Uninterruptible Power Supplies

I hate these things. Why? I hate batteries, that's why. Can't live without 'em, but they're a big pain in the fundament, and they just don't last. (Yeah, I know they're a consumable but they look and feel [and cost!] like they ought to be capital equipment... but I digress.) At one job we used to refer to UPS's as IPS's because of their propensity to take down equipment (due to overloading and/or battery failure) during power glitches that left un-'protected' systems unaffected.

These things need enough care and feeding, and I have enough of them now that it warrants having a page for them. We have:

  1. APC Back-UPS Office (280VA) (Once used on the troll, might be defunct.)
  2. APC 450 AT+ (Used on the DVR's for awhile, now on a battery-challenged laptop.)
  3. APC Smart-UPS 2000 (Used on the computer cluster.)
  4. APC Smart-UPS 3000XL
  5. Best Power FerrUPS FE3KVA
  6. Best Power FerrUPS FE7KVA
  7. APC Smart-UPS 1500 (Used on the upstairs computer cluster.)
  8. APC Back-UPS 900 XS

Only three are actively in use. The big 7kVA unit is targeted for the whole house someday, the others are, ahem, backups.

Saturday, February 18, 2006

I installed the GSA-surplus 2 kVA UPS (APC Smart-UPS 2000) on the computer cluster, using a 75 AH external 48 V battery bank. The four sealed batteries, even as seconds, cost some serious moola ($355.01!) when compared to the $35 UPS, imagine that! The UPS system is residing in the 'basement' of the enclosed rack-mount computer cabinet that graces our family room which is the hiding place of many an old computer.

The GSA-surplus APC 450 (free with the 2000) also needed a new battery, and I put another internal one in. I deployed it on the DVR's so that they wouldn't get a sudden surprise if they were recording.

Wednesday, October 24, 2007

I finally cobbled up the FerrUPS FE 3.1k 3 kVA UPS I was given free nearly a year ago. (The eventual plan is to drive this UPS with the small diesel genset, if possible, as a sustained round-the-clock modest backup power source.) It had eight 75 Ah new-ish batteries in it that had been slowly self-discharging. New, that's about $700 worth of batteries, not something I wanted to turn into scrap due to (any more) neglect. The prospect of running the 12 V battery charger serially for the next 8 days was a bit daunting, so I took the UPS's plug off and bolted a spare range cable to it and got it hooked to the 240 V range outlet via the welder/generator extension cord. That let its own internal charger do the job, it's pushing about 4 A into the 48 V series-parallel battery bank. A day or two of charging and it ought to be good to ignore for a few more months. I'm going to have to wire the cable correctly at some point.

...After work I checked, and it was running just shy of 2 A charging current. I played around with it, it puts out 120 V on one outlet, and 212 V on the other two, with no center tap. Looks like it might be rigged for putting out 3-phase voltages (120/208). That might make it interesting to try to put the whole house on. I ran a 500 W lamp off of it for a minute or so, there was no flicker at switchover. Makes sense, I think this is an 'always-on' type of unit. (It's sure loud enough to be one! Not something you'd want in your bedroom.) It drew about 14 A from the 48 V battery while running the lamp, about 4 A no-load. Plugged back in, charging current went to about 19 A for a little bit, then dropped back off to about 4 A.

Thursday, October 25, 2007

The charging current was down to about 300 mA, so I turned it off and unplugged it. More to do tidying it up, later.

I found some manuals online for a similar model of UPS. User Manual, and Installation Manual. According to these, it might not be possible to run the house on this UPS (both legs of the center-tapped 240 V) without using a rather hefty autotransformer to regenerate the center-tap. The manual says you can run loads L1-N and L1-L2, but not to run anything L2-N. Also, it's possibly only a 2 kW UPS, though 3 kVA. That's not really all that big, but it might be enough to pickle the house overnight, with care. The 4 A quiescent draw is not too attractive, but given the size of the battery bank (150 Ah@48 V) that's still some 37 hours to fully deplete the charge.

It's probably also possible to short L1 and L2 together at the house end and feed L1-N from the UPS to it. That'll feed all the house's 120 V loads, but completely starve all its 240 V loads. In fact, they'll all be rendered harmless so far as loading the UPS goes.

Monday, November 12, 2007

I bought a couple of plugs today, with these I should be able to cable the UPS up to the house both ways. (Not simultaneously, of course!)

Wednesday, November 14, 2007

I took the range plug off of the spot welder and substituted the welder plug I bought Monday. I then took the range plug over to the UPS, but I managed to drop it on the way and the brittle plastic of the retaining hook shattered. So I got to glue that. I did get the wires connected, so once the glue (cyanoacrylate for positioning and Shoe Goo for strength) dries I can screw it together.

Thursday, November 15, 2007

I made (partially) the cable-of-death for hooking the output of the UPS to the house. I used a surplus range cord and a newly-bought 30 A 120 V locking plug. Both hot 220 V legs are hooked together to the 110 V output so I had to use a short length of 10 gauge wire to bring the power out of the plug, as the two big wires from the range cord couldn't both fit into the plug simultaneously.

With it wired (though not yet insulated), I turned off the feed to the garage subpanel and hooked the UPS up instead. I was able to turn on all three banks of 1 kW fluorescent lights, but after only a few seconds they started flickering oddly then the UPS cut off due to overload. It would run two banks just fine, and I also ran up a garage door when I had only one bank on. Looks like it could work out fine, except for the fact that its 2 kW (3.1 kVA) rating is a little too small.

I then left the UPS on charge to replenish what I'd taken out.

Saturday, September 6, 2008

I plugged in the FerrUPS 3KVA again to top off the batteries, it got several hours of float charging. This needs to be done several times a year, at a minimum.

