[SI-LIST] : RE: [SI-LIST]: Long bus or star?

Jon Keeble (j.keeble@fairlightesp.com.au)
Wed, 31 Mar 1999 16:15:35 +1000

This is a multi-part message in MIME format.

------=_NextPart_000_009B_01BE7B91.B58FFF20
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

>John --
>What a great idea to series source each leg at the hub! Very, very =
nice.
>Shayle

Wow. With that encouragement, I'll try again.

If there are many arms to the star, then even halving the load seen by =
the
driver might still be too much.

Active termination might be the answer: this involves receiving with an
invertor, and feeding back the output of the invertor to its input with =
a
resistor, say 3Zo.

This does create another return wave Tpd after the incident wave, but =
this
is also absorbed at the hub.

This has the effect of moving the transmission line back toward the =
opposite
level, reducing the swing (current step) required of the source.

In other words, the idea is to trade off voltage swing for impedance
matching.

This relies on the switching point being at around half rail. Most 3.3v
logic seems to switch very close to 1.5v, with not a lot of variation =
with
temperature. So reducing the swing from 3v to 1v should be quite =
acceptable.

A related idea is recommended by Analog devices for the RDn and WRn =
lines on
groups of multiple SHARC DSPs (the opposite problem, where a good signal
needs to be generated at one SRAM from one of say 4 SHARCs).
This can be found in Chapter 11 of the SHARC USers Manual, available at
www.analog.com

Jon Keeble

------=_NextPart_000_009B_01BE7B91.B58FFF20
Content-Type: text/html;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD W3 HTML//EN">

>John --
>What a great idea to series source each leg at = the hub!=20 Very, very nice.
>Shayle

Wow. With that encouragement, I'll = try=20 again.

If there are many arms to the star, then even halving the = load=20 seen by the
driver  might still be too much.

Active = termination=20 might be the answer: this involves receiving with an
invertor, and = feeding=20 back the output of the invertor to its input with a
resistor, say=20 3Zo.

This does create another return wave Tpd after the incident = wave,=20 but this
is also absorbed at the hub.

This has the effect of = moving=20 the transmission line back toward the opposite
level, reducing the = swing=20 (current step) required of the source.

In other words, the idea = is to=20 trade off voltage swing for impedance
matching.

This relies on = the=20 switching point being at around half rail. Most 3.3v
logic seems to = switch=20 very close to 1.5v, with not a lot of variation with
temperature. So = reducing=20 the swing from 3v to 1v should be quite acceptable.

A related = idea is=20 recommended by Analog devices for the RDn and WRn lines on
groups of = multiple=20 SHARC DSPs (the opposite problem, where a good signal
needs to be = generated=20 at one SRAM from one of say 4 SHARCs).
This can be found in Chapter = 11 of the=20 SHARC USers Manual, available at
www.analog.com


Jon=20 Keeble
 
------=_NextPart_000_009B_01BE7B91.B58FFF20-- **** To unsubscribe from si-list: send e-mail to majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP. si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list ****