chris@umbc3.UUCP (07/09/87)
Hello all -
I just finished flipping through the latest issue of Computer Shopper and
got extremely sick at the prices of the Seagate HD drives. 20 meggers for
$350 including controller (for IBM PC)?!?! Come on now, SURELY a SCSI
controller doesn't cost $200. So, here's what I've been thinking....
Wouldn't it be nice to get a 30-40 meg hard drive in the $400 range, pop
on a SCSI controller, format it with somebody's software (at this point,
I don't care WHERE it comes from {send flames to /dev/null}) and be off and
hard-driving?
So here's where I'm stuck. I don't mind "hacking" a *little* with hardware
(ie, assembling HD/controller/cables). However, I am completely at a loss
as to where to get information on how to perform such a feat. Suggestions
on reliable drives (I want auto-parking, LED status light, QUITE operation,
and of course seek times <40 ms) from different companies.
Since one would think MANY people would be interesting in getting a good-
sized HD system at a reasonable price here on the net, I think a small
discussion would be ok, but please EMAIL when you can. Also, if you are
interested in seeing a summary posted, let me know that too.
ps - the main reason why I'm getting a hard drive is because I *must* use
IBM AT's at work and (I don't believe I'm saying this) they're ALMOST
tolerable with hard drives. Just think what I could do with that pile
of 70+ 800K floppies all in one location on the Mac!
--
ARPA : chris@umbc3.UMD.EDU BITNET : chris@umbc
"Better a bad reason than no reason at all!"jww@sdcsvax.UCSD.EDU (Joel West) (07/09/87)
See MacTutor, 2/87 and 6/87. One article contained a hardware
description, the second a SCSI formatter program.
--
Joel West, Palomar Software, Inc. (c/o UCSD)
{ucbvax,ihnp4}!sdcsvax!jww or jww@sdcsvax.ucsd.edulzulch@cdp.UUCP (07/10/87)
Check the February and March MacTutor magazines for information on building your own SCSI hard drive. They each contain articles by Tim Standing on various aspects of the process. I have used his CDC Wren--160 megabytes--and it is the fastest hard drive I have ever seen on a Mac (or any micro for that matter.) Larry Zulch, Dantz Software Development
czei@osupyr.UUCP (Michael S Czeiszperger) (07/11/87)
I've never actually seen a SCSI controller, so I don't have any concrete
examples of what they cost, but one thing is certain: they're alot more
expensive than an IBM standard controller. I'm currently building a SCSI
target controller using the NCR5380 chip, and it is NOT an easy task.
On the same topic, MacTutor magazine has featured a SCSI controller project
recently, although I was under the impression is was a bit buggy.
--
Michael S. Czeiszperger
College of the Arts Computer Lab
The Ohio State University
Columbus, OH 43210
{decvax,ucbvax}!cbosgd!osupyr!artsvax!czeikhayo@MATH.UCLA.EDU (07/13/87)
In article <3437@sdcsvax.UCSD.EDU> jww@sdcsvax.UCSD.EDU (Joel West) writes: >See MacTutor, 2/87 and 6/87. One article contained a hardware >description, the second a SCSI formatter program. > Joel West, Palomar Software, Inc. (c/o UCSD) > {ucbvax,ihnp4}!sdcsvax!jww or jww@sdcsvax.ucsd.edu I leafed through some non-Mac magazines, and indeed it looks like one can easily get a 40M Seagate sans controller for less than $450. This stirred considerable enthusiasm in me, but the local libaries somehow live without MacTutor and I don't know if ordering back issues from the magazine is worth the trouble. Before I do order, could someone kindly summarize what the articles said? Specifically, - what is your estimate of the controller's cost ($300? $50?) - I have no idea about how the thing works, so a very stupid question: would this "poor man's" controller significantly affect the disk's access time, if at all? - are there any serious drawbacks of this solution (compared with going the orthodox route) that the Mac/electronics wizards out there can see? As far as I know, many of us are quite adept at working with a soldering iron (I even have fun doing that), so the above may be of some interest to others too. I'll summarize mail replies (if any). Thank you very much, Eric Behr khayo@MATH.UCLA.EDU
