We had one of those unusual applications through last week. A customer had a load of 1000W that he needed to run for an hour. Simples – using our UPS calculator we can get one of our rack/tower VFI3000B units with 2 battery packs and problem solved. Or not. Problem is there is a maximum weight limit of 40kg and a dual battery solution weighed in at 120kg. At that load level, our KR3000RM-LI Lithium unit -weighing in at only 16kg – will deliver over 30minutes, hence the question: “Can you connect two UPS in series?”
Having given this some thought I was struggling to come up with reasons why not. Sure, there may be better ways, for example additional battery packs, but this isn’t an option on the Lithium units. A quick Google search came up with the “Top4 reasons you should never… 1” which intrigued. However those folk were arguing about using non-sinewave units and I couldn’t agree more with that. Our Lithium units are online double conversion, true sinewave with power factor corrected inputs. The output of these units is going to be a lot better than from the mains, and the input current waveform is a smooth as you can get (baby’s posterior and that) so there should be no reason at all not to do this.
Having brainstormed this for some time with the engineers we could only come up with 4 reasons that you need to consider. That is, aside from the fact the UPS’ must be sinewave output and power factor corrected inputs2.
Can’t argue against that. UPS 1 is not only powering the load it is powering UPS2.
For a 1000W load with say 10% losses, UPS2 consumes 1100W and UPS1 will now be compounded eg. 1210W. An extra 110W in losses as opposed to a single unit with a battery pack.
As for runtime, as the load that UPS1 has to deal with on battery is higher than UPS2 it won’t last as long. Taking good quality online double conversion lead acid systems with 6 internal batteries. One UPS will give 13minutes runtime at 1000W. Add to this an additional battery pack and you will get 32minutes runtime.
In a series arrangement, the load on the first UPS will rise to (roughly) 1100W. This has now reduced the runtime of UPS1 to 11minutes. UPS2 will remain at 13minutes giving 24minutes runtime. Not as good as adding the battery pack.
However, in our case adding a battery pack wasn’t an option, so we are able to knock this objection away.
In all our online double conversion units, there is a built-in bypass (this can be enabled or disabled in most models). What this does is bypass the internal UPS circuitry if there is an overload or a fault. If there is an overload, both UPS’ would detect this immediately and revert to bypass. Due to this there may be the possibility of lost load. This is because reverting to bypass is not instantaneous and there is a momentary drop in power during this transition. The UPS cannot revert to battery operation when in bypass and if UPS2 has no input then the output is also lost.
Good point well made Henry. However this is a fault tolerance issue and only occurs in the event of a fault. As such it can be noted and a decision made to introduce additional fault avoidance measures or to live with it.
The earth leakage on UPS systems (except for our isolated VFI-TX range) is cumulative with the load. That means if you have a load with 1mA leakage, with the UPS of 1mA as well you now have 2mA. Add to that another UPS then all of a sudden you have 3mA. This is a problem for pluggable devices as the limit for such is 3mA. However, the safety standards only require you to take some additional measures, such as supplementary bonding, or by opting for industrial connections.
So another factor to consider but not a deal-breaker.
True, two UPS (of the same ilk) will be more expensive than a UPS with a battery pack (Even with additional batteries in most cases).
But this is fairly obvious.
Absolutely. Given the caveat that the UPS’ should be pure sinewave and power factor corrected input there is no reason at all why you can’t.
In our application this was the only solution. Getting the weight down to 40kg for a 1000W load for 1 hour just isn’t possible with lead acid battery packs. Connecting two KR3000RM-Li units in series as above worked a treat. In our tests UPS1 went onto battery and UPS2 was unaffected. UPS1 eventually ran out of battery steam after 29 minutes and UPS2 went onto battery. The only thing disconcerting was the multitude of alarms going off – UPS1 emits a constant tone to warn you its just died and UPS2 starts its backup warning alarm. Having a smaller load UPS2 lasted 34minutes, thus giving the one hours autonomy that we required and weighing in at only 32kg!
1This is from power-solutions.com out of Barrington, RI, USA. Don’t know much about them except they come top in the Google search when searching for an answer to this “Can you connect two UPS in series” question. They say:
It is true that you won’t get any additional power handling capacity in this configuration, but why should a manufacturer declare your warranty invalid for connecting to a superior power source, or having a power factor corrected load connected? I’ve just shown you can increase runtime and for properly configured EPO ports on both UPS why would data centre EPO be compromised?
2Power Factor Corrected input means the current waveform is in phase with the voltage waveform and in addition implies the current waveform is sinusoidal. Such a load is ideal for all sinewave UPS systems. Non PFC corrected UPS may take current in chunks, rendering the power factor outside the rating of many UPS. This is why I’ve included this as a requirement when daisy-chaining UPS.