[time-nuts] Practical considerations making a lab standard with an LTE lite

[Dave Martindale]

The 20 MHz output should be OK, since it is series-terminated with 50 ohms
at the source and the buffer can source enough current.  The driver sees a
100 ohm load (50 ohm resistor in series with 50 ohm coax impedance) for
that 32 ns round trip time, so it will increase power dissipation (as you
note).  But the load at the far end of the coax should see a clean edge,
and the reflection should be absorbed when it returns to the source (due to
the source terminator).  Just don't look at the signal half way along the
coax.

The other outputs apparently don't have either the current drive or the
source terminator, so a long piece of coax is likely to do unpleasant
things to the edge.

In either case, if you want to run any of the signals 10 feet it's likely
better to run a very short connection from the LTE-Lite to a proper 50 ohm
line driver.  That gets the power dissipation off the board, and then you
can use drivers that give you whatever output swing you want, and which can
drive a 100 ohm load continuously so you can use parallel termination at
the far end.

- Dave

On Tue, Nov 25, 2014 at 12:23 AM, Hal Murray <hmurray at megapathdsl.net>
wrote:

>
> Said Jackson said:
> > Correct, and thats why its all a bad trade off if you have to use 50 Ohms
> > termination. Either more heat or more PN, and more circuitry.
>
> > So driving 50 Ohms inputs is not optimal here, 1M inputs are much better
> for
> > this purpose.
>
> That only works if you have a (very) short connection to the next stage.
>
> Things get interesting if you have, say, 10 feet of unterminated coax.
>
> 10 MHz is 100 ns, or 50 ns between transitions.  Coax is ballpark of 5/8 c
> so
> that's 16 ns one way or 32 ns round drip.  That's 60% of the heat as well
> as
> lots of nasty reflections.
>
> (Somebody please check my numbers.)
>
>
> --
> These are my opinions.  I hate spam.
>
>
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>