Thursday, September 11, 2008

Today I looked at a FerrUPS FE7KVA (5 kW) UPS I found listed on Craig's list for $400, with nearly-dead batteries and a manual transfer switch and a circuit breaker. (It's some $6,000 new, and that's without the switches.)

The advertisement:

Ferrups UPS 7kVA + transfer switches and breakers!—$400 (Spokane, WA)

Reply to: sale-825965501@craigslist.org
Date: 2008-09-03, 3:37PM PDT

Up for sale is this used but great UPS from Ferrups. This one supplies power for up to 7 kilovolt-amps. It uses four batteries. The control panel is removable and on a coil-cord for running tests and checking the unit. It can be used in conjunction with software on a Windows PC connected to the UPS via a DB9 serial cable.

The batteries in this unit are useable but it is recommended the buyer purchase new batteries and install them or have them installed. It is not the responsibility of the seller to provide any tech support or installation or repair services to this unit. That is all the responsibility of the buyer.

Having said that, while this unit is sold "as-is", it worked fine when last removed from service about 1 year ago. It has always been kept in a dry, clean area during operation and storage. This was a professionally used and maintained unit. Payment accepted are cashier's checks and money orders. We can discuss Paypal if need be. Please feel free to contact me (Jim) with any questions you may have.

Pictures are coming soon. Thanks! Xfer SW Front Back Opened

It is the sole responsibility of the buyer to pick up or arrange shipping of the UPS and related items included, to the destination of the buyer's choice. The seller will not ship. Local pickup is preferred.

PostingID: 825965501
The contact:
Hello there! I am sorry I did not see your first email. Thanks for your interest in the UPS.

You can come and take a look at it if you would like.

We are 1802 North Monroe Street in the old Columbia Cycle Shop building next to McDonald's. We are also right next to the Aloha Island Grill and across Monroe Street from Shari's restaurant.

You can call me at the number below to set up a time.


James Kusler, Information Technology Manager
PHONE | 509.624.1613 or 800.822.4456 x.1029 | FAX | 509.624.1604
jkusler@sound-tele.com | www.sound-tele.com | www.solaxis.com

Sound Telecom is pleased to be honored as one of the
Inc. 5000 fastest-growing companies in the United States.

Should be a nice complement to this generator, and could allow for overnight running of the house. Silent running, so to speak. (Except for the loud buzzing.) This is the big brother to the 3.1 kVA unit I already have, which is a bit too small for whole-house duty. (The no-load draw on the battery bank is too large for these FerrUPS units to be a really good whole-house overnight backup, but the price is sure right.) The battery bank I already have should be good for it to use.

When I tried to power up the unit nothing happened. It's likely the batteries have gone completely dead since it was taken out of service several years ago, but as there was no way to prove that's all that's wrong I offered $200: all I was willing to risk in case it's in worse shape than claimed. (No way to prove otherwise without a research effort.) Even if it's junk the transfer switch and breaker are probably worth about that so it was a fairly safe bet. They called later and accepted my offer. They'd said that they'd had no other calls on it at all, nor bites on eBay. For that price I'm to pay for it and pick it up tomorrow. OK, will do.

Friday, September 12, 2008

I picked up the UPS today. They took a $200 Cashier's Check (their choice, and I get them for no fee) as the preferred form of payment. I'd brought the truck, two planks, and a tie-down. They got some burly men from the bull pen (it's a phone answering service) and with much grunting we pushed it right up the ramps. It's a heavy beast, but was still manageable with enough men. I'd also brought Jill, so we had lunch at the Aloha Island Grille across the street, a place I've eaten at before when I used to work downtown. Downscale, but yummy.

...When I got the UPS home I got it out of the truck by using the hardtop hoist for Jill's 560 SL. I dropped the lid onto the car and drove it out, and backed in the truck. I laced a tow strap into a basket around the UPS (it was exactly long enough) and hooked it to the winch and lifted a bit. It was a close thing, the winch strained mightily and almost couldn't lift it. I helped by heaving up on it while I pushed the button. I then drove the truck out from under and lowered it to the floor. (I had a bit of difficulty at first when the strap's hook managed to jam the cable pulley so it could raise but not lower. I had to jack up the thing with a small floor jack and put blocks under it in the bed to release enough cable strain to get it unjammed. I then re-rigged the lifting eye to avoid the problem.)

Once down I wheeled it around into a temporary storage place behind the jacked-up 300 SDL in the center bay and took the access panel off one side. This thing is so heavy I had to use a prybar to lift the little wheels over the small lip in the concrete getting back into the garage. After blowing out the worst of the dust bunnies I found the main reason it wouldn't work is that whomever decommisioned it had done the job properly and disconnected the battery cables! (Likely as a safety matter since the unit's output wires were just bare pigtails.) Neither of the two batteries I could easily reach on the open side were flat, and I put them on charge. We'll see if there is any life left in the bank in a few days, but I'm hopeful that I can at least do some testing.

The batteries in the bottom of the cabinet are (4) Werker WKA-80J/FR 12 V 80 Ah AGM batteries. Good stuff, no sign of leakage. Replacement price on these batteries is around $100–150 each.

Saturday, September 13, 2008

This morning the first battery was still drawing 2 A and it was pretty warm and making bubbling sounds; that's not a good sign. I disconnected the charger and put it on the next one which, perhaps due to the efforts of the trickle charger that'd been on it since last night, drew somewhat less charging current than the first one had. I moved the trickle charger to the third battery.

I also removed the other access panel and used the air compressor to really blow the dust out of the works. Cough, cough!

