Radio Mobile simulation

CanterburyMedia Site to Site Link

One of the unique challenges of CSR FM is it’s structure, where it is part funded by two separate universities each with a presence in Canterbury (University of Kent and Christ Church University) and their respective student unions (Kent Union and CCCU).  Student members could be enrolled at either institution, and as such each institution has it’s own radio studio on campus – each of which has an equal chance of being put on air.

This has presented a challenge, in that live audio needs to make it’s way from either studio to the transmitter, hosted at the University of  Kent in Eliot College.  From the outset of the CSR project, the route between the two has always been over IP – there was no other reasonable option.  Up until recently, that was realised using a IP Codec – with it’s packets being routed over the institution’s networks.

This is perfectly fine, and for the most part worked very well.  However, recently CSR has had a fundamental infrastructure change to it’s audio distribution (as part of the Student Media Center project), which has made every single audio source and destination available using Axia’s Livewire Audio over IP protocol over their internal network.  This allows fantastic flexibility and allowing studios to route any source to their mixing consoles, as well as increased interoperability with our automation systems and customisable GPIO control.

However, the second studio (Studio Blue) was still connected via an IP codec link, which was not integrated into the system at all and offered limited capacity for routing over the link (only a stereo pair to and from the router).  Unfortunately the link between the two sites, over the academic networks that link the universities, would not be suitable to transport Livewire (for a number of very good reasons, lack of multicast and custom QoS being one).  It was therefore necessary that we provided a Layer 2 link between the two locations to carry this traffic, which we had complete control over.

Traditionally, you might use a leased line type circuit (ATM or Ethernet, or MPLS) to achieve this sort of connectivity.  But these options are vastly outside the realms of reasonable cost of installation and setup for the project.  So an alternative method was required.

Ubiquiti have been offering a range of very capable, and disruptively priced, Ethernet Radios for some time now.  CSR has even used these radios in the past to provide connectivity to remote locations for Outside Broadcasts, with great success.  So it was a natural step to make a permanent installation to link our two, reasonably geographically close, studios together.

5GHz link Network Diagram

The link consists of two Ubiquiti Nanobridge M5 radios, each affixed to the roofs of the respective host buildings.  We can configure them to transport VLAN trunks, allowing us to not only transport Livewire streams, but also segregated control and other data over the link.  The appropriate QoS settings in the radios and the Cisco switches at each facility ensures that Livewire data is given priority, so that real time audio is possible and reliable.

In order to prove the link was possible, we ran some simulations in Radio Mobile using the specifications of the radios to get an idea of whether the signal would be good in the area we needed.  Turns out we should get a pretty good signal between the two sites!  The simulation was run with an omnidirectional antenna, assuming that we can rotate it as required – the below plot shows areas where a realiable connection is feasible.

Radio Mobile simulation


In the past we had performed a link analysis also for the specific points – not dissimilar to one which we performed for a link to Merton Farm some years ago (pictured below)



When we ran the link analysis for this particular link, it showed that there was some “land clutter” in the way (trees, cathedrals, that sort of thing), however providing we had enough height a link should be possible.

The first radio was installed on an old TV aerial pole on Eliot College roof.  There are two CAT5e runs from the roof to the CSR FM’s comms room in the basement, which we used to connect to the equipment.  Care was taken to ensure that the equipment is not obviously visible from the ground.




The second was installed on the roof of the Erasmus building, at Christ Church University, on an existing pole for our FM aerial.  This turns out to be a very neat installation.



After alignment, we are able to get a link, over the 2km between the studios, at 180Mbps – at around -60dBm.  The wireless link is faster than the wired Ethernet down link at 100Mbps!

Hopefully this should pave the way for a better connected studio for CSR at Christ Church University

Keynestock 2014

Once again, I got roped into helping out CSRfm and this time KTV in getting their OBs from Keynstock 2014 on air.

