Stabilant 22 wins Byte's "Product-of-the-Year" Award.

In the May 1985 issue of BYTE magazine, we were pleased to note that Stabilant 22™ had won Dr. Jerry Pournelle's "Product-of-the-Year" award! Unfortunately, it was identified by the home-stereo-industry name of TWEEK™, rather than the correct name, Stabilant 22.

While that's been corrected, here's how it looked in that first review:

BYTE-Chaos Manor BYTE - Chaos Manor banner


Many magazines have special Product-of the-year features this month. I'm a mite late with mine. Of course, my rules are a bit different from other people's. I pick the products I like best. Purely subjective. And "year" doesn't necessarily mean it came out in 1984, only that I acquired it then, With those ground rules, here goes.


First choice, hands down, is the Hewlett-Packard HP 2686A LaserJet printer. That sucker has changed my entire life. It replaced the big and noisy Diablo 1620. The Diablo has served me well in the past eight years, and it's still in good shape, but I'll probably donate it to a aerospace organization because.....


The second product of the year goes by the unlikely name of Tweek, which advertises itself as a "contact enhancer. It's a clear liquid you dab into places where you suspect you're getting bad contact, IC sockets, edge connectors, RS-232C plugs, that sort of thing. It comes in a kind of hypodermic syringe with a thin flexible tip instead of a needle, making it easy to get into hard-to-reach places.

My first use was on a sticky Reset button. I didn't bother to turn off the machine, just pulled the top off the button and squirted the merest drop. VOILA!

Last week my telephone started to make horrible static noises which could be cured by violently shaking the instrument, and they'd come back. I took it apart, took off the plastic cover over the little relay contacts activated by hanging up the phone, and squirted. The noise went away. Faulty TV remote controller: squirt. Now in an audio system: squirt. So far, Tweek has cured about a dozen annoying problems. A little bit of the stuff goes long way. Get some. You'll love it.

© 1985 Dr. J. Poumelle

And two years later in the April 1987 issue of Byte, we were again happy to see our product reviewed for the second time. Again it was called TWEEK (TM) rather than Stabilant 22

In this issue he is writing about changing ROM's in his Kaypro "Big Kat" ......

Now is was about to change ROM'S. Again I'm a worrier: there are four sockets but only two ROM chips, and once you have pulled out a chip, you'd be amazed at how quickly you can be confused about where it used to be. Or at least I can be confused, and I'd rather not be. One possibility would be a Polaroid snapshot, but this is simple enough for a pencil-and-paper diagram.

I used a chip puller; getting down inside the machine with a small flathead screwdriver to pry out the chip isn't easy. Pulling the chips was no problem. Putting them in nearly drove me crazy.

I did everything right. Examine the chip. Make sure that all the pins are straight. Put it carefully in place and look again to be sure the pins are fit in the socket, and then get out a powerful flashlight, and look again. Then push in carefully and watch several pins buckle and bend.

Don't panic. Out comes the flathead alligator pliers. Ground myself carefully. Straighten the pins carefully. Now it has to go back in -

Which was where I got smart.

Two years ago I gave a product of the year award to Tweek, a contact enhancer fluid that solves all kinds of problems with PC's. It came to me as vision that this was exactly what I needed here. I still have a nearly infinite supply, - a little of that stuff goes a long way - so I fished it out of the tool cabinet and ran a bit of it along the newly straightened pins.

Presto!, the chip went in with no effort at all. In future I'll never insert chips without first lubricating with that stuff.

© 1987 McGraw Hill Inc.

We decided that once again we had better write to Byte to see if we could get the matter of just who was manufacturing the material listed in the credits, and so we sent off another letter. This was printed in the August 1987 issue of Byte on page 30 ......


Conducted by Jerry Pournelle

Say Stabilant, Please

Dear Jerry,

Thank you once again for your kind mention of Tweek in BYTE. The last time I wrote was after your "Product of the Year" award. I enclosed several samples as well as a plea for a correction. I must assume that someone mislaid the samples and letter, either in the Canadian Postal System, or at BYTE's editorial offices.

