[time-nuts] Timelab, two SR620s and losing samples

[Poul-Henning Kamp]

--------
In message <569BCEBE.20307 at erols.com>, Chuck Harris writes:

>Talk-only mode is by intention, an exclusive mode, where
>there is one talker, and one listener on the bus.

Wrong.

There can *always* be multiple listeners on GPIB, this is
why it has a three wire handshake.

The slowest listener sets the pace, individually for each byte.

>GPIB is half duplex.

Correct.

>You have to turn the line around to tell it 
>give-me-the-next-sample.

Wrong.

There is no requirement to turn the line around (= send commands)
between measurements.  EOI is only indicative, it does not require
an UNTALK+TALK sequence.

Most instruments can be commanded to measure and spew data at you
in "talk-only" fashion.

> Talk-only avoids that.

Talk-only is one way to avoid that.


As I said from the start:  Talk-Only doesn't really make that much
difference, the three-wire handshake is sufficient to cause trouble.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.