Appropriate UPS for Mac Pro (Early 2008) System?

[BillP]

I'm having trouble figuring out what the best UPS is for my Mac Pro. It's the previous generation (Early 2008), not the most recent, and I have a pair of internal drives and a pair of Dell 21-inch LCD monitors attached to it. It replaced a Power Mac G5 that was running from an APC BackUPS Pro 1000 with no problems, and for the first month or two, I didn't see any troubles.

The problem started when I began (after some long offsite backups finally completed) to sleep the Mac Pro at night and wake it in the morning. About 20% of the time (a guess), the UPS would start screeching and shut off, powering down the Mac as well, which is just the opposite of why you use a UPS. 🙂 However, it never seemed to have trouble on a cold boot, and it could run the Mac fine when I removed wall power from the UPS.

I unplugged the Mac Pro from the UPS and attached it to a Watts Up power meter, where I learned that although the system draws only about 400W in normal usage, the inrush current at startup can range from nearly 800W to slightly over 1000W (and removing the monitors from the equation isn't sufficient to help, since they don't suck much power at startup). The Watts Up records the highest power draw in a session, and from what I can tell from watching, a high inrush current is actually fairly unusual - usually what happens is that I get a high maximum on a particular boot session fairly early, and then no matter how many times I sleep and wake the Mac Pro, the max inrush won't go any higher.

So, I'm OK with the fact that I need a new UPS - the BackUPS Pro 1000 is rated for only 650W and is a little old, so it probably isn't fully capable anyway. But the question is which one to get.

The $250 BackUPS RS 1500 would seem like a reasonable step up, but it supposedly maxes out at 850W, and I've seen several maximum inrush currents over that. Then there's the SmartUPS 1500, which supposedly maxes out at 980W but costs more than twice as much, at $560. I also gather the SmartUPS has pure sine wave output, as opposed to stepped output, but I don't understand how that is better or if it's relevant to this situation.

My goal is protection of the system, not runtime. I'd be happy with a 5-10 minute runtime - just enough time to shut down or sleep. So I'm somewhat bothered at the thought of having to pay a lot more for a UPS that's massively overspecced, just because it's the only way to get one that doesn't fail instantly if the inrush current is too high.

Suggestions about how to proceed?

cheers... -Adam

[BillP]

So the current line of BackUPS devices, if the inrush current is at all above that 850W spec, will still just shut off (I've seen a couple of instances above 850W)? But a SmartUPS, if the inrush current is slightly above its spec (I've seen only one inrush draw above 1000W) won't?

I guess I'm a little confused as to why the high draw would cause the UPS to shut off when the wall circuit behind it can handle the draw - it's not as though the circuit is failing such that the UPS has to, on its own, provide that much power. And since the UPS is kicking over from the circuit to the battery very quickly in the event of a power failure, it would seem that it's normally acting as a passthrough.

Or am I just not understanding some basic chunk of how these devices are designed?

And how does the sine wave capability play into this?

cheers... -Adam

[Erasmus_apc]

If you have seen inrushes higher than what the Back-UPS RS 1500 can handle then it's hard to recommend it without some sense of hesitation. If you know the # of inrush occurrences will be limited you could certainly get away with using the UPS without worrying about an overload. Our Back-UPS units started getting retooled in mid 2008 to be more (not perfectly) compatible with higher-inrush PFC devices, by widening transfer times for state changes and calibrating/improving capacitor (smoother transitions, more power "ridethrough") functionality. This basically means the units do not hiccup or drop PFC loads nearly as often as older models do. Does this mean they are perfect with PFC devices? No, and that is especially true with larger active PFC power supplies...we still recommend Smart-UPS for those.

An SUA1500 will most certainly be able to handle your Mac Pro. Smart-UPS units have been working well with PFC power supplies on datacenter-grade servers and devices for some time now. It's really a decision that's left up to your "risk tolerance".

[Erasmus_apc]

adamengst wrote:
So the current line of BackUPS devices, if the inrush current is at all above that 850W spec, will still just shut off (I've seen a couple of instances above 850W)? But a SmartUPS, if the inrush current is slightly above its spec (I've seen only one inrush draw above 1000W) won't?

This is not necessarily what I said. I said that if you can limit any long or 1000w+ inrush events then the Back-UPS unit will be able to handle the load. I also said that they have been retooled in order to work with PFC/inrush devices better.

Our UPS systems typically have a threshold at which they will transfer to an overload state. This threshold is based on both how far over the rated capacity the unit is seeing in terms of current measurement, and how long the overcurrent event is. The Back-UPS units, as part of their retooling to be more compatible with PFC, have had their overload thresholds extended so that they can tolerate some inrush for several seconds. The Back-UPS, however, also have a longer transfer time to an on battery state (4-8ms as opposed to 2-4ms for a Smart-UPS), so they can be more prone to inrush overloads on battery during this longer period of time. Smart-UPS are also designed a little differently and basically have a higher tolerance for inrush events. However, if you were to give several seconds of overcurrent FAR beyond the rating of a Smart-UPS, it too would overload. In your case, the SUA1500 is closer to the 1000W inrush than the BR1500, and the sine wave and transfer time of the SUA1500 are better, making it a safer choice.

The sine wave output is relevant because this stepped approximation of a sine wave (also known as a "modified square wave") that the Back-UPS uses on battery can/might cause PFC power supplies to hiccup or cut off. Most passive PFC power supplies can work well with a Back-UPS but we find lots of issues crop up with Active-PFC power supplies, as they use a circuit to filter/diminish harmonics, correct input voltage, and correct power factor, which the output of a Back-UPS can cause problems with (in addition to the inrush PFC devices can be prone to generate).

The reason the the units cannot just "handle" what the wall outlet can handle is that it has to process power, charge the battery, etc and as with all smaller UPS' that are standby or line-interactive topology, there is some power wasted. There is an output power factor of the unit that reflects how the input power/kva capacity translates to "real usable power" (watts). This output factor is roughly .6 on the Back-UPS RS. A 15 amp outlet * 120 volts gets you to a rating of 1800 VA/watts assuming a power factor of 1. Because we can only use 80% of a given circuits load per NEC standard, the VA ratings is set to 1500VA, and the resulting .6 power factor brings it to roughly 865 watts. The Smart-UPS is a little more efficient, giving it it's higher wattage rating.

[BillP]

Thanks so much for the details! One last thing. I'm having trouble finding any information about the power supply in the Mac Pro. How can one tell if it's an Active-PFC power supply (and what does that mean)? Can you point me to any discussions of that?

cheers... -Adam

[Erasmus_apc · Accepted solution]

You would need to check the tech specs for your system on the Apple website. Many times they don't list the nitty gritty details like power factor and PFC yes/no/type, and only give you the voltage and amp rating. You will probably end up needing to call Apple and request the information, or check their forums.