[time-nuts] Progress of my HP Z3801A -- Ooops!

[Magnus Danielson]

From: xaos at arachne.darksmile.net (xaos)
Subject: Re: [time-nuts] Progress of my HP Z3801A -- Ooops!
Date: Wed, 30 Aug 2006 18:11:35 -0400 (EDT)
Message-ID: <20060830221136.703F9B4115 at arachne.darksmile.net>

George,

> First,
> 
> >Ah, much better. After batteling the perl package hell (i.e. finding the
> >exact right Debian package that will make it tick) I now have it
> >running, but
> >now I need to figure out how to make it tick on automatic. I also
> >commented out
> >the "cd /opt/ntp/timelord/shm || exit" line from the G1 script. How that
> >one
> >should be set was non-obvious.
> 
> This is the kind of omission that is hard to spot. Actually that
> directory is my source code gen home. I have added a separate comment
> there to clarify this.

OK, good.

> Now, to the more problematic issue of time delay between
> :SYST:STAT? invocation and receiver response.
> 
> I just wish that the information in the :SYSTEM:STATUS?
> command was available separately.

Much of it is. I have reverse-engineered the actual SCPI command structure, but
I haven't had time to test them all. Besides, I actually found an error in my
analysis and haven't had time to fix it.

> That said, it usually takes 4-5 seconds for the receiver
> to respond to this command. As a result, the interactive
> interface is sluggish and the "auto" mode lags behind.
> 
> What to do?
> 
> 1.	Send the :SYSTEM:STATUS? command and report it as-is.
> 	Problems:
> 		Time lag is very obvious but we at least get
> 		what the receiver's state is.
> 
> 2.	Send the :SYST:TIME? and at least get the time.
> 	Problems:
> 		Ok, we got the info but now we need to send the
> 		:SYSTEM:STATUS? anyway. Then for display purposes
> 		we interpolate the receiver time forward.
> 		Not a bad solution but still...
> 
> 3. Option of how the user might want this handled.
> 
> I am open to suggestions here.

First of all, not all info needs to be sharp on the microsecond so to speak.
The EFC, TI and time we would like fairly timely, where as the satellite data
doesn't need to be that timely, but we like it in snapshot form if possible.

Second, you can experiment with other baud-rates.

Third, there is a Forth interprenter in there, maybe that may put into use to
create a less verbose format but in a more timely maner? I haven't toyed with
it yeat, but I think I know how to kick-start it. ;O)

Thanks to the logging, I now see that my Z3801A has an erruption right now.
I would have enjoyed a TI plot aswell actually.

Cheers,
Magnus