My company, D.W. Electrochemicals Ltd. manufactures this patented material and sells it throughout the world as Stabilant 22 (the concentrate you reviewed a couple of years ago) and Stabilant 22A (an isopropyl-alcohol-diluted form). Sumiko Inc. (which owns the trademark Tweek) purchases the product from us and repackages and re-labels it for distribution in the home audio field.

As I started work on this class of material in 1975, and as we have spent several hundred thousand dollars in R&D since that time, seeing some other company listed as the source (and by inference, the manufacturer) of the material is, to say the least, distressing. You could contribute greatly to our personal and corporate sanity if you could mention that we make the stuff.

We sell this material not only in the computer field, but also to radio and television stations, telephone and cable companies, and avionics and airframe manufactures (to name but a few customer areas); and we are NATO suppliers as well. At present we are the only company making a material of this type.

We supply Stabilant in sizes of from 1/2 milliliter to 1 liter. The most popular form is a 15 ml Stabilant 22a service kit, retail for about $36 US; the 15 ml Stabilant 22 concentrate bottle retails at about $102.00 US. Our address is:

(This was valid until 1991 when we expanded.... and so I wont bother to list it.....)

Okay, and apologies. I first got the stuff as "Tweek" and it was as Tweek that I used it for a long time. When I got the package with two more bottles from you either the difference wasn't clearly explained or I didn't notice.

Anyway, the stuffs wonderful I used it to lubricate the gate array for inserting a math chip into the Kaypro 3 Indeed, it goes on nearly everything I put into the computers. I suppose I ought try it on my hi-fi stuff, too. Oddly enough, never thought of that. - Jerry.

© 1987 McGraw Hill Inc.

Thank you again for the kind mention

William M Wright, President

D. W Electrochemicals Ltd.

Ontario, Canada, L4C 3G4

We took the liberty of correcting a typo in the suggested US retail price of the 15 ml Stabilant 22 in the above.

In the August 1991 issue of Byte Dr. Pournelle mentioned Stabilant 22 once more. He is commenting on problems caused by changing the hard disk drive in his Cheetah 386/5


ON THE ROAD AGAINJerry Pournelle - 2

Jerry takes a new computer out where the sun shines and upgrades his hard disk drive controller card

This time, we're headed for Flagstaff for the meeting of the advisory board of the Lowell Observatory. This gives me a good opportunity to test a new notebook: I'm writing this column on a Panasonic Business Partner CF-270 which is a lightweight, no-frills 286 notebook computer with a 20-megabyte hard disk drive and a 3 1/2-inch floppy disk drive. The battery pack is removable ................

Not Broke, But...

I suppose there are other hard disk drive controller companies, but for the past few years, the two that have interested me most have been ...

For the past several years, my main system-the one I write nearly all my books and columns with, has been a Cheetah 386/25 with a DPT controller and a Priam 330-NM ESDI hard disk drive. That system has been in continuous use, being turned off for installation of new boards or if I'm going off on a trip. While there are many faster systems in the house, the Cheetah 286/25 has been plenty good enough, the DPI' controller has performed splendidly, and I've had very little trouble with the systems you'd expect, since I make my living with it.

However, the Cheetah 486/25 came with a PSI controller, and the combination was a great deal faster than the older Cheetah 386/25 had been; not surprising, given that the one was a great deal newer than the other, and a 486 is more efficient than a 386 anyway. Still, my intention was to check out a DPT controller with the Cheetah 486/25; alas, DPT kept sending me ESDI controllers, but the hard disk drive I want to use with it on the 486 is a Siemens 800-MB drive, which is a SCSI system. I do have a DPT non-caching SCSI controller, which works fine but, of course, is not anything like as fast as the caching systems.

One day, I'll get things straightened out to have a fair test, but I already know that both DPT and ]PSI performance is plenty good; and I'm not really interested in fine-detail benchmarks. That's for the BYTE Lab.

Then when Larry Alderidge put together my test copy of the Cheetah 486/33, he used a PSI controller. Finally, PSI wanted to demonstrate their newest controllers, boost ESDI and SCSI, and in particular to demonstrate mirroring.

