($40 GPSDO working well now, my mistake)

[mark03]

I decided to replace the built-in supply with a Mean Well (that name cracks me up every time ) unit, 15W.  Load testing the original supply with an HP 6632B, the voltage sank to ~ 11V @ 0.75A and switching noise also increased markedly.  The new supply is nominally 15V, adjustable to 16.5V, which will provide enough headroom for some RC filtering followed by linear post-regulation.

Edit:  Oops, just noticed (two months late!) that I posted this in the wrong thread.  Should have been here:
https://www.eevblog.com/forum/testgear/diy-10-mhz-distribution-amplifier-requirements-gathering/
which is the chassis I'm planning to use for my Trueposition GPSDO.

[mark03]

I just fired up my Trueposition board this weekend.  Like someone else reported, I could not get mine to leave wait state 19 and start picking up satellites until I issued a $SETPOS command with my lat/lon/elevation obtained from another GPS.  At least, it wasn't making any progress after 1-2 hours with a good antenna.  It's likely that some kind of factory reset would have worked also, but I was too chicken to try that.

Now it is seeing between 4 and 8 satellites pretty consistently with one of those bullet-shaped timing antennas placed next to a window.  Next step is to mount the antenna permanently in the attic.

It does seem to bounce between status 0 (normal) and 1 ("acquire"?) quite frequently.  Not sure if this is something that can be resolved with a better signal.

I also have the Lady Heather beta version v6.05 running with it (thank you texaspyro!).  It does not seem to track the EFC tuning voltage by default, or maybe I just need to read the manual?  Anyway, this has me leaning away from the STM32 controller option and towards a Raspberry Pi + LCD display running LH instead.  Is the touchscreen functionality of LH documented anywhere?  I am trying to decide whether I really need a touchscreen on the LCD for this project...

[mark03]

Do we know what the difference is between status 0 (normal) and 1 (recovery or acquire)?  In Lady Heather, when the board bounces into status 1 it is not clear whether this is considered holdover or not.  Here's an example where it was in status 1 for the middle part of the plot, corresponding to the wild swings in the EFC voltage (green trace):



I can't imagine the 10 MHz remained locked during this time, with the huge EFC spike, although LH considers the total holdover time to be zero seconds.  Possibly a bug?

In any case, I had ~ 5 hours of solid lock before this happened, and nothing obvious to explain why it happened when it did.  I.e. plenty of satellites, good CNR values, ...

[texaspyro]

Do we know what the difference is between status 0 (normal) and 1 (recovery or acquire)?  In Lady Heather, when the board bounces into status 1 it is not clear whether this is considered holdover or not. 

Ant time the state is not 0 (normal) , Heather flags that with the red line at the top  of the plot.  State 1 is not an actual holdover.   There is a message that shows the holdover time during actual holodver events.  There is no documentation that elaborates on what the states actually mean.  Most of the info on these receivers comes from what limited manufacturer data is available and from reverse engineering and scientific wild ass guesses.

The EVAL values comes from the EXTSTATUS message.   Again, no documentation what it means.   It seems to be noisy during surveys and fairly stable during normal operation.

[mark03]

Ant time the state is not 0 (normal) , Heather flags that with the red line at the top  of the plot.  State 1 is not an actual holdover.   There is a message that shows the holdover time during actual holodver events.

Hmmm.  If the EFC voltage trace is correct, then state 1 may be considerably worse than a brief holdover... because in the latter case you'd expect the controller to keep the OCXO on a fairly even keel while the GPS timing is missing, whereas here it seems probable that there is a major frequency deviation at the 0->1 state transition.  This assumes there is no other mechanism to explain why a sudden jump in EFC voltage would be needed---I can't think of one.  Unless the OCXO itself is "hiccuping"... for lack of a better term?

I guess it's a good reason for running >1 unit, to investigate questions like this.

[RoadRunner]

My distribtion Board and Enclosure

https://www.eevblog.com/forum/oshw/open-source-trueposition-gpsdo-isolated-distribution-board-and-enclosure/

[texaspyro]

No well in hell will Heather ever run on a screen smaller than 640x400.

Well,  hell may have left the gate open and Heather got in and did some tweaking.   Heather has a feature to draw text using scaled vector fonts (instead of the normal fixed sized dot matrix fonts).  Vector fonts were originally added so that if you had a very big screen and the largest dot matrix fonts were too small... they can be scaled up to 500%.   But they can also be scaled down as low as 50%.

Sparkfun now sells a 3.5" 480x320 LCD touchscreen hat for the Ras PI ($25), so I tweaked up the screen formatting some and added 480x320 and 320x480 modes and also added an option for rotating the drawing 90 degrees.  This seems to work better than expected.   I have the Sparkfun board on order... I'm a bit worried that the 16 MHz SPI interface might be a bottleneck.

