The RLL (Run Length Limited) Controller The RLL controller, with its ability to make a 20 meg hard disk into a 30 meg drive is of even more interest as the price of hard disk drives falls. At the hardware level, disk recording is an analog process. The bits are recorded in a "saturated" fashion (each bit has all the magnetism of which the medium is capable), but the recovery process must deal with very low "volume" (signal levels). Thus software algorithms must assist the hardware in order to get an acceptably low error rate. One error in 10E-12 is conventionally acceptable. Although ancient media like punched tape, mainframe 1/2" 9-track and hobbyist cassette used simpler schemes, modern high-density magnetic media have almost universally used "clockless MFM" (modified frequency modulation) on both floppies and hard disks for about ten years. The lowest-level disk routines of both MSDOS and CP/M are tied, to some extent, to this recording method; it is reasonably efficient; and, above all, the interface to a DOS computer is easy. Recently a more advanced coding scheme, RLL (run-length limited) has begun to emerge in micro hard disk systems. By tolerating greater complexity in the encode/decode process (bit patterns can have variable length, which caters to the strengths of the magnetic medium), up to 50% more data can be stored with the same track spacing and bit density. The time overhead imposed by the more complicated algorithms is negligible at the data transfer rates of our machines. New LSI chips on the controllers remove the cost, design expense and board-real-estate objections and allegedly make the new format transparent to DOS. The lure of 40-50% more data on the same hardware is strong enough to drive the market and eventually make these objections obsolete. For now, there's a slight aura of pioneering about RLL, not quite to the point where I'd recommend arrow-proof britches but, again..... The new controllers (Adaptec, OMTI et. al) are more expensive, and some drive makers say the new format stresses the drive more (Seagate says it's a no-no for their model 225), but I can't see why. We've used the Adaptec 2070A with the ST-225 in several installations, and know of many others, with nary a problem. While the RLL formatting would only be marginal on an ST-225 that just met the published specs, the fact is that Seagate builds some of the best drives in the world, and the 225's are built over-spec, so in actuality there is no problem. Neither Adaptec nor Seagate will certify that RLL will work on the 225; it does, anyhow. If you have a 225, this is a cheap way to increase your storage by 50% and get an increase in speed to boot. If you are going to add a drive, however, along with the Adaptec, you should go for the very slightly more expensive ST-238, which IS certified by both manufacturers for RLL. Minor caveats: check that whatever you buy works with whatever you will eventually want. Some RLL controllers have trouble with some tape backup units, or with Fastback. Some are impenetrable to Norton and other sector-level utilities. I know nothing about the Adaptec 2070 specifically on these matters. ONE USER'S EXPERIENCE WITH RLL 2,7 CONTROLLERS I have tried two of the new RLL 2,7 hard disk controllers, The ADAPTEC 2070A and the OMTI 5527. Here is a summary of what I've found so far. The ADAPTEC controller is a long card requiring a full size slot. The BIOS is very compatible, but the status bits returned by the controller aren't the same as WESTERN DIGITAL, XEBEC, etc. This may cause difficulty with any software that directly addresses the controller, such as some tape back-up software. This is rare and the problem is not limited to ADAPTEC or RLL controllers. The controller formats tracks with 26 sectors, 25 of which are used. The 26th is a spare sector. After doing a low level format (which is invoked by entering debug and doing a G=C800:CCC command), it scans each track. If a bad sector is detected, it is flagged for the controller and the 26th sector is used so that each track has 25 good sectors. If a track has two or more bad sectors, the track is flagged and an alternate assigned, starting with the highest cylinder on the disk. The formatting, including the verify, takes a very long time. The built in low-level format program is "self-configuring". It asks how many cylinders, heads, etc. your drive has. It should support almost any RLL compatible drive. I have used it with a SEAGATE ST238 and a TANDON TM502 with no problem. I couldn't get any documentation on programming from ADAPTEC. They will sell you a "User's Manual" for $5.00. While it is helpful for installing the controller, it's no help in programming it. I find this unacceptable. The OMTI is a (very) short card, which is also used on some "Hard Cards". It formats 26 sectors per track, all useable. Bad tracks are handled in the more conventional manner of only assigning alternate tracks. This yielded 1.2 megs more useable space on my ST238, which had no bad sectors. The built-in BIOS also has a format routine, but it requires the OMTIDISK 2.4 software to be self-configuring, unless you know how to address cyl 0, trk 0, sec 1 directly. The configuration is stored at +11A. The OMTIDISK software is free from OMTI if you call and ask. I've heard reports they required you to go through a dealer to get it, but I talked to the design engineer (on HIS nickel) and explained that I needed to support 2 different drives and he took my name and address and is sending it along at NO cost. The documentation was free for the asking and VERY complete. You can program from it. Both controllers are very fast. The data transfer rate is about 250-260KB/SEC with either controller, with the edge going to OMTI on my 8MHZ system since it will support an interleave of 1:1 vs 3:1 on the ADAPTEC. Not just any drive will work reliably. The controllers require drives that meet ST412 specs. Check with ADAPTEC or OMTI to see if your drive is approved BEFORE buying the controller. To summarize, both are good controllers. I give the edge to OMTI for the following reasons: (1) More useable space -that's why I went RLL in the first place. (2) Short card fits any slot. Not important to some, but I like the flexibility. (3) Slightly faster with interleave of 1:1. (4) Better documentation, and it's free. (5) Super support. The design engineer called me back and talked about my questions. ADAPTEC TESTED WITH THESE DRIVES According to Adaptec, they have successfully tested the ACB-2070A with the following drives: (RLL Format) Drive Model Capacity Seagate ST-238 31MB Rodime RO352 15MB RO202E 31MB RO203E 47MB RO204E 63MB Tandon TM502 15.6MB TM755 62MB MicroScience HH-325 31MB Miniscribe 3425 31MB NEC 3126 31MB 5126 31MB Mitsubishi MR522 31MB Priam (Vertex) V150 62MB V170 90MB Syquest SQ319 15MB SQ357 45MB Tulin TL226 31MB TL326 31MB TL240 48MB TL340 48MB Toshiba MK56FA 105MB MicroStorage MS212 15MB Lapine Titan 31MB Okidata OD526 31MB OD540 48MB Bull D530 38MB D550 63.2MB D570 88.4MB D585 104.5MB The Adaptec controller board has been successfully tested in the following personal computers: IBM-PC IBM-XT Compaq Portable Compaq Deskpro Compaq 286 Portable (Note 1) Compaq 286 Deskpro (Note 1) ATT 6300 (Note 2) Leading Edge (Note 1) Sanyo Eagle Notes: 1. ACB-2070A requires product P/N 401400 Revision C or later. 2. Requires ATT motherboard ROMS version 1.21 or later. There are even more advanced error-correct/detect and encoding schemes in the wings. Some, Tallgrass among them, have been using GCR (whatever that is...ED) for several years (irony: the home- brew Wozniak floppy controller on Apples has used a kind of GCR for years -- at 140K/disk!). Couple the above with the emerging SCSI interface standard (and again ... whatever that is ... ED), controllers that use 16-bit transfer and real DMA (Direct Memory Access ... Ed) (the AT doesn't), better heads, etc., etc. and it looks like the good old hard disk has many more years in it, with data transfer rates approaching that of a good RAMdisk.