Summary
Toaster is a 10-litre, 1000 W oven with two quartz heating elements, one at the top and one at the bottom. I fitted it with an ESP32 touchscreen, a thermocouple and a solid-state relay for PCB reflow and annealing. It reflowed its first board on 6 October 2026.
Motivation
I already use a hotplate for most of my boards. The Floodlamp Oven had reflowed boards and annealed actuators. I kept it intact, but wanted a few improvements in this oven:
- A steadier heating response. The floodlamp lost heat readily, and the temperature spread around its single halogen bulb was hard to judge.
- Room for larger boards. Ogma 1's 60 × 60 mm boards were at the limit of my 56 × 56 mm MHP30-style hotplate; one Teachtaire board had cold joints at the hotplate edge. Ogma 2 will be larger, and I wanted hand assembly to remain an option.
- Double-sided PCB assembly, with clearance underneath for components.
- Less fiddly loading: a compact enclosure with a rack I can slide boards onto.
- A glass door so I can see the board and thermocouple with the oven closed, and film more complex boards reflowing.
- Less outgassing from the oven itself. I eventually needed a whole extraction system to use the floodlamp comfortably in my room.
Fitting around the oven
I wanted the electronics visible. I used Fusion's generative-design tools again, as I had for Twister's frame and Sparky's camera holder. The white bracket curves around the corner of the oven, holding the display at the front and the thermocouple board and spare cable along the side.
The bracket press-fits over the two existing dials, holding the temperature dial at maximum and the timer on. I did not need to alter the dial mechanisms. The controller switches the oven's supply through the SSR; the original thermostat and independent thermal cutoff remain in the circuit.
I mounted the relay low on the side, where the body stayed cool during the initial tests. The low-voltage wiring runs outside the oven, and USB-C powers the controller. The thermocouple lead slips through the door gap.
The touchscreen
I wanted it to look like a tool. The interface borrows its colours and pixel graphics from Marathon: dark surfaces, acid green for reflow, pink for annealing and a red Stop button. I directed the behaviour and used Codex to write the firmware and LVGL interface, then checked it on the hardware.
The firmware has SnPb and SAC305 reflow profiles with ramp, soak, peak and cooling stages. The run screen shows measured PCB temperature, target, elapsed time and heater demand, with a live trace underneath. Lead-free operation is still locked pending its higher-temperature checks.
Annealing has three dials: temperature, soak time and ramp rate. The screen estimates how long the ramps will take. The same rate is requested on the way down; the controller can add heat to slow cooling, but the oven has no active cooling system. Holding the little bulb and sliding up brightens the screen. Sliding down dims it. The brightness is remembered between boots. The screen sleeps when left in a menu.
Getting it running
A bad solder joint on a MAX31855 data-out header broke the SPI connection; resoldering the pin from the top of the board restored continuity. I later squeezed the thermocouple bundle tightly enough to break a conductor. A replacement probe worked. I attached its bead to a spare Mu PCB on the rack for tuning, so the controller was responding to the board rather than a probe hanging in the air.
PID tuning and USB logging are built into the firmware. Probe faults, overtemperature and run timeouts turn the heater command off.
First reflow
My cheap Kapton tape used to burn in the floodlamp oven wherever it touched the thermocouple's fibreglass sleeve. It doesn't burn in this one, which makes me think the floodlamp had some sharp local hot spots. It also means I can tape the sleeve down and bend it so the bead presses into the board on its own.
The first real board was a Mu PCB with SMD291AXT5 leaded paste on one side. The trial profile ramps to 150°C at 1°C/s, soaks for 60 seconds, then ramps to 205°C at 1.2°C/s and holds for 20 seconds. The heater is capped at 50% for now, and at that power the oven couldn't keep up with either ramp. On the way to the peak the board fell 45°C behind the target. It still soaked at 146–150°C, peaked at 203°C and spent about two minutes above 183°C, where leaded solder melts.
When the heater switched off I cracked the door open, and the board was back below 183°C within a minute.
The 0 Ω resistors I threw down came out looking good, and so did the rest of the board.
I did not reflow the other side of Mu, because I want to be surprised when I actually have to assemble a two-sided board. I will likely raise the board slightly on spare copies, and those are in a box right now. Also, I have laundry to do.
Parts List
- Cecotec Bake&Toast 1090, 10L / 1000W, top and bottom quartz elements
- Freenove FNK0104B ESP32-S3 board with 2.8-inch capacitive touchscreen
- MAX31855 K-type thermocouple interface
- Fibreglass-insulated K-type bead thermocouple
- SSR-40DA solid-state relay
- USB-C controller power
- Generatively designed printed bracket
- Wire, connectors, fasteners and cable restraints
- Original rack, thermostat and thermal cutoff