oster@dewey.soe.berkeley.edu (David Phillip Oster) (07/13/87)
In article <458@osupyr.UUCP> czei@osupyr.UUCP (Michael S Czeiszperger) writes: >I've never actually seen a SCSI controller, so I don't have any concrete >examples of what they cost, but one thing is certain: they're alot more >expensive than an IBM standard controller. Nonsense. On an IBM you start with a drive that already speaks the SCSI interface standard. I'll repeat that, because it is important. Drives for the IBM PC market already speak the SCSI interface standard. Now, this isn't absolutely true, some of the older ones only speak SASI, the precursor of SCSI. SASI differs from SCSI in that SCSI has improved daisy-chaining, but if you don't need to hang multiple hard disks off your mac SASI works just as well. If IBM drives already speak SCSI, then why do you need a "Controller board" to attach one to a PC? Because the PC doesn't speak it, the "Controller board" just translates from SCSI to the IBM PC bus. When you put a hard disk on a Mac Plus, since the Mac Plus directly speaks SCSI, you don't need a controller board. Now, what is on that IBM "Controller board"? To answer this, I have to explain a little bit about the SCSI protocol. SCSI is basically a byte-oriented parallel protocol. A SCSI message has a header, data, and is sent a byte at a time. Today, designers use special purpose chips that decode the header. In the old days we just used to use a Parallel Interface Adapter (PIA), the same chip that was used to implement the "IBM" (really Centronics, remember them?) parallel printer interface. And the header was decoded in software rather than on the fancy NCR chip. I myself have written hard disk drivers for SASI drives using this technique. Apple used to sell a single parallel interface adapter card for the Lisa that would take either a printer or a disk. The printer was a parallel interface imagewriter (Called a C. Itoh ProPrinter back then, but identical to an Imagewriter 1). The drive was a Profile hard disk. The Profile was slow, but I am convinced that that was not the fault of the Lisa interface card, after all the interfaces I wrote ran at reasonable speed. All you need to connect such a drive to a mac is an appropriate cable. (The mac has a 25 pin connector, the drive a 50. Half of the drive's pins are tied to ground.) A bare bones SCSI driver is trivial. The ROMS use the SCSI standard to read the first few sectors into memory, these should contain your MAC disk driver. For those sites on arpanet, A generic SCSI driver kit by Vishniac is archived on [SUMEX.STANFORD.EDU] <INFO-MAC> (see your local ftp documentation.) The extra cost of the packaged Macintosh disk over those bare bargain drives goes for: 1.) a drive power supply 2.) a case 3.) software 4.) service (handling DOA drives, software upgrades 5.) cables 6.) profit. If you want to do it yourself, then you save the cost of all that. If you don't have a MacPlus or later, you'll need to add hardware inside your Mac to recieve the signals from the drive. The newer Macs have this built in. Thanks Apple! --- David Phillip Oster --My Good News: "I'm a perfectionist." Arpa: oster@dewey.soe.berkeley.edu --My Bad News: "I don't charge by the hour." Uucp: {seismo,decvax,...}!ucbvax!oster%dewey.soe.berkeley.edu
ching@amd.UUCP (Mike Ching) (07/13/87)
