Use GPS Time on PC

[Buzzman]

I like the option that STARTS with "buy this".....


Clearly need to go back to school...

lol

[muttnik]

Originally Posted by CarinaPDX

Older GPS modules output on RS-232-C the NMEA data, which was accurate to the millisecond range. Some receivers (such as the Globalsat BU-353) put out the pulse-per-second (PPS) signal on a second serial port, so if 2 ports were available time was accessible in the microsecond range. The USB pucks of the day used RS-232-C to USB adapter chips, which reduced accuracy significantly (if milliseconds count as significant). Modern modules put out USB from the module and can provide time with the USB output synched to the PPS (approx) - think sub-microsecond accuracy - with the right software and hardware. By right I mean a dedicated Pi with minimalist Linux install and GPSd. These can be used as accurate time sources for laboratories, short of atomic clocks. The reason a Pi with Linux is that it isn't dealing with a lot of interrupts unlike Win/Mac laptops and desktops which create variability.

Apologies but I question some of these assertions. If you think about it an RMC sentence is going to take well over 100ms just to transmit over 4800 bps. I don't have a huge sample size but what I looked at the offset was half a second-ish with huge jitter. The BU-353 doesn't have a PPS output. Neither does the 355. If you think about how the NMEA string is being transmitted it's not going to be microsecond accurate. You need a separate pulse output.

Also...whilst building a pi time server was a fun project and fine for your home network, how are you going to get the time out of the pi? Over the ethernet interface hanging off usb? I suggest this wouldn't be appropriate for a physics lab or some high frequency trading application.

[rgleason]

Muttnik
Do you have any idea why this guy's wifi is timing out?
I haven't been able to get to the bottom of it with him and have run out of ideas.
Open CPN won't receive XB-8000 Data over WiFi - Page 2 - Cruisers & Sailing Forums


It is something to do with his software or hardware wifi I think.

[El Pinguino]

Originally Posted by Dockhead

It all depends on what you want to do. If you're just following the dot on the map, like in your car, then IPad nav programs are fine. I have an IPad on board and use it myself, or used to -- I haven't turned it on in a few years.


If you're doing actual navigation, and even more, if you're keeping your log and so forth electronically, if you're doing complex collision avoidance, operating your radar, etc., then you need OpenCPN or something with similar capabilities, and there isn't really anything for Fapple.



I don't like playing with computers at sea, for sure, but there are certain jobs to do which require good functioning IT systems, so sometimes you have to mess around with them.

Like Ken I have a Macbook Pro.... runs OpenCPN just fine...upgraded to 5 just the other day.

Also have a PC running W7 that drives me to distraction..... used primarily for Sailmail and 'big screen' navigation.

Can run everything on the Mac including Sailmail using VM Fusion.

Its a bit like driving a car where you aren't spending half the day by the roadside with the bonnet/hood up...

I know... day late... dollar short.

[CarinaPDX]

Originally Posted by muttnik

Apologies but I question some of these assertions. If you think about it an RMC sentence is going to take well over 100ms just to transmit over 4800 bps. I don't have a huge sample size but what I looked at the offset was half a second-ish with huge jitter. The BU-353 doesn't have a PPS output. Neither does the 355. If you think about how the NMEA string is being transmitted it's not going to be microsecond accurate. You need a separate pulse output.

Also...whilst building a pi time server was a fun project and fine for your home network, how are you going to get the time out of the pi? Over the ethernet interface hanging off usb? I suggest this wouldn't be appropriate for a physics lab or some high frequency trading application.

I used the wrong number, I meant the BU-350P, that have the separate PPS signal. The "P" stands for PPS. My mistake.

I said without the PPS signal it would be millisecond accurate, not microsecond, by which I meant in that general range.

With modern GPS modules the USB transmission can be timed to provide the PPS. I don't know how it is synched with the sentence but if it is important to you I can find out. My brother is the one who writes that code in GPSd, and without that code there is no PPS synching with USB. I was actually being conservative on the accuracy as he says it is down in the nanosecond range for his test conditions, but he is accomplishing this by keeping the Pi in a thermally stable case (Rubbermaid container with lamp and thermostat) and leaves them running for days in order to characterize the systems.

