own comments.

The first bit deals with making an a standalone wifi network... which is what I did first without reading it properly....  You actually want to follow the second part of the section that deals with setting it up as an access point to an existing network. 

-----------------------------------------------------------
One common use of the Raspberry Pi is as an access point is to provide wireless connections to a wired Ethernet connection, so that anyone logged into the access point can access the internet, providing of course that the wired Ethernet on the Pi can connect to the internet via some sort of router.
To do this, a 'bridge' needs to put in place between the wireless device and the Ethernet device on the access point Raspberry Pi. This bridge will pass all traffic between the two interfaces.

Use the following to update your Raspbian installation:
sudo apt-get update
sudo apt-get upgrade

Install the following packages to enable the access point setup and bridging:
sudo apt-get install hostapd bridge-utils

Comment: reboot at this point, or the 'add bridge' bit below wont work.
Since the configuration files are not ready yet, turn the new software off as follows:
sudo systemctl stop hostapd

Bridging creates a higher-level construct over the two ports being bridged. It is the bridge that is the network device, so we need to stop the eth0 and wlan0 ports being allocated IP addresses by the DHCP client on the Raspberry Pi.
edit: May want to save a version of this text on the pi's desktop before doing the following as you'll lose the internet temporarily
sudo nano /etc/dhcpcd.conf

Add denyinterfaces wlan0 and denyinterfaces eth0 to the end of the file (but above any other added interface lines) and save the file.

Add a new bridge, which in this case is called br0.
sudo brctl addbr br0

Connect the network ports. In this case, connect eth0 to the bridge br0.
sudo brctl addif br0 eth0

Now the interfaces file needs to be edited to adjust the various devices to work with bridging.
sudo nano /etc/network/interfaces make the following edits.

Add the bridging information at the end of the file.
# Bridge setup
auto br0
iface br0 inet manual
bridge_ports eth0 wlan0

edit the hostapd configuration file, located at /etc/hostapd/hostapd.conf, to add the various parameters for your wireless network (SSID) and password (wpa_passphrase).
sudo nano /etc/hostapd/hostapd.conf

wpa_passphrase=pwdhere
interface=wlan0
bridge=br0
#driver=nl80211
ssid=NameOfNetwork
hw_mode=g
channel=7
wmm_enabled=0
macaddr_acl=0
auth_algs=1
ignore_broadcast_ssid=0
wpa=2
wpa_passphrase=AardvarkBadgerHedgehog
wpa_key_mgmt=WPA-PSK
wpa_pairwise=TKIP
rsn_pairwise=CCMP

We now need to tell the system where to find this configuration file.
sudo nano /etc/default/hostapd

Find the line with #DAEMON_CONF, and replace it with this:
DAEMON_CONF="/etc/hostapd/hostapd.conf"

Now reboot the Raspberry Pi.
There should now be a functioning bridge between the wireless LAN and the Ethernet connection on the Raspberry Pi, and any device associated with the Raspberry Pi access point will act as if it is connected to the access point's wired Ethernet.

The ifconfig command will show the bridge, which will have been allocated an IP address via the wired Ethernet's DHCP server. 
-----------------------------------------------------------
A few more observations...

Raspberry Pi v2 has no built in wifi so I added a usb wifi dongle to the original version I did.  
The Pi 3B+ has built in Wifi (both have ethernet), so there's no need to add wifi dongle.  My Pi v2 version works fine over Power over Ethernet (PoE).  The version B+ most definitely does NOT, so 2019 Owl Box will need a better power supply (and better power cable) to the top of its tree than the optimistically planned PoE only solution.


Sunday, 29 July 2018

Whats wrong with our Tawny Owl Nest Box?

Our Tawny Owl nest box was 'temporarily' retired last year because I hadn't designed it properly it was difficult to clean out the squirrel detritus.


This pic is taken while dangling precariously from a ladder approx 30ft up a pine tree. 
The box is full of sticks collected by squirrels, so no owls are going to use this... Ideally I would have designed it so that I wouldn't have to climb up another couple of extra feet to open the lid to clear it out, but I built it basically as a massive 'small bird box', and didn't scale up my lateral thinking facilities alongside.

Feeling creative I drew this up detailing all its other problems...


Its been on a 'to do' list for a while, and I was inspired by Kate Macrae's recent Tawny Owl Box re-build described here to get around to doing it.

