tarsa@abyss.zk3.dec.com (Greg Tarsa) (08/03/90)
Could someone give a quick tutorial on just how a SCSI bus
needs to be terminated? I thought I understood how this worked
(that is, the last device on the bus needs a termination resistor
plugged into it), but now I am starting to see those "plugs" that
one is to plug into the end of the SCSI bus cable. It appears
that this is used in addition to, not in place of, the last device
resistor pak. My experiences mostly center around SUN shoeboxes where
they did not used to use plugs and DEC shoeboxes, where they do seem to
use them.
What is correct?
What is the function of the "plug", is it really in addition to,
or can it be used in replacment for the last device's
resistor pak?
Finally, where can I buy these plugs in small quantities?
Thanks,
Greg
Tarsa Software Consulting
--------------------------------
33 Seabee Street
Bedford, NH 03102 tarsa@elijah.mv.com
(603)668-9226 {decuac,decvax}!elijah!tarsawschmidt%decefix@decefix.iao.fhg.de (Wolfram Schmidt) (08/03/90)
Another Question: Many SCSI host adapters for the AMIGA have external Mac compatible DB25 SCSI connectors. If you have internal drives the SCSI bus is terminated at both ends. Good. If you have external drives ... Good. If you have both the bus has 3 terminators. The host adapter sits in the middle of the cable. Shouldn't the host adapter's terminators be removed? As of my knowledge all HA have them soldered in. Will SCSI work with this 3 terminators? Wolfram Schmidt wschmidt@iao.fhg.de
terry@iti.org (Terry Hull) (08/03/90)
wschmidt%decefix@decefix.iao.fhg.de (Wolfram Schmidt) writes: >Another Question: >Many SCSI host adapters for the AMIGA have external Mac compatible DB25 SCSI >connectors. >If you have internal drives the SCSI bus is terminated at both ends. Good. >If you have external drives ... Good. >If you have both the bus has 3 terminators. >The host adapter sits in the middle of the cable. Shouldn't the host adapter's >terminators be removed? As of my knowledge all HA have them soldered in. >Will SCSI work with this 3 terminators? It should. I have a Sun SLC sitting in front of me with a HP and a Wren drive on it both terminated. The SLC provides termination also. Things work just fine for me. >Wolfram Schmidt >wschmidt@iao.fhg.de -- Terry Hull Department of Electrical and Computer Engineering, Kansas State University Work: terry@eece.ksu.edu, rutgers!ksuvax1!eecea!terry Play: terry@tah386.manhattan.ks.us, rutgers!ksuvax1!eecea!tah386!terry
stevem@sauron.Columbia.NCR.COM (Steve McClure) (08/04/90)
In article <210@rusux1.rus.uni-stuttgart.de> wschmidt@iao.fhg.de (Wolfram Schmidt) writes:
+Another Question:
+
+Many SCSI host adapters for the AMIGA have external Mac compatible DB25 SCSI
+connectors.
+If you have internal drives the SCSI bus is terminated at both ends. Good.
+If you have external drives ... Good.
+If you have both the bus has 3 terminators.
+The host adapter sits in the middle of the cable. Shouldn't the host adapter's
+terminators be removed? As of my knowledge all HA have them soldered in.
+
+Will SCSI work with this 3 terminators?
I don't see any terminators on my A2091 SCSI HA and I do have internal and
external devices
--
----------------------------------------------------------------------
Steve email: Steve.McClure@Columbia.NCR.COM 803-791-7054
The above are my opinions, which NCR doesn't really care about anyway!
CAUSER's Amiga BBS! | 803-796-3127 | 8pm-8am 8n1 | 300/1200/2400wcarroll@encore.com (Mr. New Dad) (08/04/90)
From article <210@rusux1.rus.uni-stuttgart.de>, by wschmidt%decefix@decefix.iao.fhg.de (Wolfram Schmidt): > Another Question: > > Many SCSI host adapters for the AMIGA have external Mac compatible DB25 SCSI > connectors. > If you have internal drives the SCSI bus is terminated at both ends. Good. > If you have external drives ... Good. > If you have both the bus has 3 terminators. > The host adapter sits in the middle of the cable. Shouldn't the host adapter's > terminators be removed? As of my knowledge all HA have them soldered in. > > Will SCSI work with this 3 terminators? > > Wolfram Schmidt > wschmidt@iao.fhg.de SCSI termination is not some magical item that must be there for the bus to work. It is simply a circuit that is used to eliminate reflections of signals up and down the cable. It will only do this (to any real effect) when it is at the physical end of the cable. Yes, it is possible to operate a SCSI bus with three sets of termination. Even more than three. But this is not recommended. Each termination network lowers the impedance to ground, which increases the current each driver must sink when it drives the line low. Most SCSI drivers are not specefied to drive over-terminated lines. This means that if you over terminate the bus, you will prematurely age your drivers best case, or blow them out worst case. Summary: Don't overterminate unless there is no way around it. Note: I tend not to pay alot of attention to termination and often run overterminated busses. But I don't do this on any permanent setups, usually leaving a bus overterminated for only a day or two. I have also never seen a SCSI device burn out it drivers from any cause and have never seen a bus fail because of too many terminations. I still do not recommend it and waive all responsibility if you choose to imitate my foolish actions and end up with ruined drives. -- William R. Carroll (Encore Computer Corp., Ft. Lauderdale FL) wcarroll@encore.com uunet!gould!wcarroll "Dan Quayle gives underachievers a bad name." -- Bart Simpson
jlohmeye@entec.Wichita.NCR.COM (John Lohmeyer) (08/06/90)
