It all started when…

I bought this new toy. One that I had wanted for more than a year by that point, but never could save the money to purchase. Finally did, and bought a brand new CD-R drive so I could stop wasting my hard earned money on so many of those 1.44MB floppy diskettes. Would anyone like to take 1000 floppies off my hands, by the way?

The CD-R compliments my system very well. It also performs very well, giving me very little hassle, except during a power failure. But that’s to be expected, I suppose. What is disappointing is wasting a perfectly good CD-R disc when the power does go out, even for a brief moment, and having to spend double the time recording that precious data onto a normally very safe and reliable storage medium.

Now, it has been pointed out to me that the cost of a CD-R is rather low these days. I should think that is blatantly obvious by now to most people, and when you consider the cost/MB for CD-R storage, it is very easy to see why more and more people are buying CD-R drives and media.

The above being said, you might read the rest of this page and say back to me that I am a complete loon. Fair enough, I suppose, as everyone is entitled to their own opinion. But when I bought my CD-R drive and started buying CD-R media, the cost was more than five times what a blank CD-R disc now costs. As well, in those days I was still running a slow, non-Intel 486DX4/100 system that wouldn’t allow me to burn (record data onto CD-R media) reliably beyond 2X, or 300 kilobytes per second (kB/s).

With my hydro service being what it is (many power fluctuations during the summer months) and with it already taking me half an hour to burn a full CD, and with the cost being close to $15 per disc at the time, well I just couldn’t stand losing discs to power failures that lasted a mere second or less.

The Perfect Solution

 typical APC uninterruptible power source

The power backup method of choice was an APC 400VA UPS. It was the perfect companion to my malnourished mid tower system. Sporting the same coloured case as most tower PCs and peripherals, the UPS fit right in with the scenery in my computer room. The model pictured here has much the same appearance as the one I bought, but with a smaller power rating of 200VA.

Testing began almost immediately. Admittedly I did skim through the manual briefly to make certain my methods of setup and use would not violate any of the dos and don’ts described therein. That done, I did not look back. It was time.

I strapped on my voltmeter, and waited. It didn’t flinch from 13.98 volts. Okay, it must be fully charged. “Clear!” mused my mind, recalling countless emergency room scenes on the TV.

 the 7.2 Ah battery inside the APC UPS

I pulled the plug and hit the switch to watch my computer light up as I started my timer. Fifteen minutes run time. Subsequent testing revealed ten minutes with the monitor powered by the battery as well.

Not bad, really. This little battery is rated at 7.2Ah (ampere-hours) and was kicking out an estimated 25 amps for the test duration. That is acceptable for such a small (15x10x6.5 cm) package and means that the UPS inverter circuitry is pretty efficient. Now my problem was solved. No longer would a short blip in power ruin my recording. But if it really ended there, you wouldn’t be reading this page, now would you?

Did I say Perfect?

These UPS units protect against all those malicious power line surges, sags and dropouts that are all too common around my neighbourhood. Are you afraid to leave your computer on, or even plugged in, during an thunderstorm? Not I! These babies also protect against lightning strikes, and APC offers up to $25 000 to replace your equipment if it gets damaged by lightning while being protected by a properly installed APC UPS. That is acceptable! My unit even sports a telephone line surge protection circuit. If it wasn’t for the fact that it took me half an hour to burn a disc, and my UPS only had enough juice to supply ten minutes run time…

Power outtages are frequent at my house, but they never last even a minute, most of the time. The day after I bought the UPS, I thought that my $217.50 was perhaps not so well spent. I might better have spent that money on other things, like more CD-R discs to replace all my coasters, perhaps. I envision someone at the hydro company watching my every move. Just to spite me on the day after my UPS purchase, that person shut off the power on my whole street for an extended period of time. More than an hour, in fact.

If you haven’t guessed by now I was in the process of burning a CD at that point in time. Worse, I knew that little 7.2Ah battery could not power my system long enough to complete the disc. I couldn’t watch… Ten minutes later, the lights were still off, and now my monitor, and the rest of my computer had followed suit. Darn, not another coaster!

Hope was dashed. Spirits crushed. Resentment…


I know a thing or two about batteries. I’ve been playing with them since I was a kid in all my coolest toys, experimenting with them in every elementary and high school science class.

 a huge 51 amp hour sealed lead acid battery

The general rule is, the bigger the battery, the longer it will last. On that principle I performed a little calculation and reasoned that a 51Ah battery like this one should provide me with about 70 minutes run time. That’s plenty long enough to burn a CD at 1X record speed, even if I started the process a few minutes after the hydro failed.

With a huge battery like this one, it’s pretty obvious that attaching it to a UPS which is half its size would require some sort of external connection, and said process is rated “EG,” meaning an electrician’s guidance is suggested.

I think you can guess pretty accurately what happened to the warranty on my UPS at this point… I probably wouldn’t have needed it anyway.

Captain, we need more power!

 graph of running time versus battery voltage

Check this out --------->>

The graph plainly shows that, well, my initial 70 minute run time guesstimate was a wee bit off the mark!