To make matters worse, in an attempt to adapt it a little while ago, I took the side off (side facing in the diag above) to make a 'cleaning out hatch', and promptly misplaced the whole side.  It is now what could only be described as 'drafty'.  The front ledge fell off too...

Enter Tawny Owl Box v2
Its replacement aims to address all the issues described above, and add some more cameras with the requisite illumination (IR and visible light).  Probably.  Eventually.

Saturday, 7 July 2018

BirdBox infra-red camera IR filter upgrade

Now that the bird nesting season is over, it's upgrade time! The infrared (IR) cameras in this year's boxes are getting IR filters...

and some to spare...

This year's two new bird boxes have an infrared (IR) camera with visible and IR LEDs (Robin box Birch log tit box).  There is 'night mode' with IR light  that the birds cant see, and day mode with optional visible light illumination. All non IR-light sensitive cameras have an IR filter built into the camera, whereas IR-sensitive cameras don't.

Without an IR filter, IR camera images have a pinkish cast to them in visible light, giving a permanent 'Hipstamatic' look:

Un-filtered IR camera pink colour is obvious here

An IR filter (called an IR-cut) converts an IR-sensitive camera into a conventional one that can't see IR light.  This gizmo has a bit of glass in it that blocks IR light, and can move backwards and forwards over the camera lens by changing the voltage polarity applied across it.

These bird boxes use the 8MP Raspberry Pi IR camera board.  This has a similar shape and design to those found found in many commercial CCTV cameras, so IR filters designed for CCTV camera modules can be fitted to them.  Sort of...

My first attempt to do this was the top-down camera in last years's double camera bird box.  Unfortunately, I sourced an IR-cut that had an ?M12 lens mount on it, basically a dirty great chunk of plastic poking out the font that needed to be cut off before I could use it = fiddly.  You can see my v1 test for this here from a separate project:


I wanted something easy to add to this year's boxes, and found an IR cut filter without a lens mount which fits nicely over the existing screw holes in the Raspberry Pi camera module (22mm separation).

This come from AliExpress (China, so take a couple of weeks to get to the UK), and costs $3.50 so wont break the bank.  It comes with a separate lens mount holder which I haven't used.  There's the option of 20mm or 22mm screw hole separation at order - you need the 22mm one as to fit the Raspberry Pi camera board.


This shows the underside of the camera unit that is slotted in to the top of the birdbox, pre and post IR-cut upgrade. The ir-cut is attached over the camera by M2 nylon screws and nuts, and the screws just go through the 4mm ply and pass through the camera board.

IR-cut sitting flush against the 'underside' of the bird box camera insert

I had previously designed this year's Raspberry pi zero W add-on board with an IR-cut in mind.  The voltage polarity reversal to switch the IR-cut is done with a L293DNE (£2.70).  I did  a more detailed writeup of this here which also has some example code.  The design for the wiring of this my customised add-on board is detailed here.



No birds in this, but you get the idea what switching the IR-cut in natural daylight does...


...so there are alternatives to this, eg here which has integrated IR LED, camera and IR cut - does not have stellar reviews on amazon though.

Sunday, 1 July 2018

How to make a Solitary Bee Box

We noticed a solitary bee using a gap between some tongue and groove wood cladding on the rabbit shed, which led to this bee house build...

The aim of most solitary bee house designs should be to...
  • Provide a safe place for solitary bees to nest.
  • Provide somewhere safe for new adults to emerge in the Spring
  • Allow easy access to cocoons for collection and cleaning at the end of the season.
This is my v1.0 bee house attempt:


Bee house in situ

Here is this some of the 2020 batch waiting to hatch in Spring 2021 



Here is a male hatching early season 2021.  I love the drawer since I can easily open it without much disturbance, it also makes filming this happening is really easy to do:



Hole-nesting solitary bees will use a variety of nest sites, all are variations on a long narrow opening into which nesting chambers are constructed.  Examples include:
  1. Bamboo canes: Variable sizes kept in a container of some kind.  Can source yourself or get online.  Disadvantages are you cant collect cocoons so allows disease, predator and parasite build up over time.
  2. Reeds: As above, but may be able to split to get the cocoons out.  
  3. Drilled blocks/logs.  Same issues as bamboo
  4. Cardboard tubes with paper liners.  **NEW** Update spring 2021, I got a starter set from masonbees.co.uk, see here

  5. My masonbees.co.uk holder in situ


  6. Stackable trays with pre-cut channels.  Channels cut into flat pieces of wood that are stacked, e.g.  https://crownbees.com/, wildlifeworld.  This option came out on top, with cardboard tubes a close second.