I found a link to a nice Battery FAQ that contains much information of interest. I've extracted a few tidbits (for protection against site evaporation):

Lifespan of Batteries

The lifespan of a deep cycle battery will vary considerably with how it is used, how it is maintained and charged, temperature, and other factors. In extreme cases, it can vary to extremes—we have seen L-16's killed in less than a year by severe overcharging, and we have a large set of surplus telephone batteries that sees only occasional (5–10 times per year) heavy service that are now over 25 years old. We have seen gelled cells destroyed in one day when overcharged with a large automotive charger. We have seen golf cart batteries destroyed without ever being used in less than a year because they were left sitting in a hot garage without being charged. Even the so-called "dry charged" batteries (where you add acid when you need them) have a shelf life of 18 months at most.

Here are some typical (minimum–maximum) expectations for batteries if used in deep cycle service. There are so many variables, such as depth of discharge, maintenance, temperature, how often and how deep cycled, etc. that it is almost impossible to give a fixed number.

Battery Construction Materials

Nearly all large rechargeable batteries in common use are Lead-Acid type. (There are some NiCads in use, but for most purposes the very high initial expense, and the high expense of disposal, does not justify them). The acid is typically 30% Sulfuric acid and 70% water at full charge. NiFe (Nickel-Iron) batteries are also available—these have a very long life, but rather poor efficiency (60–70%) and the voltages are different, making it more difficult to match up with standard 12/24/48 V systems and inverters. The biggest problem with NiFe batteries is that you may have to put in 100 watts to get 70 watts of charge—they are much less efficient than Lead-Acid. NiCads are also inefficient—typically around 65%—and very expensive.

Temperature Effects on Batteries

Battery capacity is reduced as temperature goes down, and increased as temperature goes up. The standard rating for batteries is at room temperature—25 °C (about 77 °F). At approximately –22 °F (–27 °C), battery Ah capacity drops to 50%. At freezing, capacity is reduced by 20%. Capacity is increased at higher temperatures—at 122 °F, battery capacity would be about 12% higher.

Battery charging voltage also changes with temperature. It will vary from about 2.74 volts per cell (16.4 volts) at –40 °C to 2.3 volts per cell (13.8 volts) at 50 °C. This is why you should have temperature compensation on your charger or charge control if your batteries are outside and/or subject to wide temperature variations.

Because they have so much mass, they will change internal temperature much slower than the surrounding air temperature. A large insulated battery bank may vary as little as 10 degrees over 24 hours internally, even though the air temperature varies from 20 to 70 degrees. For this reason, external (add-on) temperature sensors should be attached to one of the POSITIVE plate terminals, and bundled up a little with some type of insulation on the terminal.

Even though battery capacity at high temperatures is higher, battery life is shortened. Battery capacity is reduced by 50% at –22 °F—but battery life increases by about 60%. Battery life is reduced at higher temperatures—for every 15 °F over 77 °F battery life is cut in half. This holds true for any type of Lead-Acid battery, whether sealed, gelled, AGM, industrial or whatever.

Cycles vs Life

A battery "cycle" is one complete discharge and recharge cycle. It is usually considered to be discharging from 100% to 20%, and then back to 100%. However, there are often ratings for other depth of discharge cycles, the most common ones are 10%, 20%, and 50%. You have to be careful when looking at ratings that list how many cycles a battery is rated for unless it also states how far down it is being discharged. For example, one of the widely advertised telephone type (float service) batteries have been advertised as having a 20-year life. If you look at the fine print, it has that rating only at 5% DOD (Depth of Discharge)—it is much less when used in an application where they are cycled deeper on a regular basis. Those same batteries are rated at less than 5 years if cycled to 50%.

Battery life is directly related to how deep the battery is cycled each time. If a battery is discharged to 50% every day, it will last about twice as long as if it is cycled to 80% DOD. If cycled only 10% DOD, it will last about 5 times as long as one cycled to 50%. The most practical number to use is 50% DOD on a regular basis. This does NOT mean you cannot go to 80% once in a while. It's just that when designing a system when you have some idea of the loads, you should figure on an average DOD of around 50% for the best storage vs cost factor. Also, there is an upper limit—a battery that is continually cycled 5% or less will usually not last as long as one cycled down 10%. This happens because at very shallow cycles, the Lead Dioxide tends to build up in clumps on the the positive plates rather in an even film.

Battery Voltages

All Lead-Acid batteries supply about 2.14 volts per cell (12.6 to 12.8 for a 12 volt battery) when fully charged. Batteries that are stored for long periods will eventually lose all their charge. This "leakage" or self discharge varies considerably with battery type, age, & temperature. It can range from about 1% to 15% per month. Generally, new AGM batteries have the lowest, and old industrial (Lead-Antimony plates) are the highest. In systems that are continually connected to some type charging source, whether it is solar, wind, or an AC powered charger this is seldom a problem. However, one of the biggest killers of batteries is sitting stored in a partly discharged state for a few months. A "float" charge should be maintained on the batteries even if they are not used (or, especially if they are not used). Even most "dry charged" batteries (those sold without electrolyte so they can be shipped more easily, with acid added later) will deteriorate over time. Max storage life on those is about 2–3 years.

Batteries self-discharge faster at higher temperatures. Lifespan can also be seriously reduced at higher temperatures—most manufacturers state this as a 50% loss in life for every 15 °F over a 77 °F cell temperature. Lifespan is increased at the same rate if below 77 degrees, but capacity is reduced.

No-load typical voltages vs state of charge

(figured at 10.5 volts = fully discharged, and 77 °F). Voltages are for a 12 volt battery system. For 24 volt systems multiply by 2, for 48 volt system, multiply by 4. VPC is the volts per individual cell—if you measure more than a 0.2 volt difference between each cell you need to equalize, or your batteries are going bad, or they may be sulfated. These voltages are for batteries that have been at rest for 3 hours or more. Batteries that are being charged will be higher—the voltages while under charge will not tell you anything, you have to let the battery sit for a while. For longest life batteries should stay at 40% charge or higher. Occasional dips into the 20–30% range are not harmful, but continual discharges to those levels will shorten battery life considerably.