There were some not insignficant challenges. Our normal network access at Keynes was effectively cut off due to some changes to configuration.  This was quite troublesome, as we had always previously relied on this access to get our signals back to our HQ.  The outlook seemed bleak.  What we ended up doing instead, though, actually seemed to work out rather better.

We ended up creating a private “Media” network in the new Student Media Centre, and extending that from The Venue to Keynes via a 5GHz wifi point to point link (with some Ubiquiti Nanobridge M5′s that we happened to have lying around).  Once that landed we flew CAT5e to the various rooms to get their data into the network.


As for the link, CSR’s audio was sent via OpenOB with an Opus link running at 192kbps –  running on two Raspberry Pi’s each with Wolfson Audio cards.  We ended up using Wolfson’s image with precompiled kernel and updating Raspbian to get the latest bits.  However the link worked fantastically, sounded great (once we ironed out the level problems at the start) and didn’t audibly drop very much at all.

KTV were using a Teradek Cube to stream their video back to their gallery in the student media centre where a PC was set up to spit out the stream from it into their vision mixer.  All of this was going over the wifi link between the buildings.




I’ve been helping with the installation of a brand new Axia Livewire network at CSR FM. The network is a bit different to the usual installation – and that deserves it’s own post.  We’ve been using PathfinderPC to do all of the routing control.

It’s all pretty clever stuff – but we wanted to be able to extract information from Pathfinder so that we could do handy things like find out which studio is on air, and use that information to show the right webcam on the website, or use it to add further information to our Now Playing data.  Pathfinder has a way of doing this – using Protocol Translators.  Basically, it’s a TCP listener (or client, or Serial Port) which accepts and sends commands to a remote device.  The protocol is very well documented in the manual, and is very flexible in what it lets you do.

But, it’s a bit of a pain to connect to from, lets say, PHP – which isn’t really well suited to doing socket operations.  Also, you’d want to cache the results somehow, lest poor Pathfinder get inundated by people looking at the webcam!


So I decided to make a little thingy that sits in between Pathfinder and the potentially unruly web apps on the other end.

Thus, PFInterfaceWeb.  It exposes various data sources over HTTP – like a list of all source, destinations and current routes.  Also – and particularly usefully – memory slots!  This lets us query our Stack Event logic and work out who’s on air, what mics are live…. etc etc.

Oh, also, it optionally can send messages to a STOMP compatible Message Queue server whenever a memory slot, or route, changes, or custom Protocol Translator commands are sent.

At some point, I’ll make it monitor GPIO, and also Silence Detect events.

Hopefully, this will prove handy!


Betheremin 0.1

My partner, Beth, asked me if I could make her a Theremin.  So I have.  It’s called the Betheremin.

A Theremin is a musical instrument, which changes pitch and/or volume as you bring your hand close to it’s antenna(e).  The way this works is by your hand influencing the capacitance of a resonator circuit, changing the frequency at which it oscillates.  This difference in frequency creates a “beat” frequency against a reference oscillator, which can then be used to create an audible frequency or control a Voltage Controlled Amplifier.

This is a fairly complicated thing to build – and I didn’t have an awful lot of time, so I went for a more basic design which controls pitch only.  I borrowed the circuit from Harrison Instruments “Minimum Theremin” which I have used before (from a kit).  It’s a clever design which uses CMOS logic inverters to run the oscillators, and provides adjustment by tweaking the supply voltage.  There’s a full explanation of how the circuit works on the Harrison Instruments website.

The first step for me was to transcribe the circuit into CAD format, and at the same time picking appropriate parts to do the job.  In a bit of a change from my usual, I decided to source all the parts from Rapid Electronics.  It ended up being quite a bit cheaper in the end than Farnell or Mouser, since the components are all pretty common and uninteresting!  The BOM ended up looking like this:

