additions... #01

...the real world...

the uninterrupted UPS…

I've tried to keep the amount of technical terms low in the following, but such terms are unavoidable when we're dealing with electronic hardware.

As web designers we are users of such hardware, so it might be a good idea to know a little about how some of it works in the real world.

Most electronic equipment of some size is in need of a steady supply of electric power in order to function. Computers and accessories are known to draw some power, and even battery-powered items may need a recharge now and then.

Clean and stable electric energy is what we expect (or hope for) when we plug some of these items to our mains, but it isn't always clean and stable. It may not even be there...

Uninterrupted Power Supply…

A UPS is a harware-unit that keeps on supplying us with electric power (for a while at least) even when our main supply is down. Hospitals often have some rather large ones for their most vital electronic equipment and operating theaters, while large businesses may use several UPS-units to secure communications and data-systems, and private homes often have some more modest versions for their home computer(s).

When we buy an UPS to avoid interruptions, then we expect to be left "uninterrupted" by the UPS. Sounds logical doesn't it? However, most standard UPS-units for use with computer-equipment are far from "uninterrupted", and that's what the following is all about.

We may buy an UPS to avoid loss of data if, or rather when, the main supply fails. We buy these units for "protection", but we may end up with a much higher bill and a lot less "protection" than we accounted for. Some of these so called UPS-units are weak excuses for their intended use, and some are actually harmful to our equipment.

the "real UPS"…

The electric supply to our farm is quite weak and somewhat unreliable, so our backup system for computers and other electronics has been upgraded with a "real UPS". Yes, it does what it says, providing an Uninterupted Power Supply.

Ok, you can see what "real UPS" we ended up with, and follow the link and read all about it and the other ones from APC. There are probably many more well-working UPS-units produced and sold around, but we've had our unit up and running long enough to know that we've made a good choice.

It took us quite a while and a lot of research to find some completely uninterupted UPS-units, as most so called UPS aren't good for anything but keeping a few lightbulbs alive. That's what our old one is doing, and it's good at it...

signal vs. noise…

The point is that there aren't any real differences between digital signals inside a computer and the noise generated, or let through from the mains, by a fake or badly designed UPS. Some of these UPS-units are so badly designed that they generate many times more noise than the worst main supply itself, so you may get some really nasty surprises and even complete hardware breakdowns if you buy and use the wrong type.

It is simple; as all electronic equipment need time to react — and time to settle down afterwards. Isolate and test them under ideal conditions, and they'll all behave according to spec. However, when they start interacting all hell may break loose behind the panels because they are not built for interaction with anything outside specs, and each on them act and reacts according to their own—unique— specs. Not even those who sell them know what those specs are...

more signal vs. noise…

We may create our own problems by not connecting our equipment properly. Few read manuals, and most manuals are not worth reading on the issue of connecting to our mains anyway. The nicest line-up of cables may induce noise, and the worst rat-nest may work perfectly. Mine is somewhere in the middle, and work just fine.

Wrapping cables tight and nicely together isn't such a good idea when we're dealing with digital signals. It's even worse if the signals are analogue, but that's another matter. Keeping signal-cables away from power-lines is always a good idea though.

Use the right cable for its job. Distribute all cables loosely. Avoid bundling up cables that are too long. Don't expose cables to phycical stress by stretching them or putting preasure onto them, and avoid sharp bends.

Connect your power-hungry items to the right type of socket. All power-sockets must apply some minimum preasure on the plugs in order to function reliable, so avoid loose sockets and socket-stacking. I could mention fire-hasard here.

Remember that when UPS-units are used, then all mains-powered equipment that are directly inter-connected by signal-cables, down to the most insignificant accessories, must get their supply from the same source: the same UPS-unit. Failing to do so is like asking for trouble.

Most small accessories have no built-in protection and only the most rudementary stabilizers. Such items can be blown out of the water by the click of a light-switch. These accessories can easily kill ports in items they are connected to, during a disturbance in their own power-supply, by pulling their common- and signal-lines too far.

All computers and accessories acts differently and have their own time-constants, so you'll never know which one is pulling which and what way during a disturbance. You can have some serious fun with variable voltages all over the place.

The only secure way is to let all inter-connected items draw their power from one—common—power-source, and it better be a good and stable source too.

Forget about common ground in your main power-supply. It won't protect your equipment. You either connect all your items to a common ground, or none at all — and that's the only rule that counts.

If you want some extra protection: install over-voltage dischargers on your mains and communication-lines prior to all your equipment. Use some "solid ones", as lightning can/will trash anything that cannot discharge at least 30,000 Amperes — including your UPS.

switch-noise…

Most computer's built-in power-supply units, and almost all of these small power-supply units that comes with all our computer-accessories, are not designed to cope with extreme fluctuations caused by switch-noise other than the noise they are creating themselves, so the sum of interacting noise-sources ends up in our digital streams and internal supply-lines as high-speed voltage-swing, power-wawes and surges — switch-noise. Today's miniaturized circuits have very little to put up against this, so they are easily broken. You can take my word for it , because this is my real profession.

If there's any delay and/or switching going on inside a UPS when mains fluctuates—no matter how short or insignificant it may seem, then that unit won't be any good for supplying electronic equipment — no matter the low prize or whatever argument a salesperson may have. Use such units to replace candlelights with lightbulbs when the power fails, because that's what they are good for. Less romantic perhaps, but at least those fake UPS-units have something to do.

how UPS works…

A "real UPS" will not generate any noise when mains fluctuate or fails, because it is constructed to bridge the time-gap seamlessly no matter what causes the gap. And "seamless" should mean what the word says.