Life-cycle 
Female bees collect pollen.  Female bee deposits the pollen in an appropriately sized hole.  She then lays an egg and closes up a chamber containing egg+pollen with a mud wall .  She repeats the process until the hole is full with multiple chambers.  In the Summer the eggs develop into larvae, feed on the pollen stores and develop into a cocoon.  In the following Spring the new bee emerges from the cocoon.  Males emerge first so that they're available to inseminate the females, who then go onto to perpetuate the life-cycle.

This v1.0 design is based on the stackable tray type.
  • Longer holes produce more females.  Females do more pollinating than the males (who hang about at the beginning of the season to inseminate the females then die).  We made 7.5 inch long trays, with the channels cut right to the end.  Some websites had examples up to 11 1/2 inches long, but I had limited timber stocks to work with!
  • Channels are 5/16 inch deep and 3/8 inch wide, which is the recommended size for mason bees.  I used this 1/4" square-end router bit.  I wanted a round ended (or 'bull nose') bit but couldn't find one the right size that didn't cost the earth.
  • Trays were cut from one piece of 18x199x2400mm pine (mixing imperial and metric, sorry), so I got 12 trays out of one piece.
  • Each tray has 8 channels giving 96 channels in total
  • Rather than routing these individually which would be a pain, I just routed the wood into one very long tray and then cut that into sections.
  • As the channels go through to the back, there's a piece of cardboard folded over the back end (as in this video).
  • The front of the block is unevenly scorched, as this apparently that helps them find the correct hole.  Other variations of this include marking the front, or using differing colour trays.
Stacked tray block before and after scorching
  • The stack is bound together by a tie down strap with metal cam buckle (I had to google the proper name for that) picked up from the local DIY shop, but can be sourced elsewhere.
  • There's an 'emergence drawer' at the base for any cocoons we collect this year (prob a bit late, may get a batch in for next Spring).  This gives a safe place for adults to emerge.  The holes in the front allow the newly emerged bees out (drawer would be closed...).
Safe place for new adults to emerge from cocoons in  the Spring




  • A South-Facing position is recommended, we settled for S-SE 

The effect of scorching is more obvious in this pic


The box housing the bee block is made from 'stuff lying about', but mostly is 6mm ply and finished with a coat of linseed oil.  No oils / preservatives are applied to the blocks themselves.

Most of the solitary bee info stuff that I could find in the web dealt with Mason bees, but there are many others which presumably have their own nesting preferences.

Some links that I found useful
http://nurturing-nature.co.uk (watch them build, had a spot on Springwatch 2018)
http://cvgss.org/mason-bees/
https://crownbees.com/
https://beekeepclub.com/crown-bees-mason-bee-house-48-hole-wood-nesting-tray-kit-review/
https://www.masonbees.co.uk/product-page/bee-lodge
https://threehundredandsixtysix.wordpress.com/2013/02/01/green-steps-building-a-mason-bee-house/

...so no tech in this one, but I like the idea of this bee counter discussion thread, which aims to count honey bees activity around a nest entrance hole using a raspberry pi camera... Could theoretically apply the same approach with a down-facing camera on this box...

Wednesday, 6 June 2018

2018 Birch log blue tits: The odds were stacked...

Our first bluetits hatched on the day of the Royal Wedding, so the first pair were predictably named Harry and Megan.  We were lucky enough to catch one at the point of hatching, the sequence below is speeded up a bit:


Others  followed soon enough...
day 1-2 , all looking good....
We had noticed previously, that the female bluetit had a tick next to her right eye, but didn't think much of it, since we all have to contend with the nasty critters.

Tick below right eye.  Really.
The tick disappeared within a few days, but she proceeded to have a fairly dramatic reaction with a lot of swelling around the eye which was also accompanied with feather loss; she was practically bald on the right side of her head.


Adding to her woes, the male appeared to have left the scene, so we had a one-eyed single mother with 5 yet-to hatch + 3 chicks to look after... not looking so good...

She was managing for a couple of days, but fell into a pattern where eggs would hatch and within a few days the new chicks died.  Megan and Harry were hanging in there though.

Food was varied, we saw a lot of huuuuge caterpillars / grubs that are not going anywhere (except maybe by tunnelling their way out).