Mirroring involves two hard disks; every write operation is done to both of them, with the result that you have an automatic backup at all times. The notion was to install a mirroring system on my main machine, the Cheetah 386/25 and that was just attractive enough that I overcame my reluctance to mess with a working machine. The result was that Warren, vice-president of R&D at PSI, got an appointment to come to Chaos Manor to do the installation.


The prospect of changing the hard disk drive and controller on the machine I write with was a bit daunting; still, I didn't see how any disaster could come of it. At the moment, Chaos Manor is awash with machines, any one of which would be good enough as my main machine. There are two Cheetah 486!s: one. the 486/33, is still in the experimental setup stage, but its certainly reliable enough for my usual work; and there's always the older Cheetah 486/25. In addition, there's one Arche Legacy 386/33 that's very solid and an Arche 486 that we've had long enough to have confidence in. And if those weren't enough, there are a few older machines, including a Zenith Z-386/25 that's rock solid.

Clearly, some of those machines have to go back, if for no other reason than to make room. But any one of them would let me do everything that I do on my main system.

First things first, then: make backups of everything on the Cheetah 386/25. Doing that ought to be simple enough: for years, I've been doing XCOPY *.*/s/m onto the Maximum Storage APX-4200 WORM (write once, read many times) drive that's part of the Cheetah 386/25 system. I not only have all my current files copied onto a WORM cartridge, I have older versions of everything as well, since when you copy over a WORM file you don't actually overwrite anything and the previous file is retrievable. Furthermore, WORM disk files are essentially eternal, no one has seen any deterioration of WORM disks in the past decade.

There's only one problem with WORM files: I filled up a WORM cartridge some time ago, and since I have been copying with the /m parameter (meaning that it copies only files that have been written to since the last /m copy was made) I don't have all of the 250 MB on the new cartridge.

It would be possible to copy all the stuff off the older WORM cartridge and then overwrite with the files from the new one, thus assuring that I restored everything to the way that it was before we started. But that did seem like a lot of work. Time to think of something more ingenious.

As I reported last month, we have the Pioneer optical read/write disk drive connected to the Arche Legacy 486/33. The drive is on a SCSI daisy chain with Pioneer's MiniChanger CD-ROM drive, with both controlled by a Corel SCSI controller. It all works fine. So, I thought, why not make use of the Pioneer optical read/write disk system? That way, I could put everything from the Cheetah 38W25 onto one Pioneer WORM cartridge, and if anything went wrong, I'd have my reliable Maximum Storage WORM cartridges as well.

In fact, thought I, while I'm at it, Ill also make a Pioneer WORM cartridge, thus having yet another backup of the system.

First, though, I decided to make a copy on floppy disk of the very latest stuff I'd done on The Mote Around Murcheson's Eye, the novel Larry Niven and are currently working on. That way, I'd be able to only copy it onto the machine Niven works with, but also let him take a copy home.

Interlude: Clean That Machine

I couldn't make a floppy disk copy. I got "Drive Not Ready" errors. Retries did no good at all. I got out Desqview. That didn't do it. Reset the machine. No joy! Got out MicroClean's head-cleaning kit and tried that. Nope. Then I opened the machine and looked into the floppy disk drive - where I discovered an enormous fuzz ball, perfectly placed so that the disk head couldn't seek back to the home position. I remove that with forceps. I got out the compressed air can that comes with the MicroClean kit and blew out the dust and grime from every cranny of the drive.

Now the drive worked, but the machine didn't want to boot. Shake the boards a bit, move the cable around, and eventually all is well, including the floppy disk drive.

Backing Up

Next thing then, I got out an Inmac blue 25-pin cable connected one end to the printer port of the 386/25 connected the other end to a sex changer, and connected that to one of the new LapLink III "designer parallel cables. This put the two machines about 16 feet apart. I got LapLink III running on both machines and blasted everything off the Cheetah 386/33 onto the Pioneer optical read/write disk drive on the Arche Legacy 486/33, which, incidentally is running DR DOS 5.0. LapLink III in parallel turbo mode is really fast, but even so, this took a couple hours, which was all right because it gave me an excuse to play Wing Commander on the Cheetah 486/25 while things were transferring.

