Dome Node

I would use CAN bus - as long as all the nodes have a unique ID, it has lossless arbitration; i.e. even if all the nodes decided to transmit at the same time, it’ll self arbitrate losslessly and all the nodes will wait their turn.
(Read up on CAN bus, it really is elegant)

You’ll want 4 connections: VCC, GND, CAN+, CAN-. I’d recommend RJ45 patch cables, with the power pins tripled up:

1: VCC
2: VCC
3: VCC
4: GND
5: GND
6: GND
7: CAN+
8: CAN-

Then just order a bulk load of ethernet patch cables.

Bonus thought 1: I’m designing a low-CPU-load CAN bus transceiver for Raspberry Pi. It’ll be ready soon. :slight_smile:

Bonus thought 2: Making a board based on the LPC11C22 would be a single chip CAN bus uC solution (it has a built-in transceiver) and it has a ROM CAN bootloader, so a pushswitch on the PCB to trigger the bootloader mode and the Raspberry Pi can just download new firmware on, over the existing CAN bus. Easy peasy.

Bonus thought 3: You don’t necessarily need to make waterproof enclosures - conformal coating spray on the PCB will go a long way to making it condensation proof.

Bonus thought 4: I am so on board with helping with this.

Bonus thought 5: Maybe make VCC 12V or 24V and have a 5V buck regulator on each board. There are chinese buck ICs that cost peanuts each.

1 Like

Would it be of any benefit to have the stodome semi-permanently set up for testing shenanigans?

Google Earth says I would have space :slight_smile:

If you use the 328s there is of course no networking port like CAN or RS485 but there is a Uart on each. You could then take all of the Uart TX/RX pairs back to a BlackIce board.

Thank you everyone for all your help. Thank is a lot of information to conciser and discuss.
I’m not going to be at the space for the next week so will get back on this after in a weeks time.

1 Like

Update on Dome.
The following has been accepted “SHM Geodesic Interactive Kaleidoscope & Cinema”

"In 2016 (SHM) Surrey and Hampshire Makerspace built a geodesic dome. After putting away in storage, I thought there must be more uses to a geodesic dome. And now there is:
A hemispherical projection is, visible internally during the day and glowing out to the world at night. This will show a display of video, photography, sound reactive visuals and an interactive game.
Every node of the dome will have an interactive button forming a 360◦ group game or speed and reaction. It works from inside (under cover from rain) or outside with a projected AI to compete against.
I would be more than happy to do a talk, presentation and demonstrations of how it all works and the various technologies used.
If there is a particular video, image or idea anyone would like to show I would be more than interested to help. "

Test Code

Node - Test Code
Master - Test Code

Node Design
Low cost Arduino/ Atmega with MCP2515, array of 5050 ws2812b’s and a button. Possible to do at sub £10.

The current next set is to order node hardware.

1 Like

@Macaba has also been busy with his node design, using LPC11C22, 8W RGBW’s and 3W speakers.
Connection of RJ45 with pins 1,2,3 =24v, 4,5,6 = 0v, 7=CANH, 8=CANL.

1 Like

I have been working on a ‘cost effective’ dome node PCB, based around an ESP32 and a CAN bus transceiver. While the LPC11C22 IC solution was neat, I am not experienced in their toolchain whatsoever. I found finding code examples hard.
I also figured that the cost of the LPC11C22 and its supporting components was around the same cost as an ESP32 module and CAN transceiver, while also adding WiFi and bluetooth capabilities. The ESP32 has a CAN controller peripheral, and while the LPC11C22 offered CAN bus based firmware updates, we can still do OTA firmware updates on the ESP32 by utilising the WiFi.
Plus we can play with mesh networks and some other fun things.

The idea behind this design is to allow for many different devices to be populated as an option. The base board cost is around £7 per node, and standard features are:
:heavy_check_mark: ESP32 providing wifi and bluetooth connectivity.
:heavy_check_mark: SN65HVD230 CAN bus transceiver
:heavy_check_mark: 5x WS2812 ‘neopixels’ for RGB led fancy things.
:heavy_check_mark: 2x RJ45 connectors for 24V power and CAN bus
:heavy_check_mark: On-board buck regulator
:heavy_check_mark: 2x push switches
:heavy_check_mark: Power and GPIO LED
:heavy_check_mark: many headers for sensors and attachments (I2C, GPIOs, power etc)
:red_circle: A shield header for expansion boards (to be added)

Optional features by means of populating those pads are:
:question: Outer and inner facing LDRs for light sensing
:question: Heavier duty power RGB LED shield connector (exposes 24V and R,G,B GPIOs)
:question: IR Tx/Rx and (hopefully!) a ghetto distance sensor
:question: Speaker
:question: Flexible pads for main components, e.g. ESP32 allowing you to select between an ESP32 module or a dev board.