State of  
12 Volt  

It is important to realize that voltage measurements are only approximate. The best determination is to measure the specific gravity, but in many batteries this is difficult or impossible. Note the large voltage drop in the last 10%.

Battery charging takes place in 3 basic stages: Bulk, Absorption, and Float.

Bulk Charge: Current is sent to batteries at the maximum safe rate they will accept until voltage rises to near (80–90%) full charge level. Voltages at this stage typically range from 10.5 volts to 15 volts. There is no "correct" voltage for bulk charging, but there may be limits on the maximum current that the battery and/or wiring can take.

Absorption Charge: Voltage remains constant and current gradually tapers off as internal resistance increases during charging. It is during this stage that the charger puts out maximum voltage. Voltages at this stage are typically around 14.2 to 15.5 volts.

Float Charge: After batteries reach full charge, charging voltage is reduced to a lower level (typically 12.8 to 13.2) to reduce gassing and prolong battery life. Note that for long term float service, such as backup power systems that are seldom discharged, the float voltage should be around 13.02 to 13.20 volts.

Battery Charging Voltages and Currents

Most flooded batteries should be charged at no more than the "C/8" rate for any sustained period. "C/8" is the battery capacity at the 20-hour rate divided by 8. For a 220 Ah battery, this would equal 26 Amps. Gelled cells should be charged at no more than the C/20 rate, or 5% of their amp-hour capacity.

Charging at 15.5 volts will give you a 100% charge on Lead-Acid batteries. Once the charging voltage reaches 2.583 volts per cell, charging should stop or be reduced to a trickle charge. Note that flooded batteries must bubble (gas) somewhat to ensure a full charge, and to mix the electrolyte. Float voltage for Lead-Acid batteries should be about 2.15 to 2.23 volts per cell, or about 12.9–13.4 volts for a 12 volt battery. At higher temperatures (over 85 °F) this should be reduced to about 2.10 volts per cell.

Battery Aging

In situations where multiple batteries are connected in series, parallel or series/parallel, replacement batteries should be the same size, type and manufacturer (if possible). Age and usage level should be the same as the companion batteries. Do not put a new battery in a pack which is more than 6 months old or has more than 75 cycles. Either replace with all new or use a good used battery. For long life batteries, such as the Surrette and Crown, you can have up to a one year age difference.

When using a small solar panel to keep a float (maintenance) charge on a battery (without using a charge controller), choose a panel that will give a maximum output of about 1/300th to 1/1000th of the amp-hour capacity. For a pair of golf cart batteries that would be about a 1 to 5 watt panel—the smaller panel if you get 5 or more hours of sun per day, the larger one for those long cloudy winter days in the Northeast.

Conclusions? I may well be using too high a float charge on the Kohler genset's starting battery. This site essentially states that it should be at most 26.8 V, whereas I've got it set at 27.8 V. (I measured it at 27.4 V, but I did crank the thing over for some exercise less than 24 hours earlier.)

...In the late afternoon the second battery was still drawing 2 A and it was pretty warm and making bubbling sounds. I disconnected the charger and put it on the third battery. I moved the trickle charger to the fourth battery.

...Five hours later #3 was down to about 3 A.

Sunday, September 14, 2008

This morning the third battery had finished taking a charge, and it wasn't too warm or bubbling. I moved the charger to the fourth (and last) battery. The open-circuit voltage on #1 was 13.23 V, #2's was 13.24 V. (These had both had a bit of time to rest after being charged.) After only about 10 minutes of charging #4 was already down to a 3 A draw.

...By early afternoon #4 was warm and bubbling, so I took it off charge. Ready to try firing it up, I suppose. I measured #3's rested open-circuit voltage at 13.23 V.

Wednesday, September 17, 2008

I printed the FerrUPS manuals I had (User Manual, and Installation Manual, printed 8 pages to the sheet so not too many trees died for this), and reconnected the battery bank; its resting voltage is 53.8 V. Ready for experimentation!

Thursday, September 18, 2008

I opened the transfer switch. It became apparent that the short cable was the line input, and the long was the system output, and that the breaker was on the UPS output. I turned off everything and flipped on the UPS. Buzz-o-rama! It seems to be working. I checked the output voltages and they're OK. I got out the extension cable and poked the line input leads into it and everything seemed to be OK, it reacted to input voltage coming and going. The zero-load battery current seems to be around 10 A, which is much more than the smaller UPS's 4 A. I hooked up a 500 W lamp and it acted as one would expect. When powering it the battery current seemed to be around 21 A, the same increment as the other UPS. Once I could provoke it into charging it only wanted to push about 5 A into the battery, it might do more if further discharged first.

The fluorescent display in the control panel is very weak and ghosty. It is not good, barely usable in fact. I wonder if it is repairable?

...After work I tried the brightness setting of the panel, it's already at its highest. I also dug up another User Guide for the series, it looks more like this unit. Also acquired was the Tip 503 guide. I put the unit back on to charge. (According to this manual the unit only has a 5 A battery charger; heavier chargers are optional. [I believe the 3.1 kVA unit has a 20 A charger.] No way a puny 5 A charger will serve for my proposed usage pattern. The UPS itself only showed a 5 A charge, as did my clip-on DC ammeter, so I guess I should believe the manual.)