Everything went perfectly. I now had an optical read/write disk of everything from the Cheetah 386/25.

A SCSI Too Far

My next move was to make a WORM copy, but this presented a problem: I have only one Pioneer multi purpose drive. Clearly, I wasn't about to sit there a swap disk cartridges for a 250-MB transfer!

I have only one Pioneer drive, but I do have two Core interface boards, and I have lot of extra space on the Siemens 800-MB hard disk drive in the Cheetah 486/33. All I had to do was XCOPY everything to the optical cartridge to the Cheetah hard disk a then go the other way onto a WORM. As it happened, the second Corel interface board was already installed in the Cheetah 486/33, so all I had to do was physically move the drive.

Alas, it wasn't that easy. Chaos Manor has become so cluttered that there was no place to put the optical drive. Worse, it's in a daisy chain with the Pioneer MiniChanger, and the optical read/write disk drive isn't terminated; so I'd have to move both the MiniChanger and the optical read/write disk drive, or I'd have to disconnect the daisy-chain cables and install terminating resistors.

I muttered something to the effect that whenever I want to do anything I first have to do three other things and one of these will be impossible. Then I had a bright idea. Just how long can you make a SCSI cable anyway?

I got out my 10-foot Inmac blue 25-pin shielded cable and hooked it up. Incidentally, I've had that cable since CP/M days when I used it for RS-232 data transfers. Imnac prices are not cheap, but they sure do make good cables, and they deliver them fast.

Anyway, I hooked the Inmac cable to the Cheetah 486/33, thence to the sex changer, and plugged the Corel SCSI daisy chain into that. Fired up the Cheetah 486/33. Noted that I have drive E (the optical read/write disk drive) and drives F, G, H, I, J, and K (the MiniChanger CD-ROM drives). Created a subdirectory called CHEETAH3 (for the Cheetah 386/25), did "XCOPY E:\*.* D:\CHEETAH3\/s/e/v", and stood back. Data flowed. The whole thing was over in 30 minutes.

I didn't really fancy reformatting the WORM cartridge over 13 feet of cable, so I connected the daisy chain back to the Arche Legacy 486/33 and used the Corel utility. That done, I connected the drives back to the Cheetah 486/33 with the Inmac cable and did "xcopy D: \*. * E:\CHEETAH3\/s/e/v". Once again, data flowed.

Alas, it didn't work perfectly. About 100 MB into the operation, the system hung up. It took a reset to get out of it. At that point I could read the WORM drive just fine, but I couldn't write to it; I got a "Disk Full" error, although there was plenty of room on the cartridge.

Corel furnishes a bunch of utilities, one of which is a program called Repair, which fixes a damaged WORM. Using it is simple and that fixed the WORM at a cost of about 2 MB of data.


At this point I had a mild dilemma: I'd sent about half the data from the Cheetah 48(y33 to the WORM. I sure didn't want to do the same thing all over again; at the same time, I didn't want to send over another 100 MB one directory at a time. The obvious thing was to erase all the data that had been previously sent from the Cheetah 486/33 and copy the remainder to the WORM. That meant erasing about seven major directories-but each of these had subdirectories, and most of the subdirectories had subdirectories, and so on ad infinitum. I would be all day erasing those files, and, alas, DOS, Norton Commander, and SWEEP won't let you erase a directory that has data in it.

Fortunately, I have a pile of DOS shells that have come in for a review. The top one of these was Tree86. A glance at the package showed that it would indeed kill non-empty directories. Installation was simple enough. You just copy the disk. So far, so good. If you have to use a manual on a DOS shell, the designer was unclear on the concept. Tree86 passes the test. It was a snap to invoke and simple enough to find out how to kill directories with all subdirectories.

There are a lot of DOS shells and disk management utilities. The one I use is Norton Commander, in part because it has an MCU Mail utility that really works. But Norton Commander will not kill a directory, nor will it transfer a directory and subdirectories from one place to another in one operation. Consequently, I have now added Tree86 to the arsenal. So far it works fine (with and without mice), and I have had to look at the manual. There are far more complex disk management programs around, but Aldridge's Tree86 with Norton Commander is good enough.


