Testing a relay board on the bench, before opening the panel

This page was translated automatically from Italian. Read the original

Replacing a radiant floor controller with a homemade one is an afternoon’s work, if everything goes well. If something goes wrong, it’s a cold house in January with firmware to fix and an open panel. The difference is made by what you test beforehand, on the bench, while the original system continues to function. This is the procedure followed by ORFHAn, but it applies to any relay board with an ESP32 or another microcontroller.

There is one golden rule: do not touch the actual system until the complete test on the bench has succeeded.

The phases, and when the system is out of service

PhaseWhat is doneOriginal system
0document the existing setupfunctioning
1prepare and test the probes, on the benchfunctioning
2upload firmware and test the relay board, without loadsfunctioning
3test everything together, on the table, for one dayfunctioning
4replace the controllerout of service
5calibrate during the first weeksthe new one is active

Until the very last moment, the house remains regulated as before. Everything that can be tested on the table is tested on the table.

Phase 0: being able to go back

Before touching anything:

  1. Photograph everything: the controller, the terminal blocks with wires connected, the labels, the manifold, the actuators. More photos than necessary, from multiple angles.
  2. Label every wire before disconnecting it, using masking tape and a pen: note which terminal it comes from.
  3. Map the zones: which controller output commands which actuator, and which room each circuit serves. If it is not written anywhere, discover one actuator at a time by seeing which pipe gets warm.
  4. Check who turns on the pump: raise the setpoint of one zone and see if the circulator pump starts, then turn everything off and see if it stops. If the pump follows the requests, a pump enable is required: it is explained in the guide on the radiant floor circulator pump.
  5. Keep the original controller: do not throw it away and do not cut the wires short. It is Plan B.

If you are not clear on how to put everything back as it was, do not proceed.

Knowing the actuators

On the manifold, each circuit has a thermoelectric actuator, the “little piston”: inside there is an element that heats up when powered and pushes, or releases, the valve stem. Three things to know before connecting them to a relay:

If an actuator has more than two wires, the extra ones are likely an auxiliary microswitch that signals the valve is open, often used as a pump enable.

The valve actuators connected to the board relays, one per relay, on COM and NO

Each actuator goes to a relay, on the COM and NO contacts: the phase arrives at the common, exits through the normally open contact and goes to the actuator; the neutral goes directly. The NC contact remains free. On the project’s reference board, the Devantech ESP32LR88, there is also an additional reason: the filter that protects the contact from switching noise is only on the NO side.

Phase 2: relay testing, without loads

The board is powered and disconnected from the valves. The ORFHAn firmware has a test mode: while active, the board does not regulate anything on its own, and each relay is commanded manually, one at a time, via the touch panel or with an MQTT message. If you write your own firmware, a mode like this is well worth the time it takes.

What to check:

ORFHAn’s testing mode has a few rules worth copying: the pump relay does not close if there is at least one valve not open, even manually; the mode disengages on its own after one hour, so that no one leaves the system without regulation due to forgetfulness; and it cannot be activated from Home Assistant, only from the touch panel or the board’s local network.

Phase 3: the complete test, on the bench

Sensors, board, and panel work together, still on the bench. The checklist:

Then let everything run for at least twenty-four hours. Stability issues, such as an occasional restart or a sensor disconnecting from the network, become apparent over time, not in the first ten minutes.

Phase 4: on site

Choose a mild day, not in January, and a time that allows you to put the original controller back in place in the evening if necessary.

  1. Switch off the power at the controller circuit breaker, and verify the absence of voltage with a tester: do not trust the switch. Only then open the panel.
  2. Take photos again, then disconnect the actuators one by one, checking the labels.
  3. Mount the board, connect the actuators to the relays according to the phase 0 map, and connect the board power supply.
  4. Watch out for metal: a metal panel shields the WiFi. Before closing it up, test the signal with the panel closed.
  5. Keep signal wires, such as sensors and the bus, separate from the 230 V wires.
  6. Switch the power back on and test zone by zone: raise the setpoint, listen for the actuator, and after a few minutes, check by hand that the pipe in that zone changes temperature.

Two safety behaviors that should be required from any firmware: at startup, all relays at rest, meaning valves closed; and a closed valve when the sensor for its room has been silent for hours and there is no alternative. In summer, this is essential: a radiant floor providing cooling without anyone measuring the humidity will cause condensation. Remember, however, that if the board freezes, the relays stay where they were: an automatic restart in case of a freeze is the bare minimum.

Before starting

You are working on a 230 V system: if you do not have the necessary expertise, an electrician should handle the power connections. Replacing the original controller voids the system warranty. In a condominium, the pump, heating plant, and heat meters are shared parts and must not be touched without the building manager’s agreement.

What is tested

The complete installation guide, including controls and wiring diagrams, is in the ORFHAn repository.