I hand-drilled out the four heat-staked corners of the control panel, and removed the circuit board. Fairly basic, and largely surface-mount. Date codes in the 1996 timeframe. It has a LM3405T 5 V voltage regulator, a 1488/1489 RS-232 pair, a NEC D78C14GF-R75 processor clocked by a 14.7456 MHz crystal, a 93C46 64×16-bit serial EEPROM, ON Semi's 34064 reset generator, one CMOS 14024B 7-bit ripple counter, one MC1413D high-voltage NPN Darlington (7) transistor array, a TC1428CPA 1.2 A Dual High-Speed MOSFET Drivers 8-pin DIP, and a Mitsubishi M66004FP 16-digit 5×7-segment VFD controller. Plus the usual handful of R/C/D components. Oh, and a VFD display and a rubber membrane keyboard. Near as I can tell, the VFD's 37 Ω filament is AC-coupled and is getting fed around 0.2 V. There is a 1N4753A 1 W 36 V Zener diode that indeed has some 36 V across it, looks like it's being built with a voltage multiplier chain driven by the TC1428 clocked at 200 kHz. That's right in the ballpark for what the M66004 wants for grid drive. The filament voltage, though, is highly suspect. It looks like it might come out of the same voltage multiplier chain that the high voltage does.

The 78C10, 78C11,and 78C14 single chip 8-bit microcomputers integrate a 16-bit ALU, 4k ROM, 256-byte RAM, an 8-channel ADC, 16-bit timer/event counter, 2 8-bit timers, a USART and much more.

The 7810 is ROMless
The 7811 has 4K ROM
the 7814 has 16K ROM

A little surfing shows that it is customary for the two filament pins to be biased above ground so that the tube cuts off when it should, and that the net filament voltage should be around 3 V. (Your basic triode, in effect.) Well, we're seeing heavy ghosting and no brightness to speak of, so I'd lay the blame squarely on the filament drive. And that, at least, is exposed circuitry. This should be repairable, the filament itself is good.

I powered up the board on the bench, I clipped 12 V right across the input filter capacitor and the display powered up 'normally'. That's a whole lot better working environment than crouched over the opened-up FerrUPS out in the garage.

I did some more surfing, and found an Application Note for National's LM9022 VFD filament driver chip. This shows a typical circuit that vaguely resembles what's on this board, with (probably) the TC1428 taking the place of the LM9022. Everything else I could find assumed a center-tapped power transformer.

While I did all this I let the UPS charge its batteries. By the time I gave up for the night it had shut off the charging circuit, so I powered it down and put away the big 240 V extension cord. To bed!

Friday, September 19, 2008

I checked the TC1428 drive, pins 5 & 7, and it's OK. I put some 1 µF capacitors across the ones that were there at the filament connections and it improved brightness slightly, and it cleaned up the filament waveform. However, I'm not sure this is a good plan as it's still ghosty and I can now see a faint glow from the filament. All the diodes seem to be doing their thing, I'm not seeing any that appear to be dead. This is very puzzling, normally I'd expect a dead capacitor, semiconductor, or solder joint, but I can't find any of these.

The M66004 requires the –36 V drive (VP) in order to turn off the digit grids. (The segment anodes and grids are switched between this and +5 V, giving a proper net voltage for the tube.) The common-mode filament voltage needs to be somewhat less negative than VP or ghosting will occur. I need to look there next. I think I need to trace out the schematic of the voltage inverter and filament drive circuitry, it all keeps pointing back to there.

...After work I checked the drive levels out of the M66004, and the anode and grid levels both drop to within a volt of the VP negative supply voltage (–36 V). According to the 'scope, the balanced square-wave filament drive is approximately symmetrical around this voltage, let's call it –30 V to –40 V, which is not adequate to prevent ghosting, its average level needs to be several volts above the negative supply. I created a filament drive schematic by tracing the circuit. (This is tedious: it doesn't take all that long to trace out the connections, but getting it drawn with a meaningful topography usually takes several tries.)

Having seen the circuit for what it is now, I still don't see how the filament bias is really accomplished. Looks to me like it's missing a resistor or two, yet there is no sign of one.

Saturday, September 20, 2008

I spent some time making a partial schematic, in Illustrator, PostScript, PDF, and GIF formats. Thinking about the circuit I've decided that it's a non-traditional voltage quadrupler, inverting. That gives you around –36 V from its +12 V drive. Being non-traditional in form (not a Greinacher/Cockroft-Walton ladder) means that the diodes and capacitors are exposed to greater voltages than in a ladder, but because this is a relatively low voltage circuit (as opposed to multi-kilovolts) that is not an issue. I suspect that this form, with proportionally greater drive voltages on the capacitors, results in more efficient current production or something.

Anyway, the drawn circuit basically has three parts: the zener-regulated inverting voltage quadrupler, the AC filament drive, and the negative voltage (–15 V) supply for the RS-232 driver. The filament drive is what is interesting, it's taking the 12 V square-wave drive from U6 and coupling it through R8/C10 and R6/C9 to the 37 Ω filament. This results in a sufficient AC current flow through the filament to enable it to emit electrons. However, the filament as a whole has to be at a high negative potential or those elecrons won't actually be emitted. Optimally it's a few volts positive from the negative rail, which is what the anodes and grids are switched to while scanning. R5 and R7 together bias the AC-coupled filament to VP. Unfortunately I don't see how this bias stays above the negative rail, which results in ghosting. On the oscilloscope the AC filament drive is centered around the negative rail voltage, which means it's not positive enough for good operation.

Enter the modifications. Adding Ra and Rb to the circuit lifts the average filament potential (referenced to ground) a bit by pulling it towards ground/12 V, eliminating the ghosting. This lessens the cathode-to-anode voltage, however, resulting in less current flow in the tube and a dimmer display. By adding Ca and Cb in the circuit the filament current is pumped up a bit, restoring brightness.

The end result is a more usable display, but one that's still relatively cruddy looking. There's probably another failure mechanism at work, perhaps the VFD tube itself is wearing out. (I wonder if the normal expected anode operating current was supposed to pull the filament potential up a bit against its bias resistors? If the tube is now conducting less than normal that would shift the cathode potential down, causing the ghosting problem.) Regardless, it looks like these modifications will at least allow the unit to be used again.

