All posts by EEadmin

Farnsworth (Warehouse 13 artifact) – Continued

From the last post a lot has changed.

I’ve resized the cases and moved some of the aesthetics around to get every thing to fit in the case.

The display now gives connection information and displays status connected/not connected, and has the ability of displaying graphics with this information.

The software has had a lot of additions, I can now read from the SD card a set of destinations and select them using the rotary encoder wheel on the front of the case.

A destination is defined as an IP address, port and channel. The Channel is used by the UDP repeater to join Farnsworths together (

The connection has gone from just Farnsworth to Farnsworth, to connection options using a central UDP proxy, ( hopefully with the aim of getting communication out from only being in the local network. This has also added various testing additions along the way including a test card and a loopback mode, both locally in the Farnsworth and to/from the proxy for remote testing.

If you have multiple Farnsworths on a local network, they need to be set to different ports so NAT can find them for the incoming channel.

Audio is proving to be a pain. The Pico’s A2D is very noisy, and the simple amp I put on the existing PCB isn’t up to the task to overcome this. I’ve replaced the single transistor amp with a low voltage op-amp that will hopefully have a lower output impedance into the ADC.

The other issue is ensuring that the 8khz sampling and playback happens regularly regardless of the camera’s capture and incoming UDP connections.

From comments from Mastodon, the two sound clips from the original Farnsworth have been incorporated, with the “text” tone used to indicate that the Farnworth has connected to WIFI, and the “RING” tone used to indicate that a Farnsworth to Farnsworth connection has been made.

Finally got it all into the Case. This one is a Woodroffe, for obvious reasons for those that know me. When I get time ill update the older Farnsworth to the same case standard.

Just so happy I have two and they can talk to each other.

Code circuit diagram audio files and graphics are available on Github

Previous Article

Farnsworth (Warehouse 13) artifact creation

So for reasons I’m not entirely sure of myself I decided I was going to build a Farnsworth.

For those of you who are not fans of Warehouse 13, as Farnsworth is a Steampunkesque hand held video and audio communication device, named after Philo Farnsworth the inventor of electronic Television.

A picture of one from the TV series is viewable on the Warehouse13 fandom site here

Of course this is only a prop, me, being me wanted one that worked.

“The Farnsworth is a two-way audio and video communications device invented by Philo Farnsworth in 1929, soon after the invention of the television. They’re used by Warehouse agents because they are on their own secure frequency spectrum and can’t be cracked, hacked, tapped, or otherwise “broken.” However, Artie stated in Season 5 that it could potentially be vulnerable to hacks if Claudia messed with it.”

My Farnsworth, will be totally hackable, insecure and very easily broken, but working. (hopefully)

Of course I could have just put a phone in a box and extended the display, microphone etc. but where is the fun in that ?

I’ve been playing with Pi Picos a lot recently, and Id not really explored the Pi PicoW and its networking abilities.

So all I need to do is to attach a camera, a display microphone and speaker and I’d be ready to go. Oh, and I didn’t want it to cost me an arm and a leg.

Looking in my spares box I had a cheap camera the OV7670, the problem with this device is, although cheap, it uses 14 IO, this would severly limit what else I could do with the PICOW. Luckily UsedBytes( Brian Starkey) had done a pico project for robot vision using this camera, for this he had used a parallel to serial shift register to drop the 8 data lines to a single line and the PIco’s PIO to reconstruct the data. The downside was his library only ran at 80×80 pixels in mono.

After quite a lot of changing parameters in his library I got it to work dropping a 240×320 colour (RGB565) image into the Picos ram.

The Display was much easier, the GC9A01 35mm circular display is much more common, and as all I needed was to be able to write bitmaps to the display, I soon had cut down the available librarys and had a DMA running to drop the camera image to the display.

As I only wanted 240×240 on the display, I split the process into lines, rather than moving the whole buffer at once.

This also meant that I could send this data easily to a UDP Socket on the PicoW. All I do is add the line number to the beginning 240 RGB565 integers, and send this as a UDP packet.

At the receiving end, I strip off the first byte and dump the rest directly into the display. If a line is missing or corrupt (this is UDP) it gets over written in 1/5 second anyway.

This scheme also left me with 241-255 as a starting byte I could use for other purposes. Sound being one of these.