[mark03]

Cool!  Please post again and let us know how it works once you've had a chance to test.  I was tentatively planning to mount a Pi running Heather in my distribution amp enclosure, but was put off by the choice between an ugly HDMI connection, or the parallel connection which uses up every last GPIO pin on the header.  SPI would be much more tolerable if it works.

[texaspyro]

Cool!  Please post again and let us know how it works once you've had a chance to test.  I was tentatively planning to mount a Pi running Heather in my distribution amp enclosure, but was put off by the choice between an ugly HDMI connection, or the parallel connection which uses up every last GPIO pin on the header.  SPI would be much more tolerable if it works.

The official PI 800x480 screen uses it's own LCD interface connector and works quite well.

[Towger]

Have you looked at the ability to generate timecode (LTC/SMPTE) from Lady Heather?
 I believe there are already libraries for the Pi.  It could be used for driving studio clocks, giving a nice clean/large time display.

[texaspyro]

Have you looked at the ability to generate timecode (LTC/SMPTE) from Lady Heather?
 I believe there are already libraries for the Pi.  It could be used for driving studio clocks, giving a nice clean/large time display.

I looked into generating IRIG time codes, but there is too much system dependent stuff to make it practical.  It could possibly be done for a specific system /  configuration, but doing it in a generic way is not really practical.

Heather's full screen  clock displays (ZC or ZW commands) makes for a nice wall clock display.

[cdev]

You people who are using it with LH absolutely must use its nice stabilized 1PPS as the PPS for a locally running copy of chrony, NTPD etc. Its nuts not to. The GPSDO has the potential to be substantially more stable PPS than a plain old GPS. A GPS is likely to stairstep within a range of around 25 or 30 nsecs. A very short amount of time but still it matters if you are doing scientific experiments right on your Pi. (Anywhere else it wont matter, however, the network latency will eat up a small but significant chunk of that accuracy, because of the way RPI's network card is done-to shave off a few bucks, via its USB bus.) Other slightly less inexpensive SBCs with "real" ethernet cards that pair them with the nice fast GPIOs would be a better choice for REALLY precise timing use with a network because of the USB latency of the Pi's network card issue. Other users here probably know a lot more about this than I do.

However, we're splitting hairs here because all the differences usually only amount to less than a millisecond at most on a lightly loaded machine thats adjusted to minimize its network latency. You will still have a Stratum 1 time server either way. Switching from a generic GPS PPS (some are quite good) to a GPSDO ('smoothed out') PPS matters a lot but only to other applications running right on the same machine or via some very low latency link (direct connections or fiber optics - light would be two ways to do it) that can take advantage of it accuracy directly.

You can see the jitter and the effect of different changes using the 'ntpq' command.

[texaspyro]

I got in the Sparkfun LCD.  It does work, but has some nasty gotchas that make it very difficult to use with a PI.   The main issue is that dialog boxes do NOT scale to the small screen size.  That means whatever is towards the bottom of the dialog box (like CANCEL / ACCEPT) cannot be accessed!  Also once the terminal window gets more than a few lines on it, you cannot see any more output or what you are typing.

Finally, the touchscreen hardly works with user / X11 programs.  Maybe if you tap an item a couple dozen times it will register a click.  Not sure what is going on here,  the touchscreen seems to work OK with Raspbian generated stuff.  If using a mouse, the mouse pointer lags behind mouse movement quite a bit.

That said, except for the touchscreen issues, it does work with Heather.  No real issues with the screen update rate.

[texaspyro]

I did some more testing with the touchscreen... if you hold the stylus down on an item for around 2-3 seconds it registers the click... makes typing on Heather's on-screen keyboard painfully slow, though.

[tkamiya]

Just bought 3 TruePosition board for cheap.  I'm joining this thread.  My current goal is to have it boxed up with Adouino and display, and add distribution amp to it at later date. 

Has anyone tested its frequency stability short and long term?  I do not have an ability to test for this myself.

[tkamiya]

I'd like to modularize (is that a word?) my construction.  But something like this is an unknown territory for me.  Being self-controlled ovenized oscillator (that heats up), how do I manage cooling and venting? 

I've seen commercially made unit with vent holes only.  Its manual specifically says convection provides enough cooling that install it vertically and NOT to have direct fan air.

I've also seen DIYer install boards into a case with various insulation materials, and pretty much seal it. 

Any advise here from experience would be great.

[texaspyro]

Has anyone tested its frequency stability short and long term?

Note that for tau<50 secs the readings are dominated by the TICC's 60 ps resolution.

Also, the plots have a (52 second) display averaging filter applied.  The actual reading span values are around twice what is shown in the plot labels.

[cdev]

This is a new one, and it looks quite a bit better !

Is that right?

Has anyone tested its frequency stability short and long term?