Some sibling rivalry in the next clip, with Harry and Megan are fighting over a caterpillar...
You can see ?tick-related feather loss on the adult's head quite well here too.


When the adult feeds, she takes turns in selecting which chick gets fed.  One chick really wants to hang onto the food and gets flung about a bit.  An avian version of the magnetic fishing game.


As this is our first (semi) successful nest for several years, I'm not sure if the next clip showing a chick on top of its mother is normal behaviour, but she sometimes doesn't get the idea of sitting on the chicks to keep them warm...


The lack of a second adult, together with her erratic feeding lead to the inevitable deaths, while Harry and Megan soldiered on.  This clip shows her removing one of the smaller chicks that didn't make it.


Another observation from this nest was that there seemed to be quite a range of hatching intervals. Some chicks were hatching up to a week after the 'Royal pair'.  If this was an owl nest I could understand the concept of a spare/(?heir) or one saved for a lunch snack.  One theory I suppose could be that staggered hatching times maximises the chance of at least some surviving a short period of limited food.

Three big, one small.  

..so fast forward to day 6.  She removed 3 more dead chicks between 8:00 and 8:30 am, leaving two unhatched eggs and our original pair, Harry and Megan.


By this point, her right eye is looking seriously, well wrong...  I'm fairly sure this is tick-induced, however the tick itself appears to have gone.


Unsuccessful attempts at feeding include very large grubs and sunflower seeds (presumably from our feeder...)


In the end the female bird failed to overnight in the box and the Megan and Harry didnt make it through the Friday night.  She was back in the morning, even attempting to feed the dead chicks..



This clip is a great example of the 'cycle of life', so even if this nest has failed it still feeds the maggots, who in turn into flies to feed something else....



After clearing the last chick out (?Harry/?Megan), she did attempt to incubate for day or so, but eventually left.  We haven't seen her since.  We're hoping that she can recover from her tick injury without having to feed a brood.

So currently our BirchBox next box has this in it.  Black ants have finished off the cleaning up that the wrigglies missed:

Take two...

  • Days 1-14 of this nest are described here


Friday, 11 May 2018

2018 Birch log blue tits, Nest building, egg laying and incubation days 1-13

13 days in from the start of blue tit nest building, we're up to 8 eggs.  Three eggs were laid on day 6 (ouch) then approximately one egg per day.  She started incubating in the last couple of days.

Video and images captured using PikrellCam on a RaspberryPi Zero W.  This box's build is described here.

This is a slide show of nest building and egg laying using a mixture of IR and visible light.  The pinkish cast on the visible light is due to the lack on an IR-cut filter on the camera... I'll add one for next year.



Having looked at other blue tits nests, it seems that green tennis ball fluff is quite the 'in thing'.  In this one is also rabbit hair, sheep wool, straw, moss and feathers from my neighbour's chickens...

1st eggs laid on day 6

Day 9, egg in the sunshine, infra-red illuminated

Day 13, 8 eggs, incubating most of the time now

Some video clips

Female being fed on the nest while laying (not incubating in earnest yet)

This one with the lights on

All seems to be going well, other what is probably a tick above her right eye.  Goes with the territory in these parts...

Part 2 of this series is here, unfortunately it doesn't end well.... and the tick does not help...

Sunday, 29 April 2018

2018 Nesting Season, False starts and Late starters

I had almost given up on nesting this year after some promising early interest in the double camera box fizzled out. BUT all is not lost!...



Great Tit checking out double camera box
Side camera footage, female Great tit:


We had some nesting material brought in (this one has sound)


But that was about it for this box...

BUT all is not lost!

The BirchLog Box that I put up a week ago started off this morning like this:

BirchLog Box, Infra-red illuminated, blue tit. Clay blob on the floor, 8:30 am
The thing that looks a bit like a slug on the floor is a piece of air-drying clay which was supposed to help support the plywood roof, I used hot-glue in the end and didn't bother removing the clay, a piece fell out and seen here littering the floor.  I had thought to retrieve it, but in the end the blue tits cleared it out themselves.

By the end of the day, the box looks like this:
BirchLog box, visible light-illuminated, 6:30 pm
The blue tits have had a very busy day 1.  I got this nice pic of one on its way on:

Blue tit on its way into BirchLog box
Here are a couple of video clips of today's nest building efforts:

This is using the Infra-red illumination


This is is with the visible light leds on instead

Borrowing a feather from next doors chickens

Roosting overnight...




