My PhD is based upon a specialist piece of equipment that has broken. I've sent it to USA to be fixed (there is no technical support in the UK). It has now come back with another error. It has been out of action for 3 months meaning some of my data has not been collected at a crucial time as my study is longitudinal. Subjects are supposed to be tested every 5 months and this has not been able to happen. It is time dependent as they take medication and have to have the effects evaluated. I can't stop thinking about it. I am not sleeping, worrying the last 18 months has been wasted.

Damn, Nonethewiser - what a horrible thing to happen. It's almost like an act of God or something. I wish I had some words or advice or wisdom, but I'm honestly not sure what you can do about something like that happening. Forgive my ignorance, but is there no way you can change your experimental design to try and compensate for this crisis, or truncate your study? It's just it's not your fault and there appears very little you can do about it otherwise. Is there no way you can get access to this equipment elsewhere? What has your supervisor said about it? Best wishes. :-)

This seems like a big deal, to you, but really, it isn't.
That it is a longitudinal study means it's an even 'smaller' deal. Why? Well, because you've got months of data before the delay and months of data to come after the delay and your statistical analysis of your results will take care of the rest and fill the gap. If you don't know how to do this then you ask around and pick up the phone to someone who does. I spend all my time assessing longitudinal follow-ups and if you think that the follow ups all occur as intended then you are a dreamer!

Besides, have a some perspective - it's a PhD, not an intended Nobel prize. If the machine breaks down it's not your fault and as long as you show you did everything else correct then really, what's the problem? You just need to show you are competent and having a dodgy machine is no reflection on you. It's not worth losing sleep over.