>>I've never actually seen a SCSI controller, so I don't have any concrete >>examples of what they cost, but one thing is certain: they're alot more >>expensive than an IBM standard controller. > >Nonsense. On an IBM you start with a drive that already speaks the >SCSI interface standard. I'll repeat that, because it is important. >Drives for the IBM PC market already speak the SCSI interface >standard. Not true. IBM drives speak ST506/ST412. There are adapters by Xebec, Adaptec, DTC, etc. to go from SCSI/SASI to ST506/ST412. There *are* SCSI drives, eg., Seagate ST225N is a SCSI version of the ST225, but the majority of drives are ST506/ST412. >All you need to connect such a drive to a mac is an appropriate cable. >(The mac has a 25 pin connector, the drive a 50. Half of the drive's >pins are tied to ground.) A bare bones SCSI driver is trivial. The cable is not trivial to construct. It took me over 3 hours to get it done. Go out and buy one if you do this. >The extra cost of the packaged Macintosh disk over those bare bargain >drives goes for: >1.) a drive power supply >2.) a case >3.) software >4.) service (handling DOA drives, software upgrades >5.) cables >6.) profit. My out of pocket cost was ~$450 and the project took a fair amount of time. I only did it because when my Relax drive went south, I couldn't reformat it but had to go back to them for a low level format (which they did by connecting it to a IBM PC). Not being able to format my own disk bothers me. mike ching Newsgroups: comp.sys.mac Subject: Re: Making *own* SCSI Hard Drive Summary: Expires: References: <380@umbc3.UMD.EDU> <458@osupyr.UUCP> <19683@ucbvax.BERKELEY.EDU> Sender: Reply-To: ching@amd.UUCP (Mike Ching) Followup-To: Distribution: Organization: Advanced Micro Devices Keywords: mondo cheap ibm haters
czei@osupyr.UUCP (07/15/87)
>Nonsense. On an IBM you start with a drive that already speaks the >SCSI interface standard. I'll repeat that, because it is important. >Drives for the IBM PC market already speak the SCSI interface >standard. > >Now, this isn't absolutely true, some of the older ones only >speak SASI, the precursor of SCSI. SASI differs from SCSI in that SCSI >has improved daisy-chaining, but if you don't need to hang multiple >hard disks off your mac SASI works just as well. > That is interesting, because I was reading in an old Circuit Cellar article from Byte, that SASI was modeled after the IBM i/o channel. After heavy editing... >Today, designers use special purpose >chips that decode the header. In the old days we just used to use a >Parallel Interface Adapter (PIA), the same chip that was used to >implement the "IBM" (really Centronics, remember them?) parallel >printer interface. And the header was decoded in software rather than >on the fancy NCR chip. I myself have written hard disk drivers for >SASI drives using this technique. Apple used to sell a single parallel >interface adapter card for the Lisa that would take either a printer >or a disk. What chip are you refering to? I'm writing code for a custom SCSI controller built from a NCR5380 interfaced to a HD64180 microprocessor. I was aware there was a more sophisticated chip in the works, but didn't know it was being sold at the moment. (You can't be refering to the 5380 because it most definately doesn't decode command descriptor blocks.) > >All you need to connect such a drive to a mac is an appropriate cable. >(The mac has a 25 pin connector, the drive a 50. Half of the drive's >pins are tied to ground.) A bare bones SCSI driver is trivial. The >ROMS use the SCSI standard to read the first few sectors into memory, >these should contain your MAC disk driver. It seems kind of strange that one protocol could support two different physical cables. According to the ANSI description, a SCSI cable can be 50-conductor flat cable, or a 25-signal twisted pair cable. This implies that there must be 50 pin connections. I wonder why the mac has a 25 pin connector? Are the cable and connector requirements of SCSI just suggestions? (from American National Standard X3.131- 1986, pgs. 16,19) -- Michael S. Czeiszperger College of the Arts Computer Lab The Ohio State University Columbus, OH 43210 {decvax,ucbvax}!cbosgd!osupyr!artsvax!czei
martin@ssc-vax.UUCP (Peter Martin) (07/15/87)
Three of us here are interested in cheaper drives! Keep us posted please.
