That just leads to "I followed your instructions on your website and my house burned down". No company is going to open themselves up to that.

I know US is very litigious, but I am certain there are plenty of knowledge-based articles by companies in the US of A that could theoretically be tested in court this way, but really, aren't.

I mean, look for any mention of "slow-roast it for 12 hours" in your favourite web search engine (I notice a couple of those say "overnight", yikes).

They don't explain it because it basically never makes sense to daisy-chain a UPS. If you want redundancy in your power supply chain, use a dual-PSU computer. If you want longer backup times, get a UPS with a bigger battery.

The only time I think it makes sense if if you're doing a belt+braces with colo'd stuff in a datacentre.

The DC will have its own UPS system, but in the event it all goes wrong (which happens more often than I'd like) you probably want something in your cab that can give your equipment notification and time to shut down safely, and provide surge suppression and maybe some level of isolation from the inevitable back-emf caused by the rest of the hall going dark.

That's still not saying much, unfortunately — you are going with "authoritative" vs "informative" argumentation ("trust me" instead of "this is why").

Explaining what are the conditions that need to be met for one UPS not to cause issues to another daisy chained off of it would help the reader understand why that's a bad idea.

The main technical reason is that the vast majority of UPSes cannot take what they give. If you have a 1000 VA UPS that is loaded at or near 1000 VA, it will draw significantly more than 1000 VA during periods of recharging, or if you make use of the non-protected outlets on certain models, or if the model is a double-conversion type. Thus, daisy-chaining two identically-sized UPSes will never work; they always need more input power capacity than they can output.

Then there's the issue of output waveform. Cheaper UPS models put out a hideous square wave or modified square wave waveform. This is adequate for being fed into a PC's PSU to be turned right back into DC, but most UPS line status detectors will not be happy being fed in such a bad waveform and will consider the grid faulted. Thus, as soon as you have an outage, both UPSes will go into battery mode, and you will not be able to use the "upstream" UPS's capacity. I suspect that even nicer UPSes may have issues accepting the power generated by another UPS, as their voltage and frequency regulation may not be good enough to satisfy the grid stability tests.

But the larger part of my reasoning for why you shouldn't daisy chain UPSes is that it's attacking the problem of reliability from the wrong angle. By daisy chaining UPSes, you have added an additional single-point-of-failure to your system. If instead you used a redundant PSU, you would have decreased the number of single-points-of-failure.

In my personal experience, I have seen far more power supplies die than I've seen UPSes, so the first thing I would do if I wanted a more reliable power setup is go for the redundant power supply. Only then would I consider attaching multiple UPSes (but I probably wouldn't under normal circumstances).

> The main technical reason is that the vast majority of UPSes cannot take what they give. If you have a 1000 VA UPS that is loaded at or near 1000 VA, it will draw significantly more than 1000 VA during periods of recharging, or if you make use of the non-protected outlets on certain models, or if the model is a double-conversion type. Thus, daisy-chaining two identically-sized UPSes will never work; they always need more input power capacity than they can output.

It really depends on the model. The UPS I have takes uses barely any watts to recharge over a period of multiple hours.