[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gpsd-users] Trouble Receiving Time from Delorme EarthMate USB (LT-2

From: Jeff Morris
Subject: Re: [gpsd-users] Trouble Receiving Time from Delorme EarthMate USB (LT-20)
Date: Thu, 14 Mar 2019 03:19:17 -0700
User-agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:52.0) Gecko/20100101 Thunderbird/52.9.1

On 3/13/2019 12:39 PM, Gary E. Miller wrote:
Yo Jeff!


On Wed, 13 Mar 2019 08:47:20 -0700
Jeff Morris <address@hidden> wrote:

I recently came across a Delorme EarthMate USB (LT-20) at a thrift
shop for a couple of bucks, and picked it up with the hope of being
able to use it as an NTP stratum-0 time source for an NTP server on
my home network.
Ever wonder why it was so cheap?

Well as I said, "thrift shop", lol. Everything there is cheap.  :-) (I've actually gotten quite a few good deals on older gear there. Lots of fun projects. I even found a working Raspberry Pi with a case and SD card for $5 one day.) So no, I didn't really question it.

You can buy real good new GPS for around $10.

Good to know. The last time I thought about doing something like this and looked into it (probably 5-6 years ago), they were nowhere near that inexpensive. More like $100+ and I just couldn't justify it for a toy. I also tend to shy away from the dirt cheap "straight from China" stuff that the market seems so flooded with these days, since I've found so much of it to be just plain bug-infested, unsupported junk. I figured buying the older Delorme would be a better/more reliable route to take. Silly me. :-( Do you have any specific recommendations for a good unit for the $10 you quoted?

I've managed to get gpsd installed and running, and it sees the
EarthMate. The EarthMate acquires satellites and gets a fix and shows
my correct position, so it seems to be working properly. However gpsd
is reporting a date from the EarthMate far in the future. For example
for today, 2019-03-13, it's reporting the date as 2099-07-28. The
time is correct, but the date is way off. If I look at the raw time
in seconds-since-the-epoch format, I see numbers like 134439.000.
(These numbers are both from the output of gpsmon.)
Surprise, your unit has the GPS week rollover bug...

Yup, I kind of suspected that. The numbers just weren't making sense to me though. I was also hoping there might be a workaround.

I've added the following lines to /etc/ntp.conf (and removed all
other servers):

server minpoll 4 maxpoll 4 prefer
fudge time1 0.0 refid GPS
NTP Classic or NTPsec?

Classic (the one distributed by I'm running CentOS, so it's just the off-the shelf ntp package included in the EPEL repo

However, ntpq -p reports that it has never successfully polled this
Before you look there, does the cgps output look good?

Cgps output looks good, in the sense that I can see it has acquired satellites and is giving me valid location data (disclaimer: I'm new to gpsd, so I may not be qualified to determine if it's output is really "good" or not.) :-) Here are a few lines of output from it. (I have no idea why that second line seems truncated. Some sort of ncurses issue possibly):



What does "ntpshmmon" show?

Hmm. I don't seem to have an ntpshmmon program in my distro. I don't see it in any other packages in the standard repo either. Is it perhaps only included with NTPsec?

My guess would be that the bad date has the time stamp thrown away.

I'm aware of the GPS week rollover issue, and given that this is an
old receiver that I picked up at a thrift shop it may very well be
susceptible to that.
You already confirmed that with the bad date.

  However, the numbers I'm seeing don't make any
sense to me. The 10-bit week rollover issue should result in a
rollover every 20 years? The date I'm seeing is 80 years in the
Yup.  A century minus 20.

Ah! Ok, that makes more sense. I was trying to figure out the scenario under which I could end up with a date 80 years off.

The Delorme were not very good when they were new, and now they are
really old...

Well, I'm old myself and still ticking ;-) and I've always gone with the philosophy that if it isn't broken, no need to fix it, so I figured an old unit would work fine for such a basic need as just pulling the timestamp from GPS. Le sigh.

While searching for a solution to this, before I posted to the list, I came across an old thread where someone was discussing the possibility of adding a fix for the rollover bug to gpsd. I think the general consensus was that since it isn't a gpsd bug, that gpsd shouldn't be the place to fix it, plus there are challenges in coming up with a workaround that could automatically and reliably determine what offset to use. I understand the philosophy, but I do wish there was a way to just manually set the epoch date or something like that, for cheap old farts like me that just want to use the hardware they already have, and who enjoy the challenge of getting obscure, broken stuff to work. :-)

Thanks for your help! Again, if you have a recommendation for a $10 unit that will work reliably, I'll probably end up going that route.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]