SCSI bus termination always seems to generate a lot of traffic in this newsgroup. There is usually a lot of well-intentioned wrong advise in this traffic. Let me try to clear up a few points. As correctly pointed out in one response (sorry, I didn't get the name), SCSI requires one termination at each end of the bus. The characteristic impedance (Engineer talk: loosely, this means the resistance) of the terminator was selected as a compromise betweeen two factors: 1) the characteristic impedance of the cable (typically 100 ohms or less) 2) the current which SCSI drivers are required to drive (48mA at 0.5 volt) Unfortunately, it is not possible to exactly match the 100 ohm impedance without exceeding the 48 mA drive current. SCSI-1 settled for a simple 220/330 ohm termination which has a characteristic impedance of 132 ohms. This is not ideal and some of the energy from signals is reflected from the terminator which decreases the SCSI noise margin. Fortunately, most SCSI-1 systems functioned perfectly well with this decreased noise margin, mostly because they did not push the extremes of cable length, transfer rate, and number of devices. This 220/330 ohm terminator does not work as well with some SCSI-2 applications, especially when people use the new high-density connector. This is not a connector problem, but since it is physically smaller, one has to use smaller cables as well. Smaller cables tend to have lower impedance. Now we are seeing 80 to 90 ohm cables. Also, as SCSI is being used for bigger systems, we are seeing maximum length cables with 7 or 8 devices. Thanks to a lot of hard work from several people including Kurt Chan of HP and Paul Boulay of LMSI, a new, improved terminator was designed for SCSI-2 which can still be driven by 48 mA drivers, but has a characteristic impedance of 110 ohms. It also cures a nagging problem of low noise margin from a sagging terminator power line (TERMPWR). The new terminator is recommended for SCSI-2, but the old one is still acceptable if your system is not pushing all the other limits. I may have digressed a bit, but I wanted to point out that SCSI termination is important and there are consequences to ignoring it. I personally lost two weeks of work when I backed up over a SCSI cable with only one terminator. I do know of people who have damaged SCSI drivers by having too many terminators present -- this is rare. Far more often, they just get intermittent data errors (I hate intermittent errors!). The important thing to remember is that SCSI is a transmission line any time you exceed about 1 foot of cable. Transmission lines must be terminated at both ends or you will get reflections which screw up signal quality (that is, cause errors). Just because some people have gotten away with breaking these rules sometimes doesn't mean you will. Hope this helps... t impedance (Engineer talk: loosely, this means the resistance) of the terminator was selected as a compromise between matching the characteristic impedance of the cable and what the SCSI drivers can safely drive. -- John Lohmeyer J.Lohmeyer@Wichita.NCR.COM NCR Corp. uunet!ncrlnk!ncrwic!entec!jlohmeye 3718 N. Rock Rd. Voice: 316-636-8703 Wichita, KS 67226 SCSI BBS 316-636-8700 300/1200/2400 24 hours
mk59200@metso.tut.fi (Kolkka Markku Olavi) (08/08/90)
In article <626@entec.Wichita.NCR.COM>, jlohmeye@entec.Wichita.NCR.COM (John Lohmeyer) writes: |> Thanks to a lot of hard work from several people including Kurt Chan of |> HP and Paul Boulay of LMSI, a new, improved terminator was designed for |> SCSI-2 which can still be driven by 48 mA drivers, but has a characteristic |> impedance of 110 ohms. Is this design freely available? If it is, could it be posted to this group? -- Markku Kolkka mk59200@tut.fi
jlohmeye@entec.Wichita.NCR.COM (John Lohmeyer) (08/10/90)
In article <1990Aug8.063528.22915@funet.fi> mk59200@metso.tut.fi (Kolkka Markku Olavi) writes: > >In article <626@entec.Wichita.NCR.COM>, jlohmeye@entec.Wichita.NCR.COM >(John Lohmeyer) writes: >|> Thanks to a lot of hard work from several people including Kurt Chan of >|> HP and Paul Boulay of LMSI, a new, improved terminator was designed for >|> SCSI-2 which can still be driven by 48 mA drivers, but has a characteristic >|> impedance of 110 ohms. > >Is this design freely available? If it is, could it be posted to this >group? Yes and Maybe. Yes, it is part of the SCSI-2 standard. But I do not have it in an ASCII format. It is in AutoCAD format and will soon be posted on the SCSI BBS. Perhaps some kind soul has time to enter the schematic into a text editor, but otherwise you will need to get a paper copy from Global Engineering Documents (714) 261-1455 or (800) 854-7179 [I doubt the 800 number will work internationally]. -- John Lohmeyer J.Lohmeyer@Wichita.NCR.COM NCR Corp. uunet!ncrlnk!ncrwic!entec!jlohmeye 3718 N. Rock Rd. Voice: 316-636-8703 Wichita, KS 67226 SCSI BBS 316-636-8700 300/1200/2400 24 hours
gson@niksula.hut.fi (08/11/90)
In article <1990Aug8.063528.22915@funet.fi> mk59200@metso.tut.fi (Kolkka Markku Olavi) writes: >In article <626@entec.Wichita.NCR.COM>, jlohmeye@entec.Wichita.NCR.COM writes: >|> [... a new, improved terminator was designed for SCSI-2 ...] >Is this design freely available? If it is, could it be posted to this >group? The new terminator is described (including schematics and component values) in an article titled "Push SCSI Performance To The Limit" in the May 10 issue of Electronic Design. The article also contains lots of other useful information about SCSI bus impedances, reflections, noise margins etc. Disclaimer: I have no connection with Electronic Design except as a happy reader. -- Andreas Gustafsson Internet: gson@niksula.hut.fi Voice: +358 0 563 5592