A "real UPS" won't let any noise get through from the mains either. There are many ways to prevent noise from getting through, and the most efficient way to do it is by using "double conversion".

double conversion
  1. AC from the mains gets converted to low voltage DC.
  2. the battery gets charged (that's low voltage DC)
  3. the battery and/or low voltage DC is the "constant" for conversion back to stabilized AC-voltage.

Note the DC as a "constant" in the middle. DC can be stabilized and held by a battery and/or a "capacitor" of some size. The input can fail and fluctuate all it wants, and nothing of that has any effect on the output.

Now all that matters is the time the battery and/or "capacitor" can hold the DC steady. That's our backup time, often seamlessly gliding from a short-term backup-source to a long-term backup-source. A "capacitor" can hold for seconds on its own, and a battery is only depending on its capacity (size mostly) to hold for minutes or hours, or even days.

The resulting supply (to our equipment) is not influenced by what goes on in the outside world (on the main supply) at all when this "double conversion" is done properly. That's a massive filter.

how some UPS works…

Many (most) producers skip this permanent "double conversion" since "double conversion" may wear the battery-package down a bit faster. The losses or gains concerning the batteries aren't all that significant.

Another—and more important—reason for using a less robust conversion method, is simply to cut production-cost. Less heat have to be radiated away from the unit when simpler methods are used, and most of the components used can be low-cost versions. Lower cost is an important point, but the protection for our equipment is often lowered as well.

In theory: we don't need the UPS to work hard when everything is alright—which it is most of the time(?). Why not skip the parts that aren't needed(?). This is often done by using "single convertion" and "bypass".

single convertion and bypass
  1. lower the charging-current for the battery.
  2. bypass the battery when everything is normal.
  3. switch to "convertion from battery to AC" when the main supply fails.

The theory is correct in an ideal world. However, this switching between "bypass [2]" and "battery-backup [3]" has to work seamless. This is the weak point in most UPS-units. They need time to sense that power are fluctuating and/or failing, and time to react on it.

Since our main supply lines carry AC, UPS-units needs at least 1/100 of a second reacting time because of the cycle-time on our main supply lines, which is 50Hz (60Hz means 1/120 of a second). Most UPS-units relying on this "single convertion" method needs a lot more time than that.

Many UPS-units cut corners by not bridging this time-gap. These UPS-units then reacts by switching after the failure has happened, with a time-delay. The result is that noise from the main supply and the added switch-noise from these UPS-units themselves, are forced upon our equipment.

If you think these switch-noises are insignificant, then concider this: A UPS-unit reacts too late, and add its own correction to a fluctuating main supply. The result: from zero output to 200% of what your equipment are specified to tolorate. And all with full power, which is more than enough to kill a man.

For use with computers and accessories, complete uninteruption has to be achieved if a so called UPS should even be concidered for its name. If any interruption occur at any point, however small and short, then it's just another IPS-unit we're dealing with. It doesn't really matter how they achieve complete uninteruption, but a "real UPS" got to deliver just that.

how some UPS works and some don't…

In order to make the picture a bit more complete—and more complex, it has to be said that there's nothing basically wrong with either method mentioned above. They can both be made to provide a seamless stream of uninterrupted power for our equipment. Just what we want.

Both methods can also be screwed up to such an extent that they can't be used for anything but maybe as a temporary boat-anchor. Maybe that's not exactly what we had in mind.

The "real UPS" will not only help us out when our main power supply fails, but also protect our equipment from all the harmful peaks and surges that so often makes parts of our equipment fail for no obvious reason.

real world testing…

I've tested a UPS-unit which should stay within +/- 5% of 230VAC, with output-ripple and noise at max +/-5%. That's quite normal and acceptable values.
I measured short peaks at more than +/-500V, which is more like +/- 200% and beyond. That's neither normal nor acceptable.

That wasn't "real world testing" they said, so I let that UPS-units supply one of the computer-accessories this particular shop sold for real world use.
The poor thing failed more than once an hour for 3 weeks, whereafter it gave up and passed its soul to the heaven for electronic junk. A similar unit that didn't have this "UPS-protection" didn't fail at all, and are still with us today...

Well, they said that the quality of my mains were so poor that no UPS-unit could handle it! Yeah, I know the quality of our main electric supply is poor, but I wonder... isn't that when we really need a UPS-unit?
Oh, well... we've got one that works complete uninterupted now.

that's it…

I've revealed a well-working "real UPS". I see no reason to reveal more about the bad/fake UPS, since it's just one of many so called UPS available to us. They may work if they're not challenged, but are they really worth the name UPS?

Now, I need to find the switch for that small "alarm" we hear every minute or so when our main supply fails. We know when the power is gone, so there's no need for such a reminder...

sincerely  georg; sign

Hageland 07.apr.2004
last rev: 08.apr.2005


additions...

Some of these so called UPS-units are weak excuses for their intended use, and some are actually harmful to our equipment.

A "real UPS" will not only help us out when our main power supply fails, but also protect our equipment from all the harmful peaks and surges that so often makes parts of our equipment fail for no obvious reason.

Remember that when UPS-units are used, then all mains-powered equipment that are directly inter-connected by signal-cables, down to the most insignificant accessories, must get their supply from the same source: the same UPS-unit. Failing to do so is like asking for trouble.

simple case-studies:

1: bad case: mains fails, and UPS overshoots after ½ period delay.

2: good case: mains fails, and UPS continues with perfect cycles.

Addition to:


about…
…2005