Sunday, 22 April 2018

Wi-fi Small Bird, 'Birch Log' Nest Box

This birdbox is a more compact version of my previous design, but basically the same kit used as in the as the robin box described previously: 8 mega-pixel digital camera, wifi IP connection, night-vision or visible light options with added illumination, and motion activated recording facility.

Update 01/01/19: Winter maintenance = addition of Infra-red IR cut here
Update 06/06/18: Chicks and a sticky end here
Update 11/05/18: Nest building, and egg laying here


This time I've equipped a cheap commercial nest box box, rather than make my own as I liked the 'natural' look of it. This box is supplied by CJ Wildlife, it's basically a hollowed out birch log with a 32mm hole at the front, which should be good for Blue Tits and Great Tits.  Total build cost is in the region of £70.

TamperProof.  Holds off any Woodpecker armed with a philips head screwdriver

My main problem with this box is that the  lid and the back baton are attached with 'TamperProof' Torx head screws.  I cant see the point since good old philips-head screws would have done just as well.  If you're going to buy this box, make sure you spend the equivalent cost on a set of appropriate screwdriver bits....  I also replaced the back batton to give an attachment point beneath in addition to the one at the top.

There is enough of a gap between the roof and the entrance hole to squeeze in the this little fella, which fits in snugly under the lid

Lid insert with Raspberry Pi Zero-W, 8MP IR camera and 4x IR and 4x White LEDs

Birch log box with camera insert
I made two changes with this design over the Robin box version of this

Top and bottom view

LEDs are added via 2x perfboards on the underside, instead of fiddly, messy wiring on the top. The 12v to 5v conversion is done via a recom switching regulator (R-78B5.0-1.5).  I had originally planned to supply this box with a 5v feed but this didnt work well, and had to supply at a higher voltage (12v) and then drop it.  The switching regulator component alone is an equivalent cost of the Raspberry Pi Zero-W, so I've used it a reusable way rather than soldering it directly in to the stripboard.

Bodge-tastic use of PCB terminal blocks...

As previously, the unoccupied left side slot is wired for a L293DNE to switch an IR-cut filter (to be added at some point later, for a how-to see this post).

The finished product


Some inside views

Top: Infrared mode, Bottom: Visible light mode
The visible light option suffers from the lack of an IR cut filter, but one can be added once nesting season is over...

and in-situ



Monday, 2 April 2018

WiFi Robin Nest Box

I've planned a 'hub and spoke' arrangement of bird boxes.  I previously described the 'Wireless Access point' box (aka 'WAP box'), this post details a 'spoke/satellite' box.

Robin box in the undergrowth
Robin box with Great tit...

This is a ground-level open-fronted box designed with robins in mind.  Last year, a robin family successfully raised a brood of three chicks about a meter away from this new box at ground level.  That nesting site has since disintegrated, so this is hoped to be come its successor.

The camera setup is designed to be generic so that I can remove it and swop it into other boxes designed with similar dimensions, in that way I can maximise the number of boxes, while not having to equip every one with a camera.


There is a ledge at the top that holds the camera section.  This consists of a Raspberry pi Zero-W, an 8 megapixel v2 'Pi-Noir' camera (sensitive to IR as well as visible light).  Underneath are 4x each of white and IR LEDs for illumination.  Power is via a 12v feed from the WAP-box, dropped to 5v via a recom r-78b5.0 1.5 switching regulator.  LEDs are powered form the same 5v source, and switched via a transistor array (ULN2003AN), on the R side of the picture below.  This allows low, medium and high setting on each of the IR and visible light, white LEDs.  The unoccupied left side slot is wired for a L293DNE to switch an IR-cut filter (to be added at some point later, for a how-to see this post).


The white LEDs are wired in parallel with a 50ohm resistor for each LED.  The IR ones are also in parallel, but with no extra resistors.


The circuitry is arranged on a commercial prototyping board designed with the same dimensions as the Pi zero-W.  The board layout is as follows:

Protoboard layout, same footprint as Pi Zero/W
I also added pins for a DS3231 real-time clock module, since the Pi is only able to get time from the network, and wanted to plan for occasions when it loses network connection (there's not enough headroom in the box to fit them on the actual GPIO pins).
As with most of my other boxes, video capture is via motion capture software pikrellcam with captured footage/images saved to a mounted network drive (i.e. off the box).

