home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.sys.sun.hardware:5777 comp.periphs.scsi:5564
- Newsgroups: comp.sys.sun.hardware,comp.periphs.scsi
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!rpi!psinntp!psinntp!ptsys1!jjg
- From: jjg@pt.com (John Grana)
- Subject: Re: Problems with Fast SCSI-II on Sun Sparc 10 (long)
- Message-ID: <1992Nov19.010445.20713@pt.com>
- Organization: Performance Technologies, Incorporated
- References: <1992Nov15.121008.12717@pt.com> <1992Nov16.130949.1490@devon.co.uk>
- Date: Thu, 19 Nov 1992 01:04:45 GMT
- Lines: 73
-
- In article <1992Nov16.130949.1490@devon.co.uk> mcneill@devon.co.uk (Keith McNeill) writes:
- >From article <1992Nov15.121008.12717@pt.com>, by jjg@pt.com (John Grana):
- >
- >Ok...novice question. What is the difference between active and passive
- >termination? Does the active termination have a power source?
- >
-
- Passive termination uses 2 resistors - a 220 ohm and a 330 ohm per signal
- hooked up something like this: (aka parallel termination)
-
- TERMPWR (~4.4 to 5 Volts)
- V
- |
- \
- / 220
- \
- |
- +----------- SCSI SIGNAL (i.e. DATA, MSG,CD,REQ,ACK etc)
- |
- \
- / 330
- \
- |
- -
- GND
-
- Active uses a low noise voltage regulator and 110 ohm series resistors
- to terminate each signal:
-
- 110 Ohm
- -------------- |--/\/\/\-----> SCSI SIGNAL
- TERMPWR >---------| Regulator |-----------|
- | | 2.85V |--/\/\/\----->
- -------------- |
- | |--/\/\/\----->
- -
- GND
-
- The passive terminators tend to keep the line at a level more determined
- by the level of TERMPWR since there is no isolation. In addition, any
- noise induced onto the TERMPWR line (its just another wire in the cable)
- will show up at the termination point at the resistors.
-
- Active solves this by regulating TERMPWR to 2.85 volts and matches
- the ideal line impedance of 110 ohms. Be forewarned... this 110 ohm
- impedance is ok for ribbon cable, but in the Sun world we exit the
- system on shielded round cable - closer to 90 ohms impedance! So,
- ideally, you could hope that the series termination resistors are
- in sockets and try to reduce them to 100 or even 90 ohms.
-
- FPT uses a more complex circuit that uses a combination of various
- filter caps, a series resistor and a collection of planer/zener diodes
- on each line. Sorry, I can't draw that one from memory!
-
- Passive was fine for older SCSI (3-4 MB/sec). For a loaded SCSIbus (5 or
- more devices) and/or SCSI Fast (10 MB/sec) it doesn't work. Period.
- Active does a much better job for 10 MB devices but starts to
- fall after 2 or 3 fast devices and/or long (> 4 feet) cables.
- A combination of active (on the data and phase lines) and FPT on the
- critical handshake lines seems to do fine for 4 or 5 fast disks and
- a tape or cd-rom.
-
- Last but not least, you MUST have the same style of termination on
- BOTH ends of the SCSIbus to get the benefits. Active at one end and
- passive on the other is only as good as the weakest link. I have found
- that active only on one end and active/fpt at the other is not near
- as robust as active/fpt at both ends.
-
- --
- ___________________________________________________________________________
- |John Grana, Performance Technologies Incorporated jjg@pt.com|
- |315 Science Parkway, Rochester, New York 14620 uupsi!ptsys1!jjg|
- |Phone: (716) 256-0200 Fax: (716) 256-0791 |
-