At 79 minutes, the low battery alarm in the UPS started announcing imminent battery failure. I didn’t know at the time how much I could discharge this monster SLA battery before permanent damage would occur, so I became a little bit scared once it dropped below 11 volts. I knew the UPS was supposed to shut down before the battery became damaged, but these were extenuating circumstances after all.

Not to mention that I never knew until this point that the incessant low battery alarm on the UPS could not be muted until the power switch was turned off, or the UPS kicked out due to low battery voltage. And let me tell you that alarm is just a tad unnerving!

A little bit of fear and even more annoyance proved to be enough for me to manually power down the UPS at 94 minutes. My best guess by this point was that the supply would have continued to run for about 120 minutes. Now I shall never know, as I have removed a couple hard drives since then, so my power draw will be slightly less now than during that initial test.

Cause and effect

Sheer bliss! I was one happy camper—for a moment or two.

Nothing goes off without a hitch the first time and this point was proven after the test. A couple of problems were revealed:

Brainstorming through possible solutions revealed the next course(s) of action:

Right, carry on then!

The switch was easy. A little desoldering was necessary to remove the piezo alarm from the circuit board inside the UPS. Using small hook-up wire, I soldered a switch in series with one of the leads of the piezo alarm, and then back to the printed circuit board (PCB). A small hole in the side of the case and it was done. Peace at last!

Might as well have two cooling fans, one on either side of the UPS, to allow for optimal throughput. One might not be sufficient since the UPS is a sealed unit—or, it was until I got a hold on it!

Holes were punched in the sides of the UPS housing to accommodate two 50mm Pentium cooling fans. Being a popular item, they’re produced in great quantities and so cost is low per unit. They are actually slightly oversized for this application, but paying double the price for slightly smaller units was not on the agenda. It’s not like the 50mm fans wouldn’t fit, and their higher throughput rating means more air flow for better cooling.

Oompa, oompa, oompa!

Likened to people who spend twice as much on their stereo systems as they do on their cars, my computer system creates a lot of noise. With six cooling fans in the case and several separate hard drives, it’s almost not suitable for use in a hospital zone! The last thing I wanted to do was add to that noise unnecessarily.

I mean, I could have just strapped two fans to the sides, plugged them in to a 12 volt line somewhere in the supply and be done with it. You’ve been reading this page for several minutes now, and you know me better than that, right?

I couldn’t stop there. I had to have some sort of automatic cooling system on board! Surfing Dallas Semiconductor’s WWW site provided the solution. The DS1620 thermometer/thermostat IC was the only available option at the time. Surfing the site now reveals several new variants on this product.

Sporting a simple 3-wire interface, this chip is super easy to use. Not to mention that Dallas has schematics and demonstration software online, which makes it even easier to use for non-programmer types like me! I used the schematics to construct a cable that connected the standard parallel port of any PC to the 3-wire serial bus of the IC and breadboarded a simple circuit to evaluate my proposed design. It worked flawlessly!

Now I was excited.

Insert Widget A into Slot B and Fasten with Cam G

The circuit I put together for the fan control fit nicely on a 1.25" sqaure piece of Experiment Board I had on hand from Radio Shack. It sits inside the UPS battery compartment, along with various wires and jacks to connect everything from the printed circuit board to the outside world.

I used 3 pin XLR jacks for the power ports, and so followed up with a 4 pin XLR connection for the DS1620 to keep the look consistent. In the planning stages of this project, I had taken safety into account and decided I needed some sort of polarized plugs for the power connections to be sure I could not inadvertently reverse the wires hooking up the battery. I used three XLR jacks for the power connection to ensure that the current ratings for each contact would not be exceeded ever. What multiple jacks also enabled me to do was perform a hot swap from a dying battery to a fresh one, in  the wiring harness I created to connect the battery to UPS the event of extended power outtage. Hey, when you’re racking up frags in Quake 3 faster than rabbits mating, nothing should be allowed to slow you down!

All the wires needed to connect the XLR plugs to the battery makes for an attractive pipe of power pumping love slinking down the front of the battery into a length of split loom tubing that carries the wires to the battery jacks on the UPS.

Sweetness and Love

 the UPS after modifications were completed

The project took less than a day to complete once all the planning and testing was done. And what did I end up with? Well, check the photo and these added features:

The best is yet to come, however! With the DS1620 interfaced directly to my computer through the parallel port, I can easily set the low and high temperature trigger points for automatic fan operation and change them if needed to suit a new environment. Perhaps of equal benefit, I can also monitor the temperature of the rectifiers inside the UPS under load. This allows me to see how well the fans are doing their job of keeping the UPS cool under stress, and thereby prolonging its life.

You can bet I was curious and so did perform another test. This time, it was no holds barred!

 graph of battery voltage versus run time

I graphed the battery voltage as well as the internal temperature of the UPS during the test run. While I wasn’t surprised that the system worked, I was surprised by how well it did so.

With the ambient air temperature being an estimated 21 degrees Celsius, even under full load the fans kept the UPS at 25 degrees or less. The initial rise to 29 degrees shows how quickly the UPS circuits heat up under load. The cooling power of the two 50mm fans is evident with the graph showing a fast descent to 24 degrees within the first six minutes of the test.