Once I'd killed all the directories that had already been transferred, I let XCOPY do its thing on the remainder, and everything transferred to the WORM cartridge without further incident. So. All done except for one thing.

First test: see if I can run Wing Commander off the WORM cartridge. Given the 13 feet of cable in the SCSI line I was prepared for it not to work, and it didn't; the machine hung up. No problem, reset, and now copy Wing Commander from the WORM to the Cheetah 486/33 hard disk. Now run it ....

It blew up. Locked up the machine.

This was a bit frightening. Did I have a corrupt copy? Had all that backup effort been wasted? Well, I also had a good copy of Wing Commander on the Cheetah 486/33. I'd been playing the game while LapLink transferred files. So, just to be sure that nothing had happened to the machine, try that one ....

It blew up too.

That took a few minute's thought, and then enlightenment came. Wing Commander is a notorious memory hog. What had happened was that with the Corel optical read/write and CD-ROM drivers activated by CONFIG.SYS, the '486 didn't have enough memory left! Edit CONFIG.SYS to REM out loading the Corel drivers, reset, and invoke Wing Commander-there it was, in all of its glory. No problem.

Which got me to thinking, because, as you'll recall from last time, I could run Wing Commander off the hard disk on the Arche Legacy 486/33, but as soon as I tried to run it off the Pioneer optical read/write disk drive, it died. Everything said "timing errors", and we all let it go at that; but in fact, that's not right. The Arche Legacy 486/33 with DR DOS 5.0 has, even with the Corel optical read/write and CD-ROM drivers installed, just enough memory to run Wing Commander; however, when you try to use the optical read/write disk drive, it grabs just a bit more memory-and that's memory Wing Commander is also trying to use. The result is a crash.

The remedy is simple enough. The Corel software will load into expanded memory if you tell it to. Then you have to make sure that everything else goes into high memory, so that you will have a good 580 kilobytes available (with DR DOS 5.0, I was able to get the optical drive and the mouse, and still have over 600 KB). After that, you can run Wing Commander right off the optical disk, either read/write or WORM. Alas, MSCDEX, the DOS extensions that let you access the CD-ROM drive, take up so much memory that you're back to 572 KB of usable memory, and that's just enough. You can run Wing Commander from the regular hard disk but not from the optical drive. However, I'm still fooling with this, so stay tuned: I may yet find a way.

In any event, the Corel SCSI system works just fine with a 486, under both MS-DOS 5.0 and DR DOS 5.0. My next step is to get all that running with LANtastic; then I can access all those assets from another system entirely, and it won't matter how much free memory the system with the CD-ROM and optical read/write drives has. But that's for another time.

The Cheetah Operation

Now that I had everything backed up - three times - I was willing, if not precisely comfortable, to have Warren Lee open up my main machine and replace its hard disk drive controller. The Cheetah 386/25 has been stable for several years. Its a tower-configuration machine, so I don't have much reason t notice it as long as it works-and Big Cheetah has always worked.

When I cleaned the dust out of the floppy disc drive, I did take the vacuum cleaner to the rest the machine, but I'd obviously done a cursory job. Once Warren got the cover off and the machine laid on its side, it was obvious that we had a major cleaning job to do. There were dust balls everywhere. The cleanup took half an hour.

Then it was time to reformat the Priam 330 hard disk drive. I've had that drive for three year with no glitches; since it has been heavy use all the time, statistics say that it ought to be replaced. I'd a soon not do that, but I suppose I'll have to. On the other hand, the PSI media analysis and formatting program has found essentially no bad sectors, which is pretty remarkable considering just how hard the machine has been used.

I had toyed with the idea of installing either Dr DOS 5.0 or the final test release of MS-DOS 5.0 replace the PC-DOS 3.3 I'd been using. But at the last minute I decided not to, and we installed PC DOS 3.3 again. Formerly, the DPT controller for matted the disk out to drives C through L with 3 MB each, plus drive M with a few megabytes left over. The PSI controller formats only to drive through K, plus L with 24 NM. This has to do with the sector sizing and error-correction codes, none which I pretend to understand. However it is a 8 per cent loss in usable disk space.