I snapped the control panel back together again and put it on the UPS. Now it is indeed usable again!

Sunday, September 21, 2008

I was in a mood to tidy up a bit, so I dug out one of Jill's used rigid mailing envelopes and taped it inside the front cover of the UPS, with its bottom against the bottom ridge of the cover to minimize the strain on the duct tape. (Being taped down, and stiff to boot, means that it can't get sucked into the cooling fan even if the tape starts to loosen a bit.) I put all the printed material into it, and then put the panel back on the UPS.

Sunday, January 18, 2009

It seemed like it was time, so I charged the batteries on the FerrUPS systems. Got to do that periodically.

Tuesday, January 27, 2009

I nabbed a couple of big 48 V to 12 V power supplies that were discards (from an abandoned product line) from work. I may or may not ever be able to put these to use, but they'd drive low-voltage lighting admirably from a 48 V backup battery bank. (Which my four biggest UPS's all use internally [well, externally], how about that?)

Thursday, March 26, 2009

Another 'find', this time a 3 kVA UPS (APC Smart-UPS 3000XL) from an office (employing many former ISC people) that closed. Large (and heavy!) rack-mount unit with internal batteries, supposedly fairly new, and a 30 A power cord. Not exactly sure what to do with it, I like the rack mount (for future use) over the desktop APC 2000 that's currently in use, but the 30 A feed won't easily mate with my plans for the computer cluster, which I'd like to power by generator in its 110 V-only incarnation during power outages. We'll see. I stowed it in the garage and pulled its battery plug.

Saturday, April 4, 2009

With the advent of digital television and our new living room, the APC 450 has been redeployed to feed my battery-challenged (!) laptop computer and a floor lamp. We still have one DVR hooked up, but it's rarely used anymore (due to the necessity of using an external digital tuner) so I'm no longer concerned about it losing data.

Sunday, August 2, 2009

Moved the APC 2000 UPS to the new computer room, into the 'basement' cubby under the front landing. It's much more accessible there than it was in the bottom of the 19" rack. The APC 3000XL's 30 A plug actually mates with one on the back of the APC 2000, so it's plugged in there for the purposes of occasionally charging it. Which I did. (I leave its battery bank unplugged normally.)

Saturday, October 31, 2009

The power failed today and the 2 kVA UPS on the computer cluster (APC Smart-UPS 2000, with 75 AH external 48 V battery bank installed in 2006) failed to hold up very long, and when the power came back on the UPS couldn't restart until it had been allowed to sit 'off' (recharging) for several hours. The computers were also thus off for this time, since this particular UPS is always 'live'. One of the batteries had 15+ V on it, and when hit with the test load at startup it dropped to zero. Sulfated cell? The other three batteries didn't move from their 13.1 V reading during this test cycle. Sadly, one should not replace a single battery in such a set, the others are probably near their end-of-life too. Three and a half years' 'protection' out of the $350 for the battery bank? Not that great, for about $100/year. This UPS was intended to hold the computers up all day, or overnight. Not even close, this time.

Saturday, September 11, 2010

It was time again, so I charged the batteries on the FerrUPS systems. I hooked up a power plug to the big one, and got it on the wrong cable and was actually back-driving it into its outputs. It didn't seem to care much, but the control switching was weird. I put it right, no harm done apparently. The most disturbing thing is that the charger doesn't kick in right away, it seems to be on its own schedule.

Monday, September 13, 2010

I checked the APC Smart-UPS 2000 battery bank again today, and it's still bad, imagine that. Under self-test one of the 75 AH batteries drops to 7.5 V, the other three hold up fine. I need to replace that one with another good used battery. (I don't really want to replace all four with new as is generally recommended, that costs money!)

I also charged the APC Smart-UPS 3000XL.

Monday, September 20, 2010

Bought a used 75 AH sealed battery today. $40, w/90-day warrantee at Toby's. (That'll take it almost 'til Christmas.) It's obviously a different make than the others. I put it on charge. (They offer $6.50 each (in 10+ quantity) for old dead batteries. I have a lot, more than a dozen. I should take advantage of this.)

Tuesday, September 21, 2010

I installed the now-charged replacement battery, 'live'. (I didn't want to shut down the computers.) It seems to work. Its voltage is down a bit from the other three, maybe half a volt, but we'll see after it has a chance to settle in with the others. It holds up during the self-test cycle, that's better than the old one did!

Wednesday, September 22, 2010

Now that they've settled in a bit I tested the four battery voltages: the 'new' one is 13.51 V, the others are 13.78, 13.82, & 13.86 V. Significance?

Wednesday, October 13, 2010

Took the core back to Toby's, along with a dozen or so other dead batteries. (Most from a UPS at my old workplace that I grabbed when they replaced them. They'd finally all died and were just junk that was in my way.) Got $7.00/each for the non-cores, core was $10, motorcycle-sized UPS batteries 50¢ each. $117, total. It was worth the trip, and definitely helped clean out the garage.

Thursday, November 10, 2011

I charged the APC Smart-UPS 3000XL. (Memo to self: don't forget to flip on its rear-panel breaker switch, besides plugging in the battery pack and turning it on with the front-panel switch. I had it running on batteries for awhile and squawking before I figured that out. I guess it's been awhile.)

I also charged the FerrUPS systems. I don't know what it is, but the larger one's charger is definitely on its own schedule.