Later, the +/- 0.5 degree changes in the graph indicate points where I performed disk intensive operations (high points) or removed discs, such as CD-ROMs and other removeable media. This illustrates just how sensitive the system is to change. Another interesting observation is that the internal temperature rises up from 24 degrees to 25 degrees before the 1 hour mark. This is indicative of the need for proper cooling of these types of power supplies if they are to be used for extended periods of time. Since the manufacturer designs them specifically for intermittent use only, they would normally operate at much higher internal temperatures, but for a shorter duration.

There was one other very interesting observation I made with regards to that annoying low battery alarm I conveniently installed a switch to mute in times of desperation. It seems the circuit that controls that alarm is designed to automatically detect the battery capacity remaining, but adjust itself internally if it is wrong!

On the first test, the alarm kicked in around 11.35 battery volts. On this second test, it was much later at 122.5 minutes, where the voltage was 10.97V. What is significant of this value? Quite simply, that is right around the voltage that I manually powered off the UPS after the very first deep discharge test I performed back in April!

Another test was certainly needed, and here are the results:

 graph of battery voltage versus run time

Strangely enough, the same thing happened here as in the second test. The low battery alarm kicked in with about ten minutes of battery capacity left. In the second test, this was at 122.5 minutes, 10.97 volts. In this third test, the alarm came on at 102 minutes, 10.98 volts.

My original thoughts of an auto-adjust circuit for the two minute alarm warning have been quashed! With this large battery, the alarm comes on at the ten minute mark and I can only guess what this means. My thoughts now turn to a table of values for the alarm circuit. APC makes UPS products with differing battery sizes. The alarm circuit may have a set of values stored in a table somewhere and it simply uses the closest match.

You may notice a couple new things with this test. Just what the heck are those silly orange dots anyway? I let the battery sit and vent itself after the test was finished. While I did so, for interest’s sake I kept tabs on the voltage as it rose to an 11.94 volt plateau almost 47 minutes later. The four orange dots correspond to the points where the cooling fans on the UPS kicked in during these last 47 minutes.

It is evident that the cooling system is not flawless in design. In fact, the oscillations indicated at the beginning and end of the run are very much undesireable in any control system, but they are easily corrected by adjusting the low temperature trip point of the DS1620 to conicide with the ambient room temperature. Now all I need is a circuit to do that automatically! I haven’t thought out that course of action as yet, so if anyone has suggestions on the best way to implement such a feature, by all means email me at

And while we’re on the subject of power backup systems, especially battery backups, I rediscovered something I really already knew, but forgot how large an impact there is hidden beneath the surface. All the tests I’ve documented on this page are with my entire system turned on, or at least most of it, including the monitor. Doesn’t make much sense to have a battery backup without the capability to have the monitor on, after all! But here’s the caveat: By turning the monitor off while running on battery power, total run time may be extended by as much as 160%! Didn’t think a 14 inch monitor used that much power? Guess again! And if your monitor is larger, you’re using even more power to keep it going.

Yeah, so what’s my point? If you’re fragging during a thunder storm, there’s no need to stop, or to worry about your computer being damaged by power glitches when you’re using a well designed UPS. But if you’re just going to sleep and want to leave your computer on to finish that long download, or that horrendous rendering job that will take until noon the next day, the battery backup unit may come in handy if you have power problems. To prolong the life of your battery and help ensure the job finishes without interruption, set up Windows power management to put the monitor to sleep, or better still, turn it off manually where applicable.

A Word on Batteries and Charging

In case you missed surfing through some of the links above, here’s a nutshell explanation of the important stuff. First, the charging circuit built into the UPS is not designed for a larger battery than the one which comes in the box. It is the case that larger batteries require greater charging currents to bring them up to a full charge in a timely manner. This is important!

In terms of battery life, proper care and maintenance is imperative. For our purposes of using batteries in a float application, the information condenses thusly: As a lead-acid battery discharges, sulfates precipitate out of the electrolyte and collect on the plates inside the battery. At first, the sulfates are soft and easily forced back into solution when the battery is recharged. If the battery is not fully recharged in a timely manner the sulfates can harden and become permanently attached to the plates of the battery. The more the surface area of the plates gets cloaked by any deposits, the less capacity the battery has, and this also means the battery will live a shorter useful life.

Therefore, quickly recharging the battery is a must for longer battery life. The UPS charger circuit will not be damaged by placing a larger battery on it due to its constant-current design, but this same design translates to slow charge times for larger batteries and that means sulfates can harden on the plates even while the charging process is underway.

The battery showcased on this web page takes one full week to recharge fully using the UPS circuit! Always recharge a battery with a good quality charger that is rated to work well with the battery size you have chosen to use in your application. As a guideline, a full recharge ought to take less than one day, or even less than half a day.

Be careful about forcing a lot of current into a battery to bring the charge up to full quickly! Use a smart charger for easiest and best results. Higher current means greater heat, and the electrolyte inside the battery will evaporate—or boil—if you get carried away. If electrolyte levels get too low the plates inside the battery will rust, leading to less current output potential and a shorter useful battery life.