Heat pump compressor is much louder than I expected — how do I work out whether that is normal before I argue with the installer?

Add one thing that is not about the unit at all: an oversized heat pump is a loud heat pump.

If the capacity is well above the actual heat loss of the house, it cannot modulate low enough to match a mild day, so it runs hard, satisfies the demand, stops, and starts again — over and over. Every start and stop is audible, the compressor spends its time at the noisy end of its range, and it is worse in the shoulder seasons when the house barely needs anything.

A correctly sized unit on a low flow temperature runs long and quiet, which is the behaviour you want anyway for efficiency.

You can see this from the outside without any special equipment: watch the cycling. If it starts and stops repeatedly over an hour on a mild day, sizing or controls are the story and the noise is a symptom. Turning down the weather compensation curve so it runs longer and gentler often reduces the noise as a side effect.

Before the conversation with the installer, get data, because "it is louder than I expected" is not something anyone can act on.

A phone sound level app is not calibrated and will not survive scrutiny, but it is fine for relative measurements, and relative is what you need: same spot, same distance, at several times of day, noted against what the unit was doing. A cheap dedicated meter is not expensive if you want defensible absolute numbers.

What turns this into a real case is a pattern: quiet at low output, loud above a certain point, worst at a particular time. That points at a cause. A single reading points at nothing.

Also record the defrost events with times. If the frequency is genuinely excessive, that is a documented fault rather than an opinion.

You can separate these by ear and by timing before anyone gets a meter out, and the three sounds you describe are three different things.

"The whole unit changing its mind, loudly, for a minute or two" is almost certainly a defrost cycle, and it is normal. In cold damp weather the outdoor coil ices up, and the unit reverses itself to melt it. You hear the reversing valve change over with a distinct clunk, the fan stops, the compressor note changes, and there is often a hiss and some steam. It lasts a few minutes, happens periodically in the conditions that cause icing, and then everything returns to normal. If that is what you are hearing, it is the machine working.

What is worth checking is how often. Defrosting every couple of hours in genuinely icy weather is expected. Defrosting every twenty minutes in mild damp weather suggests poor airflow across the coil, a sensor problem, or an installation too close to a wall — and that is a real conversation to have.

The low rumble that comes and goes is the compressor, and here is the thing about the number on the datasheet: a modulating unit runs at whatever output the heat demand asks for, so its noise varies continuously. The quoted figure is at one specified operating point, usually a mild one. It is not a promise about a cold night at full output.

More usefully: the datasheet almost certainly quotes sound power, and what you hear is sound pressure. They are different quantities and the difference is large — power is a property of the source, pressure is what arrives at your ear and falls off with distance. Comparing the datasheet number to a phone app reading and concluding you have been misled is the single most common mistake here. Compare like with like or the comparison means nothing.

The fan is the one you can usually do something about, and it is where I would start:

• Most units have a quiet or night mode that caps fan speed and compressor output during set hours, at the cost of some capacity. Frequently not enabled by the installer. Ask.
• Airflow obstruction makes fans loud. There is a minimum clearance in front of and around the unit; check the actual installation against it.

And the one genuinely worth pressing on: mounting. A compressor sitting on a rigid bracket bolted to a house wall turns the wall into a soundboard, and structure-borne noise into a bedroom is not fixable by anything you do outdoors. Anti-vibration mounts, and preferably a ground-standing frame rather than a wall bracket, are the difference between hearing it and not.

Easy test that costs nothing: put your hand flat on the wall inside the room while the unit is running hard. If you can feel it, the path is the structure, not the air, and the fix is mechanical isolation.