Rich voltage, poor voltage: My incandescent tale
Recently some of the incandescent bulbs in my house started burning out at a higher-than-traditional rate. (Yes, we still have quite a few of them in use, in addition to CFLs and even an LED bulb.)
At first, I assumed that it was just coincidental manufacturing defects, but since the bulbs were of different wattages, and from different sources, maybe that wasn't the case. Of course, they could all be from the same factory, just with different labels and packaging.
But I have a small AC-line voltage readout in one of my outlets; it's a unit I received at least 15 years ago from Datel (now part of Murata Power Systems). Even though the date code says 1996, Datel/Murata still lists it on their web site, here. It's a tribute to the longevity of its seven-segment LED display that it is still readable, though the segments have dimmed and their uniformity is not as good as when I got it (see photo) - but hey, this is 24/7 operation over all those years, so we're talking 130,000+ hours, thus far.
This unit is the epitome of simplicity: it's about the size of a fat, rectangular AC plug, and it simply plugs into the receptacle and displays the voltage. That's it: there are no adjustments, no calibration, no user set-up, no complicated interface issues - and I am very OK with that. It does one thing, it does it well, it does it simply, thank you. If I want more, I can get the Kill A Watt from P3 International (here) or similar, but this little unit has served well.
Watching the readout on the Datel unit, I saw the AC line voltage varying between 120V and 125V. I make a quick assessment: that's a ±2% band around a mid-range 122.5V value. Not bad, right?
Wrong. While we are used to having a well-designed AC/DC supply provide better than 1% performance from an AC line of from 85 to 240V, the problem is that many products operate directly from the unregulated AC line, including the standard incandescent bulb. I read several years ago that these bulbs were very sensitive to deviations from their nominal AC design value. A quick look at Wikipedia confirmed that, for example see here. (Yes, I know that is not necessarily a credible source, but the entry seemed well researched and the cited performance reference, #83, was one with which I was familiar, though my own copy is lost.)
To quote the key part from the Wikipedia entry:
"Incandescent lamps are very sensitive to changes in the supply voltage. These characteristics are of great practical and economic importance.
For a supply voltage V near the rated voltage of the lamp:
- Light output is approximately proportional to V 3.4
- Power consumption is approximately proportional to V 1.6
- Lifetime is approximately proportional to V -16
- Color temperature is approximately proportional to V 0.42 "
Whoa, those exponentials are pretty dramatic: a few volts either way around the nominal design operating point and all four key parameters' change significantly ... think of the challenging design tradeoffs. (Also note: the entire article is a fascinating story of the development and technology of the now-humble incandescent, which we take for granted and now even disdain - but we shouldn't be so dismissive.)
The good and bad of high/low AC voltage applies to basic induction motors as well, as they usually operate directly from the AC mains. Lower voltage reduces their output and efficiency and may cause overheating and burnout; higher voltage also can be a major problem.
The good news is that many more household products are now designed to run from internally regulated supplies. For example, many appliances now use advanced motor controllers which are not as sensitive to the AC voltage, within limits. And LEDs and CFLs also use regulated supplies.
But I wonder if the designers of those cheap (excuse me, inexpensive) CFLs, as well as the pricier LED units, gloss over the challenge of designing an internal supply with good input-range performance. As any AC supply designer will tell you, it's a tricky tradeoff to do that and still maintain specifications. Even in regions with "better" AC generation and transmission systems, the nominal target voltage can be 117, 120, or 122 V (in the US) and the variations can be ±5V (of course, much of the world is on 220 or even 240 V nominal mains, and many areas have much wider variations). Both steady-state and transient values can be very far from nominal.
What's been your experience with using the AC mains directly? Have you had to design for operation from wide-range AC input? Have you had products fail or operate erratically due to nominal AC-line issues, not even counting spikes or transients?