Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cornell!batcomputer!itsgw!steinmetz!davidsen From: davidsen@steinmetz.ge.com (William E. Davidsen Jr) Newsgroups: comp.unix.xenix Subject: Re: Questions about RLL controllers Message-ID: <13139@steinmetz.ge.com> Date: 14 Feb 89 18:52:24 GMT References: <3700022@eecs.nwu.edu> Reply-To: davidsen@crdos1.UUCP (bill davidsen) Organization: General Electric CRD, Schenectady, NY Lines: 154 Bcc: davidsen@sixhub In article <3700022@eecs.nwu.edu> skrenta@eecs.nwu.edu (Richard Skrenta) writes: | Are there any dangers associated with using an RLL controller on drives | only certified for MFM? Could the controller damage the drive in any | way, or is it just a matter of mapping a few more bad tracks during the | format? I noticed there are some nasty warnings in the Seagate manual | about voiding your warranty if you use the wrong controller with a drive. You are going to get a lot of answers to this, let me supply some information which will help you put it all in perspective. RLL controllers don't "put the bits closer together" in terms of more flux changes per inch (fpi) on the media surface. What changes is the resolution of the changes in terms of time. In order to run RLL on a drive and controller combination successfully, the total timing jitter or resolution from end to end must be less than some value N in ns. If you look at the drive test sheets you may see a jitter number for the drive. If you get a chance to select a drive from a batch, take one with a small number, since this gives you a margin of error with any drive. RLL certified drive are *usually* production MFM drives which have been tested for media quality and jitter, and found to be good enough to run RLL. There are probably some manufacturers who make a whole diferent disk, but I have yet to identify one. This menas that when you get an MFM drive it may be one which didn't work as RLL, or it may be one which wasn't tested, or it may be one which passed but wasn't sold because more RLL capable drives were made than were needed to fill orders. With prices down manufacturers don't keep too many drives in stock, and will sell the RLL tested drives under the MFM label. Controllers have an acceptable range, just like drives, and one which is acceptable may barely pass the specs or be 20-30% better (smaller) in jitter. If you get a "real good" controller you can run some pretty bad disks, as long as the total jitter is acceptable. TOTAL JITTER - drive 1) rotational speed varitions a) motor b) drive train c) bearing drag 2) media a) type b) quality of application 3) heads - manufacturing tolerances 4) electronics a) response time (slew rate) of amplifiers b) voltage regulator output c) component manufacturing tolerances d) quality of edge connection TOTAL JITTER - controller 1) electronics a) response time (slew rate) of amplifiers b) voltage regulator output c) component manufacturing tolerances d) quality of edge connection TOTAL JITTER - other environment 1) quality of power supply output a) voltage b) ripple c) spikes 2) cooling a) average temperature b) temperature fluctuations c) humidity 3) other a) magnetic fields b) vibration (affects mechanical connections) c) user practices - frequent power on, etc We are talking ns tolerances here, so it doesn't take too much to change things. WHY ALL THE DIFFERENT RECOMMENDATIONS? If someone has a marginal controller (at the limit of whatever the manufacturer allows for jitter), it will take a reasonably good disk and environment to keep operation reliable. Couple this with some MFM disks which are a little wide on tolerance and "I tried and MFM disks just don't work as RLL." Note that the person with a very tight tolerance controller may report "MFM works fine, I've used any number of 'em." Both of these people are giving you their best advice, and you should be thankful they share their experience with you. If a combination is marginal a number of things will help: temperature should be kept as even as possible. When I first went RLL I had one disk which would run for a few weeks and then get more and more bad sectors. I finally noted that it failed right after I powered up. Solution? Wait 15 minutes after power on to run anything which might write to the disk. A friend had a disk which failed when he played one game... he was leaning the map against the front of the computer case and partially blocking the airflow. RLL WILL DESTROY YOUR DISK If you format RLL and it doesn't work out, the drive must be low level formatted with an MFM controller before reuse. Virtually all of the stories originate with someone who tried to do a DOS (high level) format and failed, concluding that the disk was broken. Some controllers seem to be unable to do a low level format for some reason, and this would make them unable to reformat a disk which had been used as RLL. I have only seen one case where a drive could not be reformatted on a known good controller, and that one suffered a massive mechanical failure of a bearing. IT WORKS FOR A FEW WEEKS AND THEN GRADUALLY GOES BAD This *usually* means you have not locked out all of the bad tracks. I've had this happen myself, and it was usually cured by locking out just a few tracks. It seems that under both DOS and Xenix a few unmapped bad sectors cause additional problems. I can't explain it, just observation, if it doesn't start it doesn't spread. Also, disks running RLL definitely do go bad with use. So do MFM disks, cars, trains, and people. Everything wears out, and a failure after several years doesn't mean that RLL has warped your bits. The only that which gets better with time is wine. WHAT TO DO? If you can find a dealer who is understanding, you will probably be able to get a 10 day return on a controller (that dealer may not be the cheapest price in town). When you low level format run the surface test(s) for 3-4 hours, and be sure to let the disk warm up for 15-20 minutes before formatting. If the test can be run continuously, run them for a few days! If you get vast numbers of errors, you may just have a combination which doesn't work. Anyone who claims that any MFM drive will work "all the time" as RLL probably is an optimist. You can save a lot of money by using MFM drives with an RLL controller. In most cases they will work, and reliably. Even going with a high cost RLL drive from a big-name manufacturer will not absolutely insure that you get reliable operation (it does improve the chances, though). I have had good drives not work, and a number of cheap Seagates work for years. I have used RLL and even the Perstor ARLL (90% increase) controllers on MFM drives without trouble. If you can afford to go top drawer all the way, buy all RLL components (or better yet, ESDI). If you can't afford the money to take some risk that the combination won't work, don't even try it. If you fall in the middle (most of us do), give it a try, if you install it carefully the odds are in your favor. An Adaptek controller is about $150 (back of PC Week) and the WD is about $250. There are others in between. Most of them are less than the diference between an MFM drive and the same drive rated RLL, and easier to install. -- bill davidsen (wedu@ge-crd.arpa) {uunet | philabs}!steinmetz!crdos1!davidsen "Stupidity, like virtue, is its own reward" -me