Designation Description Part Order Code Qty
Batt PP3 battery clip 1 X PP3 PCB Battery Holder 18-2990 1
C1, C9 100pF 100pf 2.5mm Npo Ceramic Capacitor 08-0940 2
C2, C4 0.01uF 10nf 2.5mm X7r Dielect Ceramic Capacitor 08-1000 2
C3 0.1uF 100nf 2.5mm Y5v Dielect Ceramic Capacitor 08-0275 1
C5, C6, C7, C8 10uF pol Forever 85°C 10U 25V Radial Lead Aluminium Electrolytic Capacitor 11-0220 4
CR1, CR3 1N4148 1N4148 Silicon Diode 150mA 75V DO-35 51-0100 2
CR2 1N4004 1n4004 1a 400v Silicon Rectifier Diode 47-3136 1
Enclosure Hammond ABS Enclosure Multipurpose Translucent Blue (121 x 66 x 40mm) 30-3779 1
JACK 6.35mm jack 2 Pole PCB Standard Jack Socket 20-1390
ONLED1, ONLED2 White LEDs 3mm Warm White LED 750mcd 55-2228 2
PITCH-CTRL 1K Pot 1K 16mm Linear Potentiometer 65-0705 1
R1, R11, R4 270k 270K 0805 1% Chip Resistor – Pack of 100 72-1025 3
R3, R12, R5 22k 22K 0805 1% Chip Resistor – Pack of 100 72-0907 3
R2, R8,R10, R9 27k 27K 0805 1% Chip Resistor – Pack of 100 72-0917 4
R6 100R 100R 0805 1% Chip Resistor – Pack of 100 72-0590 1
R7 470R 470R 0805 1% Chip Resistor – Pack of 100 72-0667 1
RV1 10k preset pot 3306W-1-103 10K ±25% 6 mm Bourns Vertical Cermet Trimmer Potentiometer 67-0642 1
PCB screws Pozi Countersunk Self-Tapping Screws No.4 9.5mm Pack Of 100 33-3415 1
SW1 Power switch Spdt R/a PCB Slide Switch 76-0271 1
U1,U2 CD4069UBE 4069ub Hex Inverters 83-0398 2
US1 US2 DIL socket Tube of 34 14pin DIL Socket, Narrow7.62mm Without Central Support 22-0108 1
VR1 LP2950ACZ-5.0 Lp2950cz-5 Micropower Regulator. 82-0680 1
KNOB 16mm knob Re’an 16mm Soft Touch Knob with White Pointer 32-0470 1
ANTENNA FASTENING Spacer – Nickel Plated Brass – 5.5mm AF Female To Female- M3 x 10.0mm – Single 49-0747 1
ANTENNA FASTENING Pozi Pan Head Machine Screws BZP M3 6mm Pack Of 100 33-2300 1
ANTENNA FASTENING Steel Washers BZP M3 Pack Of 100 33-1760 1
PCB SPACERS Pack 25 6mm Nylon Spacers 33-3634 1

Total cost, £18.04.

I decided to go for 0805 resistors, simply because it saved on space and they were cheaper (as well as having some in stock already).

The enclosure is a nice translucent blue Hammond box, which lit up with the white LEDs will look awesome.  I designed the PCB so it’d fit into the box and use the mounting posts already provided.

I decided that I wanted all of the parts of this project to be mounted onto the PCB – including all of the connectors, switches, pots and battery boxes.  The PCB would be mounted to the lid of the box, rather than inside the body, using nylon standoffs to give enough space between the PCB and the lid to fit the PITCH ZERO potentiometer.

For mounting the antenna, I used a hex profile spacer and a machine head screw, sandwiching the pad and providing a screw thread flush with the lid of the box (after drilling a hole).  The antenna I got from eBay, with a M3 standard thread.  This part is the main problematic part to source – I’ve yet to find someone who sells telescopic antennae with screw thread ends in any volume, or any “standard” antenna.  I guess there’s not so much of a need for these these days – perhaps a better solution would be a “rubber duck” antenna with SMA connector or similar – but those connectors are quite expensive and I wouldn’t be taking advantage of the benefits of coaxial barrels.