Last week the APC 450 started beeping and acting up, it was no longer functional, either. I opened it up and found one of the two 6 V 12 AH batteries was dead and swollen, the other measured and looked OK. Time for new ones! The last ones were bought in 2006, at Toby's, so I guess it's time. I hate that they die so regularly. I also hate that this UPS is very badly designed, mechanically, as you have to take it completely apart, removing both the circuit board and the transformer from the case, to replace the batteries. Ugh. Anyway, two new batteries were $63 over the counter, double-ugh.

Saturday, July 14, 2012

I was in the wiring closet today, adding a network drop for the downstairs TV cluster, and decided to finish hooking the main UPS (APC Smart-UPS 2000) up to the gear that needs it. I scared up an orange extension cord and snaked it out under the hatch door to feed the house's main network switches, that was pretty easy. I found another orange cord and cut it in half, then snaked it through the sill plate between the two wiring closets and wire-nutted it back together. I hooked the troll (sprinkler/lighting computer) up to the UPS this way, decommissioning the APC Back-UPS Office (280VA) that used to be on that computer. Its console terminal is also on the UPS, though it's never powered on. The sprinkler/light driver and the X10 interface remain on the regular power source.

Tuesday, September 30, 2014

It's been far too long, when I tried to put the two 3kVA reserve UPS's on charge neither was happy. Eventually both would start up, but I may have left these alone for too long. We'll see.

...Neither battery is in very good shape, but both did eventually take a charge. The problem child was, again, the 7kVA UPS, which purely refused to go on charge, even when left overnight. Its battery was at about 49 V, and stayed there. According to the little front panel, the charger was in "Dsbl/Off" mode, and I couldn't get it to change. Like I said before, the charger seems to have a mind of its own. The display seems pretty dim, I wonder if it's on its way out. I am unfamiliar with the failure modes of VFD's. (See my earlier repair attempts.)

Wednesday, October 1, 2014

I power-cycled the 7kVA UPS, with the battery circuit breaker disconnected this time, and it came up in an Alarm state as one would hope. Turning on the battery breaker didn't please it, so I power-cycled it again. This time the charger was in "Enbl/Off" mode, and after a time it went on charge. Finally! I went through all 140-odd display elements on the controller, and I found that there's a 240-second charger delay, #128. That 4-minute delay explains a lot of my general frustration with the charger, but does nothing to explain why it never went on charge yesterday. Anyway, it's charging now, at its rated 5 A value. The battery voltage was up around 52 V, and rising slowly. (The remote controller and the serial console are both set to 1200 baud, 8N1. Worth remembering for the future, if the little remote gives up completely.) I set the UPS to its LineCondition mode, so that it wouldn't discharge the battery if the power should fail during this charging exercise.

The FE7KVA model seems to be in production again, and its new retail price is around $8,500!

...After work the charging current was down to around 600 mA, I'll give it a chance to continue to drop before I shut it down.

Thursday, October 2, 2014

The charging current had dropped to around 200 mA by morning, so I shut it down and moved the power cord back over to the 3kVA unit. It started charging at around 18 A, and dropped off to around 11 A or so while I watched. It's apparently not done yet! Unlike the 7kVA unit it doesn't have an integrated display to show these things, so I used the Fluke clamp-on DC Ammeter to determine the charging rate. (I believe the 3kVA is just as capable as the 7kVA, and obviously has a beefier charger option to go with its larger external battery, but lacking the control panel you need a serial terminal to access the measurements and settings.)

...After work the charging current was down to around 1 A, I'll let it go 'til morning.

Friday, October 3, 2014

No, still charging at around 600 mA, I'll give it some more time.

...After work the charging current was down to around 400 mA, which is about 2× what I stopped the 7kVA unit at, but while the bank is about 2× the capacity I'll still let it go 'til morning one more time.

Saturday, October 4, 2014

Down to around 300 mA this morning. Good enough, I shut it all down.

Saturday, February 17, 2018

Recently I acquired an APC Smart-UPS 1500 at Goodwill, about $7. Naturally the batteries are toast. (Two 12V 17AH SLA batteries.) One was open-circuit, but putting 50+V on it got it to take a bit of trickle current. I eventually got both batteries to take enough current that the UPS would actually start up and pass its battery test, though it can't even power a 100W light bulb, and the hold-up time is dreadful. The batteries are badly sulphated. Mail-order can supply another pair for around $60, but I'd like to see if there's a way to recover the ones it has. I ordered a de-sulphating pulser from the internet somewhere, about $12 for the kit.

The idea is that if I can get this working again I can put it on my upstairs computer (MacPro2), which is located nowhere near the main UPS.

Wednesday, March 28, 2018

After some time of messing with the 1500's batteries, and getting them back to a dribble, it became clear that they'd never work again. So I shopped around and found a pair for about $60 online. They came a couple of days ago and I put them in, on charge. Yesterday they passed the self-test, so today I buttoned them up and installed the UPS on the upstairs computer cluster. Call it about $68 to have brownout protection on them. I may or may not hook up the USB cable.

The powerline network adapter can't/shouldn't go on the UPS, so even if the machines stay up they'll be offline. Oh well.

Tuesday, November 6, 2018

The MacPro2 machine has been moved downstairs to the computer room, releasing the APC Smart-UPS 1500 from duty.

Friday, November 16, 2018

I put the now-idle APC Smart-UPS 1500 on the relocated Xerox Phaser, which is definitely against recommendation. (Heating elements are high power! It definitely threw overload indications while the printer came up.) We'll see how this attempt goes, the printer was quite the strain on the UPS. We don't want the UPS for when it's operating, what I'm after is keeping the printer alive while it's asleep should the power go out. So long as the printer can operate OK through that UPS, when the power is on, we should be good.

Thursday, June 27, 2019

The APC 450 UPS in the living room seems to have given up the ghost. It might be the 'new' batteries, which are now 7.5 years old, but given that the reason for the UPS in the first place was the bad batteries in a Pismo laptop, which was retired years ago, there's no real reason to resurrect this UPS at this time.