If you have some free time, I would very much appreciate your help reviewing the board. I plot the schematic to PDF here and the whole project is in that Git repository as a KiCad 5.0 design.

DomeNode Mechanical.pdf (473.1 KB)
Ideas for fixing and board sizes.

As we have the board now in the space and being soldered (Pictures to follow).
The fixing system is awaiting plastic diffusers and 3D printed brackets to be completed.

It’s time to look further into a software. As we have ESP32 with CAN bus and WIFI the best plan would be to OTA updates.

Nice find Andrew, that code looks simple to intergrate into our sketch.
I’ve found more documentation on the update process, the main requirement is that our sketch cannot be more than half the size of the flash space on the ESP.
It also gives us two options, update using the Arduino IDE’s OTA function or via a HTTP server. I think the second option might be worth a shot, as we can host the files on the Raspberry Pi and the example code lets us check sketch version and MAC address, so we could have the nodes check for an update on power up.

EMF Camp Dome in action.

This is a quick summary of what we have learnt from version one and it’s first real test.

  • Future PCB designs should have a additional 5v converting component as the Buck module did not produce a clear/ consistant enough supply.

  • EMF-Camp is not great for OTA updates. This is because mDNS & Bonjour is disabled. A online server based system would be better suited.

  • Waiting for WIFI before continuing boot is not reliable. Suggest trying to use CAN BUS.

  • All 16 nodes have been working although some not consistently. Need to investigate why.

  • 24v wired CAT5 worked well. CAN bus need new firmware to work.

  • The 3D printed brackets where not designed for the 5way nodes. Need to re-look at fixing locations.

For a 5v converter module try a Velleman VMA404 2.5 amp module from Esr Components. It’s output voltage is variable and therefore needs to be preset( whether this could be done without a load) before connecting up. Have ordered 4 modules at £4.75 ex vat and delivery which have been at the boiler room for 2/3 week’s now. Max input voltage is 35v( not 40v) limited by rating of the surface mounted capacitor. Tried looking for fixed voltage modules, but other than the Velleman website itself and wading through Radiospares/ RS Components, I don’t know any other stockists without wading through endless website menu options trying to find what we want. Would give a vote to Esr Components for the quick access with the least hassle. Finding it a nuiscance that Velleman don’t sell to individuals.
Further notes are using Schmitt trigger modules which could be built from simple logic gate ic’s for a more predictable single light level ldr single dumb function interface with a microcontroller board. I have made further notes about using logic Schmitt or flip-flop ic’s for mechanical/conventional debounced switch interfaces. For the ldr there’s also a circuit using a CA3140 op-amp with a signal si diode connected across the op-amp input. This circuit however operates at a higher voltage than 5v. It will therefore need a voltage translation transistor circuit on the output.
Checking Velleman website, for the psu module, and found inconsistencies between circuit diagram component value and what’s actually fitted/ supplied. User instructions state fitting an additional heatsink if over 10W of power is drawn which will be likely in this case. A probable reluctance to use these modules arrised because of the small pcb footprint which is non compatible with 0.1 inch matrix board and difficulty of additional heatsink attachment due to the regulator component being surface mounted on the pcb.Have downloaded this info on smartphone.
As an alternative, will be looking at Quasar Electronics website for psu modules. Update have found some modules on Quasar’s website but they are probably the linear type. They also stock from similar manufacturer 's as do Esr Components. Have also soldered 1.3mm veropins to one of the Velleman VMA404 module. If someone else is brave enough to evaluate the use of these modules…

Psu problems? Would a intermediate 9-12v regulator stage solve problem in stepping down the voltage from 24 volts. It would lessen the dropped voltage across the 5 volt regulator possibly allowing greater current to be drawn. As from the Github Pdf diagram, noticed that the regulators are in cascade. Current draw from the 3.3v regulator stage (800mA max)coupled with the voltage drop across the 5v regulator maybe compounding problems with power dissipated across the 5v regulator. Wonder if use of the 78S 2amp series linear regulators could be used here for a proposal i.e. a 78S12 × 2 forms 1st stage followed by a 78S05 × 2 2nd stage then the 1117 regulator. Depends on what current at 5v the microcontroller board can withstand without substituting 2 1amp 5v regulators for the 5v stage.
Update: Add further LM350/338 regulator for 12-15volt intermediate step down stage. All are linear regulators. There is the LT323AT on Digikey which has a 20v maximum input voltage limit.