Sound is via a single transistor amp (probably needs to be a better amplifier) into the pico’s A2D sampled at 8Khz

So sound is given a starting byte of 250 and 10mS of sound is sent in the same way a line is sent every 10mS

At the other end the sound packet is unpacked and pushed into a FIFO buffer so it plays for 10mS, if the buffer empties then silence is played.

The Case outer is 3d printed from black PLA, and the two front plates are made from black acrylic engraved on a laser cutter with infill of acrylic paint.

Loads still to do, Ill update here as the project progresses.

The code is available on Github

Continued here…

The Gaussfest 2023 badge

How not to make a badge for an event.

Every year I make a badge for the Nottingham Gaussfest

Previous years I have just done a simple laser cut badge with a battery powered flashing LED, but this year I thought I’d do something a bit different.

My first idea was pushing the boat out a little too far, and after receiving the PCB’s I realised it would never work ( I don’t want to say what it was, as I have ideas for sort it in coming years)

So, as I was losing time quickly, I decided that I’d build a simple lightning detector. At the Gaussfest there are many, many machines that make sparks. So the idea was, when the exhibitors are making sparks, their badge lights up, what could be easier.

If I was going to do this, the badges MUST be there and working on the day. No point in having a lightning detector badge on a non-teslacoiling day.

I roughed up a circuit, based on Extreme kits lightning detector. To keep the LED on for a few seconds I used an NE555 timer.

The circuit was tested on bread-board and if I used a CMOS NE555 would run really well with a 3V supply. (Non CMOS 555’s need 6v AND take a lot more current)

I wanted to use through-hole as Extreme Kits’s lightning detector circuit used almost all of the same components but, I didn’t want ugly through hole wires on the front of the PCB, so I opted for through hole components, but surface mounted.

Happy with the circuit, I edited the PCB layout so I could get two boards in a 10x10CM order, which is the most cost efficient way of getting these PCB’s made with my supplier.

I also ordered 50 NE555 (CMOS) as the best prices came from china the PCB’s and the chips should arrive at roughly the same time.

After a couple of weeks the PCB’s and chips arrived and I quickly built one up and put in a CR2032 3v battery

It didn’t work, nothing…

Nothing I did would make it work, but everything seamed fine. Eventually I put it onto my PSU and tried it at 3V from there.


Scoping the output, it worked, but wouldn’t drive the LEDS, I tried every variation of LEDS I had, nothing worked (apart from a VERY dull glow).

On a hunch, I tried the circuit at 6V

Everything worked fine. Eh ???

I then occurred to me that my CMOS NE555’s (from China) may not be CMOS, I swapped one for a lone known good chip I had, the circuit worked fine. (insert swearwords here)

With not enough time to reorder CMOS 555’s, I needed a solution.

Well, CR2032’s are 3v and 3.2mm thick, CR2016’s are 3v and 16mm thick, I could fit two 2016’s in a 2032 socket giving me 6V. This worked, but the bottom battery would short. A small piece of tape sorted that quickly. The only down side now is that 2×2016’s have half the current capacity of 1×2032 and the non CMOS 555 chips are a lot more greedy. In a trial, the badges lasted 14 hours. I only needed them to last 8. I was hoping for a few days where I could put in the batteries and pack the badges. Adding a pull away tab which could turn them on, on the morning of the event sorted that. Finally I had a solution.

So, I built a batch of the PCB’s

Half of them worked, the other half refused to do anything. (insert other swearwords here)

On closer inspection the second PCB on each panel wouldn’t work.

Referring back to the PCB design I spotted my error, The fills would against ground, as I’d copied the PCB the signals for GND were no longer GND so the fills avoided the pads. All the GND connections were not connected. I didn’t have time to reorder the PCB’s

Luckily, my PCB house had done a couple of extra PCB’s, so I had JUST enough. The two organiser badges though, had a pretty addition of gold wires on the back joining all the GNDS together.

Finally a box of working badges, with only a few days to go. The only problem was I ran out of time, and I never got to test the badges with a real tesla coil or Wimshurst.

Luckily on the day they performed pretty well, If I was going to be harsh, they triggered occasionally on static discharges with peoples clothes, and their range was poor when they were close to the chest, but 1-2M away from my and other tesla coils they lit up wonderfully.