Monday, December 21, 2020

I bought ($125.88 total via Amazon) a well-rated CTEK US 7002 battery charger to use on the idle UPS batteries. (Probably too little, too late.) It's supposed to recondition batteries, etc. I put it on the first battery in the FerrUPS FE7KVA. By evening it was 'done', and I moved it to the next (of four) batteries in that cabinet.

Tuesday, December 22, 2020

The second FerrUPS FE7KVA battery was charged, so I moved on to the third. This battery triggered the desulphation cycle, we'll see if it comes out of it.

Wednesday, December 23, 2020

Even after three attempts #3 wouldn't take a charge, so I moved on to #4. It actually seems to have a partial charge still on it. I got out the 'new' 5A bench power supply and attached it at maximum voltage (18+V, current limited to 2A) to see if it could jar #3 loose. It was taking ½A at that voltage, we'll see if it resurrects at all.

Thursday, December 24, 2020

By morning the terminal voltage on the bad battery was down to 16V, and it was drawing the full 2A. I think the voltage needs to come down a little more before the CTEK would engage with it. (The #4 battery claimed full charge by this time.)

By mid-morning the voltage was down to 14.4V, and the CTEK battery charger would engage. The battery would power a brake lamp with no problems. I put away the bench power supply, and left #3 on charge. Success?

I went around the four batteries again, and all acted like they were fully charged. (A brief cycle to full on the CTEK.)

Friday, December 25, 2020

The CTEK could do nothing with the battery on the diesel almost-generator. Too far gone, I guess. It did charge the riding lawn mower's battery with no problems.

Saturday, December 26, 2020

I moved over to the FerrUPS FE3KVA, which has 8 batteries in it, not 4. They all seem to be flat, under 2V, and the CTEK wouldn't engage with them. So, out came the bench supply again. I got #1 quickly to the point where the CTEK would start charging. #2 was more recalcitrant, starting with only 300mA at 21V. After many minutes the current crept up to my 2A limit, and the voltage started coming down. It's very intermittent, though, which is a bad sign. I'm sure that at least one of these batteries is destroyed, perhaps all of them are.

After some 45 minutes the voltage was below 12V, so I moved the bench supply over to #3, which came up to that point quickly. #4, ditto. I moved to the lower deck, and #6 was slow to charge. I really need to decouple the two decks for this procedure, as one deck can act as a parasitic load on the other as I bring up the charge piecemeal.

After several hours I had all but one of the batteries brought up to a 2A draw, and I disconnected #1 from the CTEK to see its state of charge. 7V! That is not good. All of those other batteries are 6V or less at this point, even after I got them taking current. I broke the circuit on the upper deck of batteries, that should effectively decouple them all from each other.

Sunday, December 27, 2020

#1 had dropped back down to sub-2V overnight, and #7, freshly taken off the bench power supply, was still only in the 7V range. #8 was still in this range, and the CTEK would engage against it—we'll see how it does there. I put the bench supply back on #1. These batteries are not healthy!

#8 failed to charge, and #1's voltage (at 2A) was dropping. I suspect that the whole upper deck's batteries are all toast, as #2 clearly has a cell that overheated badly in the past. I moved the CTEK to #7, and the bench supply to #5.

Monday, December 28, 2020

This is the third day, and none of the FE3KVA's eight batteries will take a charge. Not looking good.

Wednesday, January 19, 2022

Recently I paid $20 for an unused in-unopened-box Back-UPS XS 900. The box was very dusty, and at a thrift shop, and I had vague ideas of putting it on our internet access radio if it worked out, but the age was quite suspect. When I opened the box the 24V battery pack was dead-dead-dead, and wouldn't take a charge. At 24V, with no visible center tap, none of my usual battery chargers would work with it. I put the old half-wave rectified line voltage through a 4W light bulb charger/destroyer on it, and eventually the voltage across the battery dropped to 24V as the cells reformed. It still wouldn't operate the UPS, so I put the Heathkit IG-18 square-wave generator on it, set at its maximum (100kHz) rate and (10V) voltage, in series with a 30V power supply. According to the oscilloscope there was megahertz-level ringing on the edges, adding substantially to the prominent square wave voltage that was riding on the DC. (The Simpson 260 meter on the battery pack shows 24V at all times now.) Over the course of a week or more the amplitude of the square wave diminished, as the battery's impedance dropped, and the self-test of the UPS held up longer and longer. Today the UPS actually fully started up (no load).

Ever-hopeful, I put it back on my homemade desulfator. Maybe I didn't waste $20 and a bunch of time? The UPS came with all its cables and auxiliary operating software. PowerChute? It has a USB port rather than the older relay contacts.

Saturday, January 22, 2022

While getting ready to try a load test I managed to drop the XS 900's battery pack, and it went open-circuit on me. I peeled apart the battery pack, and found one of the two standard batteries was still good. (I tested it in a large flashlight, and it was happy.) The other battery shows intermittent connection if I bang it down on the workbench, but I think I have to admit failure. Doh!

It's looking like LiFePO4 batteries are just about to the point where I could consider upgrading to them from lead-acid batteries. They are lighter and have more life cycles, and can be discharged further without damage. But, still a bit spendy.

I revisited the APC 450 UPS, last used in2019, and its batteries (in service for 7.5 years, and idle for another 2.5 years) were completely open-circuit. (I.e. solidly sulfated.) I tried the nasty-ass charger, and it was getting a dribble of current in at some 80V. After a couple of hours it was lighting the bulb pretty well and the batteries (2× 6V in series) measured in the 12V range. I then connected the fancy CTEK US 7002 to it, and it recognized that the batteries needed de-sulfating (flashing orange light) and it entered that operating state. I guess we'll see.

Return to Site Home