[time-nuts] Set time on Solaris computer from HP 58503A

[Bob Camp]

Hi


> On Dec 14, 2014, at 4:26 PM, Hal Murray <hmurray at megapathdsl.net> wrote:
> 
> 
> gigneil at gmail.com said:
>> Based on my recent testing - including Solaris - you will be better off with
>> the Internet unless your USB adapter is far better behaved than the several
>> I have here 
> 
> That depends,  (TM)
> 
> How good/bad is your network connection?  Mine gets over 3 seconds of queuing 
> delays.  I challenge anybody to find a USB device that bad.
> 
> There are only 2 or 3 significant vendors of USB-serial chips.  I think they 
> are reasonably well supported by all major OSes.
> 
> USB is polled, so interrupt latency turns into polling latency.  I think the 
> polling cycle is 1 ms for slow things like serial ports.  

On at least some (= the ones I’ve seen) of the serial devices, they will continue to buffer if they have a character coming in. Put another way, a string sent at 19.2 Kbaud will likely transfer as a block rather than a character at a time. Is this  at the bus or the driver level? - who knows. The result (in Linux / Win-dooze / or OS-X) is that data comes in in bursts. 

> Maybe 1/4 ms for 
> faster things like disks.  At the system level, faster polling means more 
> overhead and slower polling means bigger buffers.
> 
> Many of the low cost GPS units use the SiRF chips.  They have a wander of 
> ~100 ms.  I said wander rather than jitter because it's very slow as in 
> hours.  You can't filter it out with a 10 or 100 second sample.

.. as in a pps output is not  necessarily  a *useful* pps output. Only useful outputs count. We’ve been over the why and the how of chips that put out pps’s way late a number of times. Simple answer - it didn’t matter in the firmware design spec. 

Bob

> 
> The old Garmin GPS-18-USB (not 18x) units had pretty good timing.  No wander. 
> Unfortunately, they weren't very sensitive.
> 
> 
> -- 
> 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.