[time-nuts] GPS receiver weirdness
Sean Gallagher
sean at wetstonetech.com
Thu Mar 10 10:15:53 EST 2016
Good morning time-nuts,
I just sent this email to Olivier who was our contact over in France
with Heol Designs. They made the replacement board with the same form as
the Trimble Ace III that would work with our buggy TS2100 machines and I
have recently run into problems and I wanted to share this with you all.
Forgive me for the length I was really just doing a brain dump of
everything I've noticed so far:
"
Hi Olivier,
Okay, so I spent all day yesterday testing out various things with one
of my interns. This is a bit confusing to follow even when I was trying
to explain it in person to my boss, so if you have any questions just
let me know.
We have two TS2100 machines. One is from Datum and the other is slightly
newer from Symmetricom (Who bought Datum and is now owned by Microsemi).
We were using the Datum 2100 for the testing purposes since it happened
to be the one plugged in but they are both running v4.1 firmware. I also
just hooked up the newer one and plugged in your receiver to it with the
same results which I'll explain in a second.
Besides the two operational servers with the bc635PCI cards in them we
have a couple extra servers and about 20 spare 635 cards. I have the
spare servers off to the side and disconnected from the production
environment. They are there for testing purposes for things like this,
and we are trying to upgrade software too.
We purchased 3 N024 receivers from you and had one in the hooked-up
2100, and one each in my two operational servers attached to the 635
frequency processors.
I should mention that we currently have what we are thinking of as 4
"generations" of GPS receivers. Besides yours, which we know is the
newest, we are basing these generations mostly on the appearance of the
cards as there are no dates or anything on them. Things like the size of
the components and the "shininess" of the cards.
Gen 1: I only have one of these and I think it came out of the oldest
2100 but I'm not 100% sure because I can't remember.
Gen 2: I have a few of these. One was in the newer 2100, and others were
on 635 cards and laying around
Gen 3: Shinier and newer looking. I have a few of these as well
Gen 4: The Heol design cards.
Generation 1-3 cards all had the rollover issue a few months ago that
caused us to purchase the units from you.
This is where things start to get weird.... A couple of weeks ago I went
to switch some things around and have my 2100 output timecode into one
of my production servers. When I did it and went to audit the server
against the second server (still running GPS) it failed because the time
was 1 second fast. This was when I first noticed something was wrong. It
may have been like this earlier but since we weren't really using the
2100 I hadn't looked closely at it.
This led to yesterday and me trying all of the GPS receivers I have in
the 2100, and also trying them in my test environment too with 635
cards. Results when compared to nist.gov and the GPS time displaying on
the monitors for my servers were:
TS2100
Gen 1 card: Time and date are correct, the UTC offset of the card
is +17
Gen 2: Time and date are correct, UTC offset is +17 (we did have
one that failed to ever get time)
Gen 3: Time and date are correct, UTC offset is +17
Gen 4 (Heol): Date is correct, time is +1 second fast, UTC offset
is +20
bc635PCI frequency processor cards
This is a little more difficult because I currently cannot get ANY
of my receivers (Gen1-4) to work with any 635 cards in my test
environment. For the most part it looks like the 635 is recognizing that
a GPS card is hooked up but it's not translating any data.
I can say in the two production servers that each have one of the Heol
cards hooked up to a 635 card that the time and date are both correct,
and the UTC offset is still +20.
The actual UTC offset is currently +17 since the leap event a few months
ago. So needless to say I am thoroughly confused as it looks like
nothing should be working on paper at this point but somehow is.
And I'm not sure if the 635 cards have run into some kind of end of life
issue that is causing them to not work with GPS anymore, and the only
reason my 2 production servers work still is because they haven't been
powered off or some other issue....
"
Respectfully,
Sean Gallagher
Malware Analyst
571-340-3475
More information about the time-nuts
mailing list