This clip shows the box in situ cycling through various illumination settings:


A couple of observations from the transition from 'testing to the real world...'
Pi Zero-W wifi reception once closed into a wooden box drops off somewhat.  Running the following command gives some idea of reception levels:  iwconfig wlan0 | grep -i --color quality
Output is: Link Quality=44/70  Signal level=-66 dBm

The impact of this is that I cant set PikrellCam to record video captures direct to the network drive, as it all keels over.  Handily there is an archive option, so I just need to keep an eye on it and archive to the network when need be.  I also dropped the quality of the live stream which seems to have stabilised it. 


UPDATE: A updated implementation of this kit is detailed, in 'Birch Log' nest box described here

Thursday, 15 March 2018

Wireless access point bird box

My current nestbox cameras are mostly raspberry pi-based with either a webcam or raspberry pi (RasPi) camera module to record and/or to stream digital video to a web-browser or CCTV software on a separate PC.  Power is via 'Power over ethernet' (PoE) where one ethernet cable provides power and a means to transmit the digital video.  Cable runs are approx 50m and 80m.

Using this setup, I'm limited to one camera per cable run:


Solution: The Wireless Point access bird box or 'WAP-Box'

I've taken an existing raspberry Pi (Raspberry Pi 2, model B)-based bird box, added a wifi dongle and set it up as a wifi access point (as per these instructions).  This is the 'WAP-Box', which helpfully can still be used to run its camera as well as being a wireless access point for other nearby devices.


In this setup, 48v PoE ethernet cable goes to the base of the tree.  At the tree base is a junction box where the cable is split out to ethernet-only and power, dropped to 5v using a TP-Link PoE splitter.  The 5V output is then used to power a  Mini USB hub using a  female USB to female barrel socket adapter, and the barrel connector supplied with the TP-link PoE splitter.  The usb hub can be used to power up to 4 devices. (note that this PoE kit also includes the power injector used in the house).
So, this didn't work well.  I ended up dropping the 48v feed to 12v, an splitting this out to a 'junction box' at the bottom of the tree as I didnt fancy climbing a tree each time I wanted to  connect/disconnect a device....  12v feeds are ...

1) Recombined with ethernet for wired, PoE and sent to the 'WAP-box' which is up a tree.  This a based around a wired ethernet Raspberry pi 2B with wifi dongle.  The 12 v PoE feed powers the Raspberry Pi another TP-link PoE splitter with voltage dropped to 5v   The power cable was hacked from micro USB cable (keep the micro end to be used later) and a DC barrel connector.  I couldnt find a link for the PoE injector I used, but it came with a print server ages ago and had been in the 'may come in useful' drawer.



2) The additional WiFi 'Satellite NestBoxes':
Birch log box small tit box described here
Robin box described here

These take a 5v 12v feed via cables made from a length of black 2-core bell wire and the same microusb and DC barrel connectors linked to above.  The cables power a Raspberry Pi Zero-W + Raspberry Pi v2 IR camera module combo in each box which will have a couple of IR and visible LEDs to for illumination. The satellite boxes are pre-configured to connect to the WAP-box, and thus can save any captured video direct to my home network, so I dont need to worry about the little Raspi ZeroW's running out of space on their SD cards.
This didnt work well due to lag / poor network connection.  In the end I ended up writing video locally to the PiZero's SD cards, and archiving off daily to the local network, which can all be automated.

This all currently works on the bench with 4 concurrently running camera setups using PikrellCam software, and will be going up in the next couple of weeks once their various boxes are completed.  In testing, the Pi Zero-W can 'see' the WAP-box up to about 30m away - my main concern is that if the 12v cable runs are too long the voltage drop may lead to instability of the Pi Zero-W's, so that is likely to limit the cable run distance rather than the strength of the WiFi signal.  I also need to have a think about the current load on the 48v power injector at the house end...

More to follow once some boxes are up...



Of note is that the Raspbery Pi foundation recently released a new version of the highest performing version, called the Model 3B+.  This is interesting this context as it could potentially make a better WAP-box core since it has a faster ethernet connection.  Currently, simultaneously live streaming 4 Raspi-zeros via the existing Pi2 Model B is a bit choppy, which may be due to a bottleneck at the WAP-box's ethernet connection... or this could just be me using this as an excuse to play with one...

Update:
Satellite box 1, 'Robin Box' described in this post
Satellite box 2 'Birch log box' described in this post