Came back (from B&Q) yesterday morning and noticed fairly load bursts of LF hum at regular intervals - put it down to next door neighbour doing some arc welding in his garage. Then went upstairs to bathroom and nearly wet the floor when the LOUD 50 Hz buzz went off behind me. The bath was a mess of sprayed water and black carbon dust and there was a continuous trickle of water from the base of the Mira Event Power Shower. Went to the bedroom to remove power - the electrician had not thought to provide an isolating switch so had to find a screwdriver to lever the fuse out of the spur. Removed shower casing and discovered motor too hot to touch, carbon dust spread around, and water coming out of top of pump case. Had wife make temporary fuse connection so I could see what was happening - motor not turning but powered up even although water control was in off position and microswitch contacts open. Violent arcing from brushes.
So turned off hot and cold water and went back to B&Q (40 miles) to buy some press-fit isolation valves - the plumber had not bothered to fit any. (See other thread about HBP pipe and Speedfit).
The shower was used the previous evening at about 11pm and operated normally. We left the house about 9:30 am and returned about 11:15. The motor has a Klixxon thermal fuse (110 deg C) which was opeating after about 3-5 seconds then resetting a few minutes later. The temperature reached had melted the impeller within the pump and caused other damage.
PCB checks out working on bench (Triac, diac, varable pot) and the microswitch works normally. The microswitch does not remove power but is in the gate circuit of the triac. Putting a variable resitor across the microswitch I found that about 30k was sufficient to switch the Triac on - so possible a combination of carbon dust and moisture /could/ have started the destruction process. The microswitch was protected with a rubber cover and appeared clean enough. With the shower turned off the outlet is blocked to the pump would have been under a strain.
Checked resistance between adjacent armature segments and found one place giving high reading (about 32 ohms instead of about 4). Not sure if this could have burnt out once the pump jammed or was the initial cause of the motor stopping and overheating. There are score marks in the pump casing which would indicate that there was rotation when the failure started.
Still puzzled that the thing started by itself - suspected mains surge had blown triac but it is rated 600V and the PCB still works.
13 years is not bad but I would have preferred a more dignified death.Pictures of damage here:-