I just took apart an old 8842A that apparently was connected to excessive voltage--at least 1500V---with some power to back it up. The entire front end protection--4 MOVs and 2 fusible resistors--were incinerated. I removed the burnt parts and subbed in 2 regular 1K resistors (to be repaired properly later, of course) and the meter works properly and is in cal. No evidence of any external damage, the circuit board is unharmed (but sooty) and I'm sure nobody died. I'd say that is 'robust' although I can't quantify that.
As I have said, the first DMM I ever owned was a brand new Fluke 8000A bench meter. Fairly costly and on par with what you may find free at Harbor Freight. This meter has little to protect the sensitive custom ICs. A simple mistake and you're changing $70 or so (back then) for new ICs and a handful of precision matched passive parts. Could it be repaired? Sure, I repaired it more than once. But the downtime and cost drove me to retire the Fluke and move to HP3468A. An old story.
While I would guess every meter I have evaluated could have been repaired, given enough time and money, once the main controller is damaged I scrap them. In very rare cases, I have gone so far as to replace ICs to recover them. Only the 121GW prototype, Summit/TPI 194II, Yokogawa TY720, and UNI-T UT181A all fell into this category.
There have been some really low end meters, like the UNI-T UT90A that have such poor clearances, they sustain a lot of damage but I don't test with enough energy to finish them off. Not that I am suggesting these are robust meters but they have been repairable. The UT90A may be the best example of that.
Without looking at the data, I would guess somewhere around 60% of the meters were scrapped. I am not suggesting that these products wouldn't be a fine choice for some people. I'm not here to push any brands or sell meters. However, personally, I don't have a use for a meter that is damaged as easily as my first Fluke. The transients I would encounter in my early days are not representative of what I experiment with now days. For me, the more punishment the meter can take on the bench without downtime, the better. It's not about safety for me but uptime.
I've had a few people comment on my degrading the MOVs over time. After a member here also made the same suggestion, I ran a long term high voltage test on my Fluke 87V where I applied 1.6KV DC. This meter certainly fulfills my needs for a robust meter but its lacking in other ways.
Today, I use the 34401A if I need that performance or am running some experiment that I have automated or want to log the data. After discovering the BM869s and the Fluke 189, if I am working on anything that I have any concerns about HP, I use these handheld meters as I have no doubt that they would survive levels that the old HP would not.
***
SP