Note that for tau<50 secs the readings are dominated by the TICC's 60 ps resolution.

Also, the plots have a (52 second) display averaging filter applied.  The actual reading span values are around twice what is shown in the plot labels.

[tkamiya]

Is it true that this particular GPS lacks sensitivity compared to more modern ones?  While I can get all 8 satellites with GPSTM, I am getting 5 to 8 on this. 

I am little concerned that we don't know what status 1 mean....  Mine flips between 0 and 1 fairly often.  If it means locking and synchronizing is compromised, it is unusable.  I read TexasPro commenting it probably doesn't mean fully unlocked. 

[cdev]

Tkamiya,

Mine was only 'really' insensitive at the beginning when it was still figuring out that it was in the US rather than in Japan. After a bit of time running off of an outdoor antenna it stored the data and got a minimum of five or six sats fairly consistently after that, which is enough to maintain the time with a high level of precision. Thats why they picked the Furuno timing unit which is known for that ability.

Now my GPSDO (when its on) is running off of an antenna slapped on top of an AC unit, outside but underneath an overhanging roof with its metal gutter. It works fine there. It also works okay (not great, but it maintains lock) with the antenna indoors as long as the antenna has a good ground plane.

No, its not as sensitive as my Skytraq GPS which is much newer. That never loses lock even with a passive indoor antenna. Its incredibly sensitive.

Nor is it as sensitive as my ublox GPS which is also quite new.

Its probably around the same as my Mio with sirfstar III chip, a PND, a car GPS.

It likes to have a good sky view if possible.

Its gotten better with time, likely because it trains itself as to its surroundings, case, position, etc.

I should leave it on all the time, but haven't been because mine is still not in a permanent case.

I am probably going to encase it in an extron case I have, sharing the inside of my video distribution amplifier's case.

That seems like the best plan for now.

---------------

There are some TruePosition GPSDO videos on YouTube now!

The first one is by YouTube user CircuitValley who also is a user here.

[tkamiya]

I am kind of recalling GPSTM took a while to lock first few times.  So hopefully, this one will "learn" and stays locked. 

I have an 1U case I can use.  I just have to decide if I want to keep GPS and distribution amp together or separate.  Right now, I only have 5 equipment that need external sync.  But this is expected to grow.  I thought an ability to remember the last location was generic to all GPS....

I actually have an ability to have an outside antenna but Kind of figured with shortened cable length will cancel out attenuation by going the attic route.  Also, my 4 way splitter has a built-in amp to compensate for the loss. 

My setup is on desk and not in case.  Because the OCXO gets quite hot, I didn't want to keep it running unattended.  Just a bit scary.  I hope to encase them in next few days.  Then, I will probably interface it with Lady Heather and keep it running.  I want to ultimately rack mount all of them and keep "house clock" all together.

Thanks for the video.  I thought the first video was yours....  no?

[cdev]

GPSDOs have ovens so they all get warm.

They have to. Mine gets toasty warm, but not really hot. Not painful to touch. If its working properly it should have a stable frequency. Its going to keep the oven at whatever temperature it needs to to keep that frequency where it needs to be. It has to be hotter than any normal ambient temperature, otherwise it wouldn't have any ability to go lower below that even if it turned the oven off. So it has to be quite warm.

As far as sensitivity, if you are using a patch antenna, make sure it is on a flat, horizontal conductive ground plane, as large as is practical.

A CD/DVD is around the minimum size that I think is enough. But a little bit larger than that is even better.

[tkamiya]

No, but I don't think this kind of "Bullet" antenna depends on ground plane.

[cdev]

Do you know what's inside that particular bullet? If it uses a quadrifilar helix, no it doesn't need a ground plane and perhaps could even be mounted sideways and still 'work' kinda sorta.  Thats a big advantage of them. They work in any direction. They will reject the first order (odd) reflections coming from any direction.

Your photo looks like it should be rotated 90 degrees to the right.


You can mount it to the inside wall but your GPSDO would likely work much better if you put it outdoors and away from metal things.

Wood also will attenuate the signals a fair amount in that direction and you likely will be able to see it in the pattern built up over time, All the satellite in that direction, a good chunk of the sky, will be weaker.

if you use software (like rtklib or lady heather) that records SNR in a manner you will see this- Both will show you the polar plot with signal strength.

QFH antennas can be surrounded with metal junk and still work okay. But they do best just on a pole that supports them from below.

If you use a 90 degree elbow, if you don't put a common mode choke ferrite bead right where the feedline leaves the antenna you may get a lumpy pattern from the transmission line going off to the side.

This is a useful resource to see what happens: http://gpspp.sakura.ne.jp/anteva/anteva.htm
 

[tkamiya]

Would you or someone else tell me what I2C address Pakrat code is expecting?  Not all LCD boards use same address and source is not available.