Suggest using a 30v transformer? Minimum rating of 100VA to reliably deliver 2.5A allow ing for harmonic loading. Probably deliver close to 40v peak when rectifyed and smoothed on no load. Therefore needs a regulator to prevent damage to node display module. The microcontroller will need a pre regulator stage as well. The Velleman VMA404 modules will need their input electrolytic capacitor changing for a 50v working type. Without line isolation diodes I don’t know if cascaded switched mode regulator modules would work because of intermodulation problems of each switching stage.

Sounds good, I think a more robust PSU for the whole system would be good. It would be even more ideal if we could add some sort of short circuit/overload protection. @Shackbob602 is there anything we could utilise for this kind of thing?

I’m looking forward to getting this project going again for this year’s EMF! Absolutely thinking of doing a v2 board for the nodes with some improvements:

  • better on-board power supplies. The buck converter board was a bit dodgy, and difficult to tune properly to 5v. We should add a separate 5v regulator for the neopixels and 3.3v regulator for better stability.
  • Make it all one sided where possible for reflow soldering
  • Loads of other ideas :slight_smile:

I have 20mm fuseholders at the Boiler Room. Not sure what size to use though space requirements may eventually influence choice. About the pre-built VMA404 modules- the maximum power rating Unheatsinked is 10Watt. Saw some brass/gold cadmium plated stick-on heatsinks at the Boiler Room. Wondered if a conventional through hole TO220 regulator could be substituted or a circuit built up as the connection foot-print is small on a 0.05 inch matrix. Also have some 50mm tall cool rail H cross section heatsinks with a large flat finned heatsink across the top.As for the VMA404 switched mode modules I’ve stuck one of the forementioned heatsinks to one. It also has 4 1.3mm veropins soldered to itself ready for parent board mounting. Left itself on the top of other heatsinks between Matt’s bench and soldering area. Now need 2 pcb’s to be traced out for best measure, one linear composing of dual 78S24’s in parallel (check this with datasheets).The other would use 2 cascaded switched mode VMA404 modules with isolation diodes. Also found a 200VA dual wound 15volt transformer that I had- will bring on Friday. Use the 15 v centre tap to power the buck converter. Also make use of indicator LED 's for power good indication and local load test. Suggest use of ATX supply colour coding with green LED for 24 volt. A 24 volt supply suggests 2 to 3 series connected car batteries for a remote power option.

I have both the 12v and the 24v 78S series linear regulators. There is a method of increasing the current by the use of an external PNP pass transistor (TIP36C). A circuit also exists with 2 external transistors with one used for current protection (in this case a TIP42C). However this may increase the dropout voltage to nearly 6-7 volts as the datasheet shows a 3 ohm power resistor in series with the regulator. Current is set by an extra resistor in series with the TIP36C transistor. Maybe marginal with 15v AC rms supply. Possibly ok with 30v AC rms supply for 24v. Have a 200VA 15 volt dual wound transformer and 2 8amp in line bridge rectifier components. Also have UF54 series and SB5100 ultra fast/ Schottky barrier diodes for switched mode regulator module line isolation. Think these switched mode modules could do with an extra capacitor across the module’s input. I noticed a discrepancy between values given in the module’s schematic and what’s actually supplied/fitted. Pdf info vma404- schematic.pdf. Components now at Boiler Room. Could bring HD instrument case , unfortunately non IP rated, to house this circuit and finally clear bench. For a reduced overload current a 7824 regulator could used instead of the S series for 1 to 1.5A. In the overall circuit an overload would cause the TIP42C to conduct raising the voltage on the TIP36C’s base turning it off reducing the current. Not sure if this action would occur in a cyclic fashion giving rise to possible RFI as with a twin back to back transistor current regulator. Wonder if a strategically placed capacitor would solve this potential problem.

For the microcontroller psu, can 2 switched mode buck converter modules be cascaded with line isolation diodes. If not a slight downgrade in current output would be a buck converter module and the 78S05 2A regulator. Specs say 170uV of noise. There’s also the buck module and the LT323AT regulator, but at over £4 and only 1 to 2 suppliers I’ve not choosen this option. 17/1/20 Delivered forementioned components to the Guildford Boiler Room except the 78S05 regulator. Components are now with the displays in the Dome Node box.

Decided on the buck converter linear voltage regulator option. Set the buck converter output to 9v then feed the 5v and 3v3 regulators from it’s output. Hope the 3v3 regulator would still operate. Seems like isolation diodes have been used between the switched mode and linear/ lo-drop regulators and this will be still the case.