Tim Shimeall <tim@uci-icsd> (02/09/85)
Can anyone give me information on how to build voltage protection units for micros? Specifically: is it difficult to put together a unit which will protect my system from voltage dips? I have some information on surge protection, but nothing on dip protection. Is dip protection too difficult for an amateur electrician? Would it be better to bite the bullet and purchase a comercially available unit? Any suggestions/experience as to which commercial units work well? Tim
Sam Hahn <Samuel@SU-SCORE> (02/09/85)
I have a related question-- I've investigated UPS's for my Compupro, a rather large and sensitive beast, but those I've found suitable seem invariably to be priced around $1400 to $2000. I have no power engineering background. Could someone please explain to me why an uninterruptible power supply necessarily is so expensive? I've been targeting UPS's with fast switching time, or else one which is on-line continuously. I'd like one which can power my Compupro, which s fairly loaded with 12 boards, and my Quantum 40 MB disk, minimum. I can live without my printers or terminals , though 1 terminal is a necessity. Among the names I've looked at are: Clary, Sorensen, Computer Power, Power General, Best Energy, Frontier Technologies, LorTec, Isoreg, Electro Vector, Triad-Utrad, Elgar, Topaz, and Jefferson Electric... Thank you. -- sam hahn [software weenie]
john@hp-pcd.UUCP (john) (02/10/85)
<<< A good UPS does not come cheap because it is not a trivial task to generate a high power 60 hz sine wave in phase with what the power company is no longer supplying. Power supplies take a lot of stress between a varying source and a unpredicable load and they need a wide stress margin to handle it. The annoying thing about UPS's is that after you go to all the trouble to convert your battery power to 110 VAC is that the computer turns around and converts it back to +-12 and +- 5 VDC to do most of the work. It would be a lot simplier if computer power supplies were designed with a 12VDC inter- mediate voltage that the user could attach to a Diehard for battery backup. This approach would be more expensive for the user who didn't want a battery backup but a lot less expensive for those who did. As CMOS prices drop down we may even see a move toward computers that require only 12VDC from a wall transformer. A system like this would be easy to backup. John Eaton !hplabs!hp-pcd!john
crm@duke.UUCP (Charlie Martin) (02/10/85)
There is a way to do an uninterupptible power supply that will work for short periods that I have always wanted to try, but never had the need/ambition... what you would do is find the cheapest 110/60 alternator you could, and a cheap electric motor, and build a GREAT BIG HEAVY mucking flywheel; drive the flywheel with the motor, then drive the alternator with the flywheel. When the power was working OK, then you have a mechanical loss between the line and the load -- so it costs a little more to run the thing. When the power changed over time (voltage or freq) then the flywheel's inertia would smooth the change. In addition, stored energy of rotation would provide a short-term source of power which would keep you up at least long enough to save and shut down. I've never tried it, as I said -- if you do and it works, let me know. -- Opinions stated here are my own and are unrelated. Charlie Martin (...mcnc!duke!crm)
johnl@ima.UUCP (02/12/85)
A motor-generator with a heavy flywheel should work fine, so long as you don't blow all of your fuses while spinning the flywheel up to speed. This is the traditional approach to UPS, by the way. IBM provided such a motor-generator standard with each 360/91. John Levine, ima!johnl
doug@terak.UUCP (Doug Pardee) (02/12/85)
> what you would do is find the cheapest 110/60 alternator you could, and > a cheap electric motor, and build a GREAT BIG HEAVY mucking flywheel; > drive the flywheel with the motor, then drive the alternator with the > flywheel. Oh, no! Let's please not bring back the MG (motor-generator). I listened to the racket of those infernal things for enough years back in the "good ol' days". Between the MG's, the 1200 card-per- minute card reader, and the 1000 line-per-minute line printer, I dang near went deaf. -- Doug Pardee -- Terak Corp. -- !{hao,ihnp4,decvax}!noao!terak!doug
dtynan@cadvax (Dermot Tynan) (02/13/85)
[Eat your heart out!!] Well, you've hit upon a very large issue there. The field of UPS's (Uninterruptable Power Supplies) is pretty extensive. There are a myriad of issues involved with power conditioning. One can start with an EMI filter (Electro-Magnetic Interference), and work up to a UPS. Basically, an EMI filter extracts the noise from the AC line. Although it could be said that it also helps OverVoltage and UnderVoltage to a certain extent. I bought a power strip with four outlets that had an EMI filter built-in. I can't remember the manufacturer, but I *do* remember it costing somewhere in the $50-70 range. After hooking it up, and four hours editing I got my money's worth when the refrigerator kicked in, and my terminal got zapped (it wasn't plugged into the power strip). The system stayed up, without even a