A badge being triggered by a discharge from my Shake sphere (that’s the background noise)

So my tips for designing a badge for an event.

  • Start early.
  • Nope, earlier.
  • Order the components and PCB’s and test them all, while you still have time to reorder.
  • Have a backup plan.
  • it will cost more then you think, budget.
  • Put a power switch on your badge. They are cheaper in time than messing with pull out tabs or populating batteries.
  • If something goes wrong, don’t panic, there are many people out there who may be able to help.

Nottingham, Gaussfest 2023

Another great day at Papplewick Pumping Station playing with all things high voltage. A very eclectic collection of high voltage exhibits again this year, and a big thank you to the exhibitors and visitors that literally came from all over the world to this event.

And of course a huge thank you to Papplewick Pumping Station and their volunteers for lending us their steam hall, boiler room and coal shelter for the day.

Here is a few photos and videos, Ill post more as I get them.

Pico Lynx

A Pico Emulated Camputers Lynx

The Camputers Lynx was a classic, although not massively popular computer in the early 1980’s. It featured a z80 at 6Mhz and full colour graphics, with none of the BBC micro’s (or others) Graphics mode shenanigans. And for its time many advanced capabilities like 96K memory, a disk system (later upgradable to CP/M 80), and a proper keyboard. Unfortunately it hit at a bad time, the Spectrum was much cheaper, but much less capable, and the BBC was being pushed massively by the BBC (unsurprisingly). It’s eventual down fall was that its brilliant graphics were both quite slow and due to its strange memory banking system, rather hard to program. Eventually it lost out due to the lack of game titles compared to the Spectrum and BBC Micro/Electron.

But I had one and I loved it (latterly I had three)

Camputers Lynx (By Retro-activity, CC BY-SA 3.0,

I already have a PCB that Ive used for Pico Emulation of a number if things, so that was my go to as a start.

I had asked for permission to use the code by Charles Peter Debenham Todd retrogubbins (as there is no licence on his GitHub pages) and he graciously let me use his code with no restrictions.

I based my Pico design on his PALE emulator and his ESP 32 emulator

One of the first issues I had was that the Z80 emulation that he used did not have any licence either, so early on I swapped it to LIBZ80 © Gabriel Gambetta ( 2000 – 2014 which is currently maintained by EtchedPixels as part of the z80 emulation kit

I’ve used this library before in the RC2040 So familiarity was on my side too.

Swapping over the emulator and porting the ESP32 code to the Pico was amazingly straight forward, I needed to write some code to DMA out the RGB banks to a display (240×320 OLED) and I quickly got a Lynx prompt.

Unfortunately it was in Yellow, and the emulation stopped there. I took me a while to get past this, until I realised that the Lynx Beeps, between writing the Red and Green banks, and before writing Blue. I hadn’t implemented the code for the sound / speaker.

Adding this in gave me not only a while logo, but two power up beeps, and (after spanering in serial to the keyboard emulation ) a workable basic.

So time to make a case.

I took the measurements from a real Lynx and scaled it to roughly fit the keyboard I had with the PicoPuter. It came out as roughly 1:3

Of course Not everything fitted correctly, so after a few goes.

I got one that fitted and held the PCB’s in roughly the right place.

Of course I needed a monitor, so I did a quick 3D print of an IBM 5151 style monitor. This gave me the space to the right to hide the display connector. and as the original was only 12″ didn’t look out of place with the small Lynx Case.

For the keyboard I wanted actual working keys.

So I made an escutcheon from grey sprayed card

and laser cut/engraved a set of small keys to fit. Because I couldn’t get grey acrylic I used transparent acrylic, on top of a grey sprayed key holder.

The effect is grey keys with white painted idents.

Not 100% happy with these, but I have a set for working grey keys, on a grey Lynx shaped case. For a more accurate layout I’d need a custom Lynx keyboard PCB.

The whole case is only 120mm wide to give you some idea of scale.

The two PicoPuter PCB’s sit inside and a small battery, because it was vital for this to run as a portable 🙂

The Ribbon cable is the feed to the display, which is mostly hidden by the case and the display itself.

and attached to the back of the display by a small daughter board.

Notts Gaussfest – What is a Gaussfest ?

          What is a Gaussfest ?

