Fluke 89iv or 289 for unattended fast transient capture?

[Bakaman]

I have intermittent failures that may be down to power spikes, so I want to log max peak voltage across the ac supply.
The 89iv has fast max peak re recording mode, but what happens if left in that mode and the batteries go flat? Is the peak and its time stamp written to memory before gracefully shutting down? I presume not, but don't want to exhaust a set of batteries trying it.
An alternative might be to use the logging, and let it record any unstable events, and it will gracefully power down after the batteries die. Does anyone know the minimum length of instability that the 89iv will detect and log?
My plan was to leave the 89iv and a 289 on site and collect them after one week, one meter to capture fast transients, whilst the other just logging supply voltage and recording and unstable events.  The batteries will die in that time, so graceful degradation and reading preservation is key.
Is that plan fundamentally flawed?
FYI, the target system is 230V single phase, and we are getting occasional blowing of input surge protection fuses on our rectifier modules. The fuse is F15A online and upstream of a 320V MOV that starts to conduct at around 800V peak, so that is a likely to be a high-energy spike that nobody has looked for before.  Thoughts on a transient recording strategy if the 89iv and 289 won't work well enough? My goal is not to capture the 'biggest spike, but to see how clean and variable the supply is generally. If we are getting 400-500v peaks wrt earth then it would seem plausible that even larger peaks are possible, in which case perhaps hiring a PQ analyzer would be justified on a follow-up visit.

[coromonadalix]

89IV are the 1st version of the 187 189 series,  for longer logging on 189   

you can add a battery pack  BP-189  when you can find them ... they can provide up to 400hrs of logging with bigger cells

but alas their price as risen


but as you wrote  for transients  they maybe not up for this task, scope with the right probes and data logging software would be better