TIL: 31st January 2026 — three phase heat pump calamity

Due to russians systematically trying to destroy our energy grid (and along with it making millions of people freeze to death during an exceptionally cold winter at sustained -10C to -30C), there are frequent and lengthy power outages — some last for days. Even when the grid is online the voltage per phase goes between very low (170V) and very high (280V) periodically. This is obviously not great for the various electronic devices that are designed for the 230V standard here in Europe.

To address this issue it's popular to install central voltage surge protection devices, e.g. locally produced and very reliable Zubr relays. It's essentially a simple microcontroller that switches an integrated polarised power relay based on the configured high and low thresholds (and some extra options, like a turn on delay).

In most private houses it's typical to have a three phase grid connection with 230V between each phase and neutral, and 400V between the phases. So most installations will have 3 voltage control relays, one per phase to deal with grid quality (especially in rural areas).

This works great in most cases. My case, though, is not the one as there's a three phase device in it — a heat pump with compressor and fans that have three phase electric motors. It turns out, the professional electrician™ did not consider this when designing the breaker panel years ago and I didn't understand it well enough.

How did I find out? One of the phases got 280V and the relay switched it off, but the single phase devices on this circuit kept switching on and off very quickly, some small USB chargers burned out. This left me (not a professional electrician™) puzzled as there was a definite "click" of a relay disconnecting on the threshold.

Thing is the heat pump kept working and none of its own protections triggered. Eventually I noticed that the switched off relay had a LED indicator (for the relay state) blinking quickly when the heat pump motors started humming. And it clicked (no pun intended).

The two connected phases were sufficient to start the motors, which induced current on the third (disconnected) phase. Normally, this wouldn't have happened as the heat pump would detect an open circuit. But in this case the circuit was closed by the relay as it created a new circuit which included the heat pump and all the devices connected to the power outlets on this phase, completed by the neutral.

(diagram created using generative AI)

So the relay actually created a dangerous situation instead of protecting the connected devices. To do it correctly, there should have been a separate three phase relay (or a controller + separate 3-4 pole contactor) with only the heat pump downstream from it. It would cut all the phases in case of issues with any of them.

I didn't have a sane option to install that, so I chose to connect the heat pump directly to the input, omitting the relays. Which is mostly fine as it includes over/under-voltage protection internally anyway. This works well now even when the grid quality is far from perfect.

Lesson learned: avoid pitfalls by learning enough about everything so that you could verify someone else's work I guess…