That wasn't good news, but then there was worse the, machine wasn't booting properly.

Like the Old Days

First possibility: the battery. It hadn't been change in a couple of years, so it was time anyway. The Cheetah came with a holder for four AA batteries but I've for some time had a Raytheon lithium computer battery, which is supposed to be a lot better and more reliable. We installed that.

Still problems. The machine wasn't acting right Warren Lee looked a bit nervous. I was thinking about which machine I'd use to get the next day' work done. Still, it was a bit early to give up. The machine had, after all, been run more or less without maintenance for three years. Time to be systematic.

We removed all the boards and disconnected all the cables. Cleaned all the cable connectors with alcohol. Cleaned the board contacts. Got out my Stabilant 22, the wonderful contact treatment liquid that comes from D W Electrochemicals in Canada, and ran a line of that along every board and on all the cable connectors. When we were finished, we put the whole thing back together and voila! All of which proves what I've known since S-100 days: these machines do take a bit of preventative maintenance, and Stabilant 22 is one of your best friends!

© 1991 McGraw Hill Inc.

In the October 1991 issue we were mentioned once again in Dr. Pournelle's column.


Pioneer does It Again

The Pioneer six-pack CD-ROM MiniChanger is great. We've had it in operation for the best part of a year now, on a number of different systems. At present, its daisy-chained with Pioneer's read/write multipurpose optical disk drive run by a Corel SCSI board, on an Arche Legacy 486 running DR DOS 5.0. It had always performed flawlessly. I've had three different CDROM programs open in three different Desqview windows and jumped back and forth among them; no problems, and as I said in the original review, it changes drives a lot faster than you'd expect it to.

The other day a mysterious package arrived from Pioneer. No cover letter, no PR hype, no press release. Just a card from a Pioneer engineering supervisor, a plastic tube containing two very large ROM chips, and a single sheet of instructions on how to take the MiniChanger apart and replace the old ROM chips with new.

The change took about 10 minutes, because the ROM chips are big and we bent the pins inserting the first one. After I straightened them out and got out the Stabilant 22 contact enhancer and connector lubricant, we didn't have any more problems. We put the machine back together again, loaded it with some highly visual intensive CD-ROMs like the wonderful Impressionist disk, and tried it out.

It's really fast. It may not be faster than the newest Hitachi CD-ROM player but it's not noticeably slower. Accesses that used to take many seconds are now nearly instantaneous. Accesses that took over a minute now take a few seconds. I always did like the MiniChanger, even though it was slower than the Hitachi. Now it's even better.

Incidentally, the Pioneer MiniChanger will work just fine with a Mac.

© 1991 McGraw Hill Inc.

Martin Heller also mentions it in January, 1992 in his column.

BEYOND DOS Martin Heller

Windows 3.1 API's, C++ libraries for Windows, and hard drive adventures
Our thanks to McGraw Hill for permission to reprint the material from Byte Magazine


"Oh what horror's I have seen in that chamnber"
        - James Joyce

Being a beta tester is both good and bad. You get to suggest things that would be useful to you in the final product, but you have to put up with the inevitable horrors - bugs and synchronization problems. If you can use one beta program, fine - you can keep that under control. But I'm currently running test versions of Windows, Word for Windows, and Borland C++. It gets out of hand. Not that any of these is bad - no, I wouldn't go back to previous versions for comparability................

Alarms and Excursions

A few months ago, one of my ESDI hard drives developed the alarming habit of spontaneously spinning down and dumping its heads. Click-thunk, and suddenly DOS wouldn't be able to see the D partition table. Usually a reboot would fix things. The problem seemed worst in the morning, but leaving the machine on all the time didn't seem to help. I was seriously considering that one day it would spin down and never power up again.