glitch! Basically, an EMI filter will cut out such things as the "snow" one see's on a TV when some electrical gear switches on. A "brownout" from the power company is, however, a different story. A UPS on the other hand, will retain the power anywhere from 10 minutes to two hours. This is similar to a battery-backup (which is essentially what a UPS is!!). In order to find out what exactly you need, you have to analyse your current setup. For example, how much current are you using, what devices or peripherals are you worried about, etc. Normally, it's not feasable to plug such things as printers or terminals into a UPS, as the power requirements go way up. Most computers (should I say *some*??) have a rudimentary EMI filter built in. I would advise that you check, though. Also, a lot of today's personal computers use "switching power supplies" (ie, Apple, IBM). Switchers can adjust for a low voltage in a relatively short amount of time. Also, a large electrolytic capacitor on the voltage rails will cure undervoltage of a short duration (again, the values in question have to be determined). As for a UPS, I would suggest use of one of these, if you suffer from a lot of junk in your 110v supply. This interference is normally due to a DIY neighbour with a power drill, or a nearby industrial park. In this case, I would advise a call to your local power company, as they may be able to help you, by putting you on another phase (this should cost less than a UPS). The price of a UPS is generally governed by two criteria: 1/ Supply holdup time (how long the computer stays up, after the lights go out!). 2/ Current/wattage requirements (how much you have hooked up, and how much current does it eat). Generally UPS's are used by people who don't want their system to crash in the event of a power failure. Examples would be banks, hospitals, etc. Your best and cheapest solution would be to purchase an EMI filter and go with that. If you want to build your own, take a trip down to Radio Shack, or your local electronics dealer, and ask for an EMI filter. Installation is pretty straightforward. Otherwise, most computer stores will sell you a power strip with built-in EMI filtering. Be very selective, there are as many specifications as there are filters. I don't expect this note to explain away your power problems, but I hope it gives you an idea of the different techniques. From my own experience most power problems with personal computers are due to EMI rather than severe brownout (the power supply *should* take care of minor ones). If you have any questions or problems, send me mail, but be a bit more specific. - Der Tynan
munck@mitre-bedford (02/13/85)
> what you would do is find the cheapest 110/60 alternator you could, > and a cheap electric motor, and build a GREAT BIG HEAVY mucking > flywheel; drive the flywheel with the motor, then drive the alternator > with the flywheel. Seems to me we ran an IBM 370/155 that way back in '73. I think it also needed AC at something other than 60Hz; maybe 400? Of course, this was easy with the flywheel: just use a 400Hz generator. Maybe you could use a 5V DC generator and eliminate your power supply. I question the "GREAT BIG HEAVY" flywheel. Energy storage with flywheels is going in the direction of relatively small, light wheels made of exotic fibre materials and spinning at fantastic rates (100,000+ rpm) on magnetic bearings in a vacuum. Energy is added and taken out by magnetic coupling. Of course, it's all in an armored box to handle the shrapnel when the wheel breaks up. Consider what happens to the "BIG HEAVY" wheel when a bearing seizes. -- Bob Munck, MITRE
jbn@wdl1.UUCP (02/14/85)
There's nothing wrong with that idea; there is commercial gear for that purpose, but it tends to be on the large side and noisy. Some years ago I worked at a Univac mainframe site which had such a rig; the MG set was a vertical-spindle unit about 3' in diameter and 4' high, delivering about 50KW or so. Being synchronous (3 phase), it would ride over dips in the power line and totally isolated the computer from nearby source of electrical noise, which included - heavy arc-welding equipment, - a magnetic-squeeze former (a device for squeezing metal tubes, bands, or bearings onto shafts using giant magnetic pulses produced by discharging a large capacitor bank into coils of copper busbar), - a test stand for locomotive transmissions driven by a 1000HP electric motor and loaded by a water-cooled Prony brake which cycled through start-forward-stop-reverse cycles at about one per minute. We had NO power problems at that site. John Nagle
peterb@pbear.UUCP (02/14/85)
Most bearing types will give ample warnig that they are going to seize. this can be high heat or shrilling or in real bad cases, smoke.
rf@wu1.UUCP (02/15/85)
Charlie Martin (mcnc!duke!crm) writes:
There is a way to do an uninterupptible power supply that will work for
short periods that I have always wanted to try, but never had the
need/ambition...
what you would do is find the cheapest 110/60 alternator you could, and a
cheap electric motor, and build a GREAT BIG HEAVY mucking flywheel; drive
the flywheel with the motor, then drive the alternator with the flywheel.