Nottingham Gaussfest is a yearly meeting of a loose group of friends sharing a common interest in high voltage equipment and Tesla coils.In the USA these meetings are always referred to as Teslathons, a reference to Nicola Tesla who’s tesla coils always feature prominently at these events. We decided, being European, we would adopt Carl Friedrich Gauss as the scientist, due to all the devices also required a high magnetic field and we wanted to differentiate ourselves. Hence the term Gaussfest.

The equipment that appears at the Gaussfest is entirely dependent on the Exhibitors present, and their current interests. Usually there will be a collection of various types of Tesla coils, and a collection of static electricity generators like Van-De Graff machines and Wimshurst machines. But each year someone surprises me with something different, or new.

Tesla coils are the most common high voltage device we get, are essentially a large transformer. Near the bottom of the tesla coil is a radio frequency generator either electronic or mechanical, driving a primary winding with a small number of turns. The Secondary coil (the tall one) has a much larger number of turns of around 800-2000 turns typically. And at the top if this, a topload where the high voltage comes out. We get tesla coils of all sizes from a 50mm high, to over 2M tall at the Gaussfest.

The Static generating machines our exhibitors bring are even more are varied. The common ones are a Van de Graaff machine, uses a belt to transfer charge up to the top dome where it is stored, and the Wimshurst machine uses a pair of contra-rotating disks to multiply charge and store it into a pair of capacitors. Again the sizes of these machines vary from 200mm tall to 1.5MAlthough the electrical discharges from these machines look dangerous, they are entirely safe if viewed from a distance, and our exhibitors are very experienced with the operation of their machines.

Or Exhibitors are happy to talk about how their machines work and the problems with their construction, although if they are preparing them for a demonstration, or have problems please give them “thinking room” as their safety, and the safety of their equipment is paramount. All our exhibitors are all amateurs and make and demonstrate high voltage equipment for fun. The Gaussfest is their opportunity to display what they have been working on.

Commercial high voltage engineering is expensive, due to this many of the exhibits on display will have been made totally or partially from household items, surplus, and any available electronic equipment, often with items multiplied up to deal with the high voltages/currents that are required. To keep the cost down, the equipment is usually run at the very maximum of its tolerances. These devices are rarely engineered for their longevity, so failures are quite common, and this trade-off is very much part of the hobby. Due to this, we do not have planned schedules for the demonstrations, as equipment WILL FAIL (and be repaired) during the day, although we will endeavour to have something always happening.

When the Gaussfest moved to Papplewick in 2017, we were very conscious that for us to do our activities, Papplewick pumping station must get something back, So Papplewick allow us to meet, and we let the public see what we do and give demonstrations to give back to the Pumping station. As much of what we do can be dangerous, this trade-off is some restrictions in access to the station to ensure your safety.  

Nottingham Gaussfest

Gaussfest registration

Shake Sphere

A Hand held Electrostatic generator by A D Moore

In his book “Electrostatics; exploring, controlling, and using static electricity” A D Moore describes a simple electrostatic generator described as a shake sphere. Very little other information is given.

Luckily A.C.M. de Queiroz gives a simple description and picture of a built one here the operation is copied below.

Because I can never follow instructions and I always want to make my life harder, I deided to try to build a folded version. From the above it is reasonable to see that the earthed block in the centre is common so can be split, as long as its electrically re-joined.

This allows the inductors and collectors to be stacked next to each other rather than have a couple of trailing wires, and as the block is earth, then the wire joining these two blocks will be earth, much more efficient than two wires with high voltage. So I designed this.

The Collectors were 3D printed, then foiled and burnished to make then conductive. The two tubes are 200mm long and made from 25mm (OD) PVC electrical conduit. The two areas where the inductors would sit were taped. and 2 20mm steel balls were used as the charge carriers.

This was constructed and a simple test, with low humidity we could get sparks from the collectors after 30-40 rocks forward and back from the ball bearings.

For ease of use, a handle was added and a couple of insulators to keep the earthed wire away from the charge collectors.

This will produce 10mm sparks on a good day, and is an interesting demo, but practically useless for anything else

Pico Cray – Small scale distributed computing.