I got religious about doing my backups, and realized that my 60-MB cartridges were not ideal for backing up a gigabyte of material. It took me all day to do a single set of backups, and I was not about to make multiple sets. With a little push from the folks on BIX, I bought a bottle of Stabilant 22A. I cleaned and treated all the drive cables; better, but not fixed. Finally I treated the hard drive controller card's edge connectors and all the socketed IC's; better, but still not fixed. Finally I dripped some Stabilant into to ribbon cable connections inside the drive and Eureka! I've had no more trouble.

Intel OverDrive

I first heard about it from a reader, and then I got an Intel OverDrive chip myself. It really works.

I have two Cheetah systems, a 486/25 and a 48CV33. Both of them work splendidly. The 486/33 is my regular machine, the one I'm writing this on. When Intel sent me an OverDrive chip, they sent me a 33 MHz version, which will change my 486/33 into a 486DX2/66. What that means is that the bus speed remains the same, but the internal speed in the chip is doubled

Because the Cheetah 486/33 is my main machine, we installed the OverDrive in the 486/25. 'Me chip comes with a small chip puller, but we found that useless: the puller bent before the chip was removed. Then it was back to the screwdriver and being very careful.

Once the 486/25 was removed, we popped in the OverDrive. Alas, because of the configuration of the Cheetah board, we didn't push it down quite hard enough: you could see a little of the pins. When we turned on the machine, it didn't work; the next time we used a bit of Stabilant 22 contact enhancer and lubricant (great stuff) and seated the chip firmly. This time it worked fine and continues to do so.

The 486/25 is now faster than the 486/33. I Will now go out and buy an OverDrive 25-MHz chip and put the 33-MHz chip into my 486/33. When I do that, the Cheetah 486/33 will probably be the fastest machine in the house. Its only rival will be the Gateway 2000 4DX2-66V, and my suspicion is that the Cheetah is going to be slightly faster. The Cheetah is one great machine. Alas, the company is out of business.

Note that we made no changes to software or BIOS; we just swapped chips. The results are just wonderful. The Intel OverDrive chip gets a User's Choice Award. Recommended.

© 1993 McGraw Hill Inc.

From the July 1993 issue of Byte ...

THE DOS 6 QuestionJerry Pournelle

I'm sure DOS 6 will eventually catch on. But until I have some applications that need it. I'll use DOS 5.O, QEMM, and caching controllers.

About the time you read this, they should be doing flight tests for the DC/X out at White Sands. DC/X, which stands for Delta Clipper X, is a flying scale model of the SSX spaceship that General Graham, Max Hunter, and I have been involved with. The goal is to have a ship that will fly into orbit without dropping off any stages (and thus will be able to take off from any location, not just rocket ranges), return, refuel, and fly into orbit again without any refurbishing. It should also have the capability of surviving an engine failure on takeoff. DC/X won't do that - it has only four engines, and you need at least eight - and it won't make orbit, but it does test many of the concepts needed for a proper spaceship.

We went down to the rollout at the McDonnell Douglas plant in Huntingdon Beach, and it was a pretty impressive thing to see a spaceship - even a model - rolled out like we used to roll out airplanes. We won't have access to space for the rest of us until we have spacecraft that operate like airplanes, and DC/X gets us one step closer to that.

At the rollout, DC/X program manager Paul Klevatt said that this was the first project he'd ever seen in which the software development was at the long pole in the tent. DCVX is controlled entirely by computers - they're using a 32-bit, 4.5-MIPS computer with off-the-shelf flight-control hardware such as the F-15 inertial navigation system-and the programs, being part of a Department of Defense project, are written in Ada.

Originally there were plans for them to develop software from scratch, but there wasn't enough money for that; which, I suspect, may have been a blessing in disguise, because it led them to off-the-shelf CASE tools. One of these was Matrix X from Integrated Systems. This starts with a graphical representation and develops actual Ada code. According to Klevatt, "Our coding error rates are much lower than on previous projects and debugging times have been much shorter."

Anyway, she's a beautiful ship and I'll be sure glad to see her flying.