This is the oldest uninterruptible power supply. It is known as a
"dynamotor" or "motor-generator set". They work well, but are expensive
and, well, mechanical. These days they are mainly used in places with HUGE
power requirements which require AC power and can't afford any power
outages; big computer centers, mainly.
Usually, dynamotors are built with diesel engines connected to their common
shaft via clutches. When power fails, the energy stored in the flywheel can
be used to start the engine & there is not even a change in the phase of the
output current.
Randolph Fritz
UUCPnet: {ihnp4,decvax}!philabs!wu1!rf
glen@intelca.UUCP (Glen Shires) (02/16/85)
> A motor-generator with a heavy flywheel should work fine, so long as you > don't blow all of your fuses while spinning the flywheel up to speed. > > This is the traditional approach to UPS, by the way. IBM provided such a > motor-generator standard with each 360/91. > > John Levine, ima!johnl I seem to remember some big mainframe that used its hard disks as a "motor-generator with a heavy flywheel". For a few milliseconds after a power failure, it could generate power from the motion of the hard disks and have just enough time to dump core onto one of the disks. -- ^ ^ Glen Shires, Intel, Santa Clara, Ca. O O Usenet: {ucbvax!amd,pur-ee,hplabs}!intelca!glen > ARPA: "amd!intelca!glen"@BERKELEY \-/ --- stay mellow
robertm@dartvax.UUCP (Robert P. Munafo) (02/24/85)
> I question the "GREAT BIG HEAVY" flywheel. Energy storage with > flywheels is going in the direction of relatively small, light wheels > made of exotic fibre materials and spinning at fantastic rates (100,000+ > rpm) on magnetic bearings in a vacuum. Energy is added and taken out by > magnetic coupling. Of course, it's all in an armored box to handle the > shrapnel when the wheel breaks up. Consider what happens to the "BIG > HEAVY" wheel when a bearing seizes. > -- Bob Munck, MITRE Remember that the size and mass of the flywheel really does not matter. Whether it's big and heavy or small and light, it is still going to be used to store the same amount of energy, and so the same amount of energy will be released when the flywheel fails. -- Robert P. Munafo ...!{decvax,cornell,linus}!dartvax!robertm
henry@utzoo.UUCP (Henry Spencer) (02/25/85)
> I question the "GREAT BIG HEAVY" flywheel. Energy storage with > flywheels is going in the direction of relatively small, light wheels > made of exotic fibre materials and spinning at fantastic rates (100,000+ > rpm) on magnetic bearings in a vacuum. ... Quite true. But try to buy one! Great big heavy flywheels are available off the shelf; fast high-speed composite flywheels are generally found in research labs, not suppliers' warehouses. -- Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,linus,decvax}!utzoo!henry
jc@sdcsvax.UUCP (John Cornelius) (03/08/85)
Well over 90% of the voltage abnormalities that will ruin your day (and your computer) are either over/under voltages of magnitude <nominal>+- 50% or spikes/dropouts/noise lasting less than 20 milliseconds. IBM did a study about 7 years ago and concluded that the ability to ride through 2 cycles of peculiar power without the peculiarity reaching the machine would take care of about 90% of the power related maintenance calls. The general form of such protection is with isolation/regulation transformers. These fall into 3 types: Transformers that do one or the other Multitap isolation transformers that switch secondaries with SCRs Ferroresonant continuously regulating isolation transformers. My prejudices lie with the last of these because they are superior to switched designs in extreme cases of voltage abnormalities and those are the ones that make maintenance companies rich. The maintenance company that we use buys Isoreg transformers from us and installs them on micros that they are maintaining under contract. When the contract goes away, so does the Isoreg transformer. When the customer objects he sells it to the customer at 130% of his cost. He claims that these things get rid of virtually all of the undiagnosed intermittent problems. The Isoreg product goes for about $.50 per watt for 750VA and above and a little more for smaller units. They tend to run warm and the larger ones have fans for cooling. They are heavy and not particularly pretty but if you have irreplaceable data on your disk they are worth their weight in gold. Isoreg is in Littleton Massachusetts. Their phone number is (617)486-9483. John Cornelius Western Scientific