The Pico Cray is a cluster of 8 Picos and a controller Pico. All Picos are connected in parallel by an I2C bus. The controller is designated by one GPIO being held low, the processor Pico’s addresses are allocated automatically. The Controller Pico has a 320x240px display connected by SPI and a couple of buttons. As all of the Picos run the same software updating the cluster is easy.

Currently the cluster is programmed to run a Mandelbrot program. With each processor Pico calculating 120 points before returning them to the Controller.

Pico Cray

More RC2040 settings

The INI file

The ini file sets the defaults for the Rc2040 and substitutes for switches and buttons on boards that have none.

It allows you to select the ROM, RAM settings, and CPM settings for the emulation.

Using SIO based software

Requires CPM Inc Transient Apps SIO2.img,, 24886009.BIN and rc2040.ini for SIO2 as below

Rom Details

ROM has basic at address 0x0000 (000) ROMsize 0x2000
ROM has basic at address 0x2000 (001) ROMsize 0x2000
CP/M / Basic via monitor at address 0x4000 (010) ROMsize 0x4000
Small Computer Monitor at 0xe000 (111) ROMsize 0x2000
More detail at []

RC2040.INI for SIO2

Config with no switches and other emulation settings for SIO2

idefile = “CPM Inc Transient Apps ACIA.img”;
idefilei = “”;

a13 = 0; // Address switches 0=0x0000 100=0x8000 111=0xE000
a14 = 1;
a15 = 0; // ROM file as ROM source
romfile = “24886009.BIN”; // source for Rom Loading – see a13 a14 a15
romsize=0x4000; // Size of ROM

port = 1; // Console port 0=UART or 1=USB

serialtype = 1; //SIO selected
inidesc=”SIO using 24886009.BIN”; //describe the ini

[SPEED] //vPico overclocking *1000 Mhz
overclock = 250; //overclock the PICO at 250 x 1000

Using ACIA Software

Config with no switches and other emulation settings for ACIA

Requires CPM Inc Transient Apps ACIA.img,, R0001009.BIN and rc2040.ini as below

Rom Details

Rom has basic at address 0x0000 (000)
CP/M via monitor at address 0x8000 (100)
Small Computer Monitor at 0xe000 (111)
More detail at []

RC2040.INI for ACIA

Config with no switches and other emulation settings

[IDE] idefile = “CPM Inc Transient Apps ACIA.img”;
idefilei = “”;

a13 = 0; // Address switches 0=0x0000 100=0x8000 111=0xE000
a14 = 0;
a15 = 1;

// Size of ROM

#ROM file as ROM source romfile = “R0001009.BIN”; // source for Rom Loading – see a13 a14 a15

[CONSOLE] // Console port 0=UART or 1=USB
port = 1;

serialtype = 0; // ACIA=0 SIO=1; ACIA selected
inidesc=”ACIA using R0001009.BIN”; //describe the ini

[SPEED] //Pico overclocking *1000 Mhz
overclock = 250; //overclock the PICO at 250 x 1000

Other INI options.

jumpto = 0x2000; // non standard start vector (e.g not 0x0000);
ramonly = 1;// ram only (no rom, 64K load from romfile);

pioa=0; // set the IO address of the 8 bit port

ide=0; //Turn off IDE
iscf=1; //enable cf file as idefile, rather than the .img format

inidesc=”Broken INI file”; //ini file description to show at boot

trace = 0 // trace details in RC2040.c

CPM Manager / CPM Tools

You can edit the contents of the SD card .img file using CPM Manger and directly add or remove files in the CPM img.

CPM Manager is available here.

If you are using Linux (or windows), you can use the CPM tools to add or remove files. CPM tools are available here If you are using CPM tools, you will need a diskdef files. Disk defs for the RC2040 format(s) are available here DiskDefs


8 bits of a simple IO post are brought out to the RP2040 GPIO as an input/output port. The port is configured on the fly, So if you execute an IN instruction, the port becomes an input port, (with pull ups). If you execute an OUT instruction the port becomes an output port. (The Original RC2014 port is arranged as 8 outputs and 8 inputs, but there were only 8 bits available. )

Switches and Buttons

Details are in the Circuit diagram . Switches and buttons are not required. But give you direct access to the ROM addresses (top 3 address lines) and the buttons allow for a z80 reset, Dump and other functions.

A complete kit of parts for the RC2040 is available here from Extreme Kits