Of course ethernet would not be used in lab conditions for precise time measurements, although it could be used as a very accurate network time server. Precise measurements would be done using the GPIO pins on the Pi, which would provide the most direct (i.e. least buffered) connection to the cpu. This is certainly accurate enough for anything that doesn't need atomic clock levels of accuracy. Admittedly there are many that are skeptical, until they actually invest the $100 and try it themselves.

Greg

[Dockhead]

OK, thanks to helpful suggestions on here, I found several small apps which can update the PC clock with time data from GPS. Here is a list of them: GPS Time Sync Software for Windows OS – OH8STN Ham Radio


Some of them will update the clock every minute, so this is a great solution for me.


HOWEVER, they all seem to need serial data from a serial port. I'm getting my 0183 data over Ethernet, in GPSD protocol.


So it would seem to be a snap to configure a virtual COM port on OpenCPN, configured to transmit $GPRMC, but this function was built for autopilots, and among the various 0183 sentences you can transmit over the handy dandy virtual port, $GPRMC is not among them!


Anyone know a workaround? Or could the OpenCPN wizards add this sentence, to the virtual COM port?


In the worst case I could buy a cheap GPS dongle, or add a serial connection from my plotter, but the OpenCPN virtual COM port would be tarts for this application!! I bet I'm not the only one who would like to do this.

[nohal]

Originally Posted by Dockhead

So it would seem to be a snap to configure a virtual COM port on OpenCPN, configured to transmit $GPRMC, but this function was built for autopilots, and among the various 0183 sentences you can transmit over the handy dandy virtual port, $GPRMC is not among them!

Not sure I understand what you talk about, but if it is the dialog to add filter to a connection in OpenCPN, look about two centimeters to the right of the list of the predefined sentences, click on the Add button you find there and add whichever additional sentence not already there.

[Dockhead]

Originally Posted by nohal

Not sure I understand what you talk about, but if it is the dialog to add filter to a connection in OpenCPN, look about two centimeters to the right of the list of the predefined sentences, click on the Add button you find there and add whichever additional sentence not already there.

Duh! Thank you!!


And it WORKS!


I had to install Comm0Comm virtual serial ports to get GPSTime to connect to OpenCPN, but that was a snap.


Now O is sending out $GPRMC over virtual COM6, and GPSTime is listening to the paired virtual COM7, and the clock is updating once per minute!! Success!!


Thanks a lot, Nohal!

[rooiedirk]

You might as well tell OpenCPN to de the job. No need for extra utilities and virtiual com ports.https://opencpn.org/wiki/dokuwiki/do...iguration_file
Although this will be done only once per session.

[Dockhead]

Originally Posted by rooiedirk

You might as well tell OpenCPN to de the job. No need for extra utilities and virtiual com ports.https://opencpn.org/wiki/dokuwiki/do...iguration_file
Although this will be done only once per session.

I need it done more than once per session.


GPSTime is NOT doing the job after all. It acknowledges receiving the data and claims to have updated the clock, and it APPEARED to be doing so, becacuse the clock was still running. But then the clock stopped and it became evident that the update was not working, despite the claims of the app


So back to the drawing board.


Perhaps this app is not working with Win10. I will try some other apps.

[Dockhead]

In parallel with trying to get the clock reset once permit to GPS time, I am trying to get the internal clock working. I have taken the box apart and found the CR2032 3v lithium button cell.


The leads are soldered or perhaps welded onto the cell.


My multitester makes 3.00 volts -- does that mean that the cell is not bad?


Why else would the bloody clock not be working?

[CarinaPDX]

The leads are spot-welded onto the coin cell. You, or at least a technician, can replace the cell by de-soldering the leads and soldering in a new cell with leads. If you are planning on keeping this computer for long I think it makes sense to solder in a 2032 battery holder, which would allow easy cell replacement. I did this with an old ham radio. Needless to say it is a lot harder to find a 2032 with welded leads than a plain cell.

The 2032 cell should test at between 3.0-3.4V so it should be working but my guess is that it may be close to EOL if not actually there, and the clock circuit may not like the 3.0V.

Greg