Pete 206-773-4270
--
Peter Martin WD9HAD NORMAL MODE: uw-beaver!ssc-vax!ssc-bee!martin
ABNORMAL MODE: 206-773-4270
WEEKEND MODE: 206-631-3304
>>> SCCA Flagging and Communications worker <<<fjo@ttrde.UUCP (Frank Owen ) (07/15/87)
> I leafed through some non-Mac magazines, and indeed it > looks like one can easily get a 40M Seagate sans controller > for less than $450. This stirred considerable enthusiasm in > - what is your estimate of the controller's cost ($300? $50?) You should be able to get controllers for ~$150. You may also be able to get old surplus SASI controllers for less than this. ("Computer Surplus Store" has a bunch of SASI boards for $79. They advertise in the back of Byte Magazine. ph.(408)-280-1740. You'll also need a power supply, case and cables to make a working model. > - I have no idea about how the thing works, so a very stupid > question: would this "poor man's" controller significantly > affect the disk's access time, if at all? The really cheap controllers might not be able to read disks formatted at a 1:1 interleave. (Read entire track in one revolution). The one I got (DTC 520) is limited to a 2:1 interleave factor. This is only an issue if you want to use the drive on an SE or MacII, because the MacPlus is not able to keep up with a 1:1 interleave factor anyway. The MacII (I think) can. (Although it does NOT have a DMA for the drive, it has a faster processor, so it SHOULD be able to keep up with the 1:1 data rate. ) In most cases, however, the greater measure of disk speed comes from the "average seek access time" of the drive itself. This measurement pretty much determines the price of the drive in question. > - are there any serious drawbacks of this solution .... The drawbacks are that you are really on your own in terms of getting the software working to drive your particuliar board/drive combination. You also don't get fancy things like disk-spoolers, and backup utilities (unless you want to spend the time writing them yourself, or find them availible in the public domain) Frank Owen ..ihnp4!ttmcsa!ttrde!fjo
mentat@auscso.UUCP (Robert Dorsett) (07/17/87)
A couple of comments. First, you do NOT want to buy a separate controller
board. Seems that if you buy a Seagate drive with an "N" suffix (e.g.,
ST-255N) you get a built-in SCSI controller. Second, I have a friend who's
put one Seatgate 40-megger and two 255N's on his Mac. Apart from having a
power supply with too low a wattage (he did not count on having three hard
disks when he built the first 255N), he hasn't had any problems. Third, it
seems the Seagates are the same drives Apple's using in its HD SC series. Thus,you can use the Apple formatters and utilities on the Seagates. I've heard
conflicting information on this, i.e., that Apple uses another brand, but the
Apple utilities DO work on the Seagate.
No, I haven't built the Seagate--yet. I plan on doing so within a week or
so. I'll keep the net informed.
--
Robert Dorsett {allegra,seismo}!sally!ngp!walt!mentat
University of Texas at Austin {allegra,seismo}!sally!ngp!mentat
mentat@ausco.UUCPjjg@walden.UUCP (John Grana) (07/28/87)
In article <472@osupyr.UUCP> czei@osupyr.UUCP (Michael S Czeiszperger) writes: > > What chip are you refering to? I'm writing code for a custom SCSI > controller built from a NCR5380 interfaced to a HD64180 microprocessor. > I was aware there was a more sophisticated chip in the works, but > didn't know it was being sold at the moment. (You can't be refering > to the 5380 because it most definately doesn't decode command descriptor > blocks.) > Take a look at the Western Digital WD33C92/93 SCSI Interface Controller. I have written the code for a SCSI<->VMEbus host adapter using both the NCR5385 (a little more sophisticated then the NCR5380) and the WD33C92. I would term the NCR538[05] as first generation chips and the WD33C92 as a 2nd generation. For example, using the "Select and Transfer" command (assuming you don't handle multi-threading), you load the command desc. block into the chip, load the destination target id, then issue the Select and Transfer command. At that point the WD33C92 will handle ALL the SCSI bus activity (Selection, Command Out, optional Data In or Out, Message In and finaly Status In). At completion, the chip interrupts with the SCSI status byte. The 33C92 is used for differential interfacing, the 33C93 has the 48MA drivers built in (for single ended , i.e. Mac). >> > >It seems kind of strange that one protocol could support two different >physical cables. According to the ANSI description, a SCSI cable can >be 50-conductor flat cable, or a 25-signal twisted pair cable. This >implies that there must be 50 pin connections. I wonder why the mac >has a 25 pin connector? Are the cable and connector requirements >of SCSI just suggestions? (from American National Standard X3.131- >1986, pgs. 16,19) > > Reading those pages... "The nonshielded SCSI device connector shall be a 50-conductor connector" etc.,etc. (it also says SHALL be 50 when talking about the shielded). It seems Apple took some liberty in cable specs. for the sake of space limitations. It just means your getting less grounding in the cable (i.e. don't try running a 25 meter cable between the Mac and drive/controller). - jjg -- John Grana ..!seismo!rochester!cci632!walden!jjg ^ | for a little longer...