After many hours of careful routing, tweaking and massaging, I had a PCB ready to go:


There’s a power slide switch at the top side, which looks like it’s going to interfere with the battery box!  But, I thought, why not put that on the other side of the board, since it’ll work exactly the same.  Remember that ….

The “bottom” side of the board, is actually the side that’s going to be facing upwards in the box.  This means all the components are on the open side of the board, except the LEDs and the potentiometer.  I didn’t want there to be silkscreen on the side facing upwards, so I put the markings (the “Betheremin” text) in the copper layer, and peeled back the solder mask – which once the board has been through the gold flashing process gives a nice gold text :)

So, PCB design, I needed to get them produced.  I had a few options, but I had previously used BatchPCB to great success.  Thus, I decided to try out it’s successor, OSHPark.  They accept Eagle files (no faffing around with Gerbers, yay!) and they give you a great automated preview when you upload your design files.  I was kept informed throughout the process, from sending them to the fab house to sending the boards back to me.  The price was very reasonable, and considering the boards were shipped to me in the UK from the USA, came pretty quickly.  Also the packaging, like the PCBs, is awesomely purple :) I will definitely be using them again!

This is what I was shipped from OSHPark:


Gorgeous!  I’m very pleased with the quality.  The routing is very smooth, the soldermask is very consistent. The registration on the silkscreen was a little off, but that’s to be expected really.

After a little filing around the edges to make up for the snap sprues, and the slightly tight tolerances around the curved corners, it fit into the box perfectly!

Some soldering later:


In this side shot, you can see how the hex-spacer works to let you screw the antenna into the lid, and where the pot is attached to the board on it’s back.



The pot was secured to the PCB using normal self adhesive pads – mainly just to stop it putting strain on the pads.  The shaft on the pot could do with being a little shorter – so that the knob doesn’t float quite so far away from the panel.

Note the slide switch.  I put it on the wrong side … doh!  It interfered with the battery box a little, which – rather than putting the switch on the other side of the board – I filed the battery box a little to make it fit.  Doh!

The LEDs point up the board, allowing the solder connections to reflect the light to produce an interesting pattern when the device is on.


Overall, it looks like this:


After some fairly tweaky calibration you can control the pitch from about 12 inches away.  If you are grounded well, then this works even better and you get more range.  It produces a mostly sawy sine wave (pretty much an RC oscillator sort of curve really, not surprisingly).  I did try to put a RC low pass filter on the output (by putting a 1uF cap between the output and ground) to see if that’d make the output more siney – but unfortunately it did attenuate the output rather significantly at the frequencies we were interested in (between 0 and 1500Hz ish).


In REV 0.2, if I make another one, some things to note:

  • Put the slide switch on the right side :p
  • Make RV1, the calibration trim pot, multiturn as the single turn one was very tweaky to adjust
  • Make the thermals thinner and the pads bigger (desoldering is a pain otherwise!)
  • Increase the spacing between the 0805 resistors
  • Don’t put anything under the battery box :)
  • Make the holes to mount the battery box big enough to clear a No 3. self tapping screw, so the box can be secured to the PCB properly.
  • Increase the tolerances on the PCB borders so it’ll fit without needing filing!

Future enhancements:

  • Put a proper amplifier on the output, with gain control
  • Provide a volume control
  • Break out PWM output to control other things.
  • Generate MIDI/OSC messages over USB?!

Overall this was quite a fun little project, and something I’ve not done for quite some time!  And immensely satisfying that not only is it complete and works well, but also that it’s going to be used productively!

Hello, World

Good day, fine visitor-person.

This is a brand new blog, and much like a new notebook I will write some information about me on this first post.

My name is Chris Roberts, I am a Broadcast Engineer by trade – but have had an interest in Technology since forever.  I’ve done a degree in Computer Systems Engineering at the University of Kent, and I, as of this writing, work for a “large international broadcaster” in London.

I have a history of being overly attached to whiteboards, cakes and lab coats.

That’s pretty much it for now! Expect a new post soon about whatever project I happen to be daydreaming about at the time.

Chris Roberts MEng. Broadcast Engineer