Everyone runs out of disk space, and it doesn't matter how much you have. Ezekial, my original Z809 system, had twin 64-KB floppy disks to hold both programs and data, and I can recall thinking how luxurious double-density 8-inch floppy disks would be. Now, even with twin 330-MB hard drives, plus network access to the Pioneer read/write optical drive, I sometimes find I have to stop and shift things around to install a new program. I can imagine what it must be like for people who don't have the hardware users I do.

One answer to the disk-space problem is compression systems, and the latest of these is DOS 6, which includes both file compression and memory management. While Microsoft's special introductory offer will be over before you read this, I suspect DOS 6 will still be far and away the lowest-cost way to get these goodies.

DOS 6 works, or at least I had no great problems with it; but for some reason, I don't get a warm feeling .......

There are some problems Stabilant 22 won't solve, but I'm astonished at how many it does take care of. The Pioneer CD-ROM drive resides back in the cable room (a horrible place infested with monsters), where it runs off the Cheetah 386/25 we use as the server for the W4WG network. When I had problems installing Speedcache+, I found I needed to get a telephone back into that room, which meant stringing together phone lines using those wonderful little gizmos that let you do that. You expect that if you have seven different telephone wires connected end to end, you'll get a noisy line, and indeed that happened, so I used Stabilant 22 on each connection. The result was blessed silence. Byte cartoon I periodically use Stabilant 22 on all my phone connections, including modem phone lines.

In case you don't know about Stabilant 22, it's a contact enhancer. It comes in a little bottle that isn't cheap, but don't worry about that. A little goes a long way.

It helps eliminate noise in telephone lines, but for me the most important use of this stuff is inside my computers. I have a lot of fairly old equipment. My main machine is a Cheetah 48CY33 with an Intel OverDrive CPU which makes it in effect a 486/66. You may recall that the Cheetah 486 won my 1990 User's Choice Award as the most useful machine of the year. The previous year's winner was the Cheetah 386, which is useful as a network server.

Problems never come singly: when my attempts to install Speedcache+ crashed the Cheetah 386, recovery was complicated because the Cheetah 486 began to act flaky at reboot time. Sometimes it wouldn't boot at all - even from the "panic" boot disk, and if you have not made yourself one for every machine you have, go do that now-sometimes it would tell me the hard drive was all properly formatted, and then it began telling me that there were no hard drives at all.

"Don't panic," I kept telling myself, as I thought about all the deadlines I'm facing. I had everything backed up on DAT (digital audiotape) using the Palindrome backup system, so if worst came to worst, I could install Palindrome in a Gateway 486 and let it transfer over my whole working environment. It would cost a couple of hours, but it would be no disaster.

I didn't want to do that because I like this big Cheetah. It's not quite as fast as the Gateway 4DX266V because the latter has a local bus video, but unless I'm doing very complex video images, I can't tell the difference. The Cheetah has worked fine for years.

I don't like opening up computers without need, but it was clear I'd have to get inside this one, so I did. It was dusty in there, and cables were bunched up in a way that might have been blocking airflow. I moved the cables. I also reinstalled the CDROM drive. It's the one that comes with Creative Labs' Multimedia Upgrade Kit; a fast reliable CD-ROM drive. Highly recommended. Alas, when I first installed it, I hadn't any proper rails for the hard disk cage, and I up with gaffer's tape. This time I found some rails and did it right. Then I vacuumed things out and got a fan blowing into the case while I tested things.

This time it booted from a floppy disk, and once booted that way, it could find the hard drives. But it wouldn't boot from the hard drive.

OK, that's progress, I thought; so let's see what else I can do. The machine was filthy in there, so I took all of the boards out. Some of the boards-including the caching controller-had discoloration's on the contacts. I got out the Stabilant 22 and used that to polish up every board contact: then for good measure I used it on all the cable L contacts as well.

This time when I fired up, everything worked fine. I confess that before I actually put the cover back on, I used the reset used the reset switch several times and powered the system on and off a few times. My lack of faith wasn't justified. It worked every time. If it ain't broke, don't fix it; but was next time you have to do any troubleshooting, use Stabilant 22. ... If it doesn't fix things, you will at least know that the problem hasn't been dirty contacts, and that's always worth knowing. Highly recommended.


© 1993 McGraw Hill Inc.