Poor man's GPSDO experiment

[Mike S]

Not much time left until the eclipse, and I still don't have my station ready.

Tried a quick and dirty way of making a GPSDO, described here by Ricardo CT2GQV:

http://speakyssb.blogspot.com/2019/10/si5351-gps-disciplined-oscillator-with.html

He had the 1PPS output of Neo-7M module programmed to 24MHz (an even divider of the internal 48MHz clock).

It kinda works, but the quality of the Si5351a output is not good, I can hear some hum and noise, and even the frequency is off by several Hz. So I looked at the 24MHz output of NEO-7M with an SDR receiver and HDSDR. If set to 24MHz exact, the signal has sidebands spaced about 70Hz apart, slowly moving around. If program the PPS output to 24,000,070Hz, the sidebands are gone on the spectrum, but there are clicks in the SSB demodulated audio, again changing in frequency. Looks like the circuit inside NEO-7M is not a PLL, but some digital algorithm removing or adding pulses to the output pulse train to keep the average frequency GPS locked. The raw pulse train (before adding/removing pulses) is generated by some kind of VFO which is not very stable.

Hope this saves some effort for people who are tempted by the simplicity of this approach.
Mike AF7KR

[John Ackermann N8UR]

Hi Mike --

Yes, the "RF" output from the u-blox GNSS modules* has a horrible
spectrum. It's generated by a numerically controlled oscillator and I
think that in effect there are pulses added/dropped as necessary to keep
the average frequency correct.

The spectrum also changes depending on the frequency and there is a
"magic number" frequency that has a lot fewer spurs than others, but
even that is too grungy for direct use in analog systems.

To use RF timepulse outputs for radio applications you need run them
through a clean-up loop with a crystal oscillator to improve the phase
noise and reduce the spurs. Depending on the performance you want, that
can range from a simple voltage-controlled crystal oscillator to a
relatively expensive TCXO or even oven oscillator.

73,
John

* There are a few exceptions. The u-blox LEA-M8F module has a 30.72 MHz
TCXO output that is quite clean, but it's a quite obscure and pretty
expensive part.
----

> --
> Please follow the HamSCI Community Participation Guidelines at
> http://hamsci.org/hamsci-community-participation-guidelines
> <http://hamsci.org/hamsci-community-participation-guidelines>.
> ---
> You received this message because you are subscribed to the Google
> Groups "HamSCI" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to hamsci+un...@googlegroups.com
> <mailto:hamsci+un...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/hamsci/894a70e8-9c82-4292-9fff-3180bbf8d05an%40googlegroups.com <https://groups.google.com/d/msgid/hamsci/894a70e8-9c82-4292-9fff-3180bbf8d05an%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Terry Bullett]

You will find out with these type of measurements that phase noise in your LO is your worst enemy.  A good temperature controlled quartz oscillator has decent phase noise.  Most signal generators have terrible phase noise.  And this phase noise passes strait through your receiver into your data like a dose of salts.  The very process of applying discipline to an oscillator for long term frequency accuracy necessarily adds phase noise.  The usual way to do this is with a phase locked loop, and the PLL determines how the phase noise is 'spread around' the frequency domain.  Every GPSDO maker has their own scheme for how they 'spread this manure around'.  You can spread it thin then it is everywhere in small quantities, or you can pile it up in one place but the pile is high and deep.  Look up the term 'Allan Variance' from NIST.

Even WWV itself has this problem. Their frequency reference is a cesium source which is 'disciplined' by the cold ion fountain maser in Boulder.  The cesium sources in Ft. Collins are frequently 'nudged' by  comparison to the Boulder source.  The phase noise from those 'nudges' is spectrum spread around according to some algorithm developed at NIST.

Also note that almost all GPS time receivers are single frequency, and the ionosphere is the major component of the GPS error budget for single frequency receivers. What happens to this error is the choice of the receiver designer, but the major components are in the height and time dimensions.  Some receivers optimize on time if they know they are not moving.  The eclipse WILL create an ionosphere hole and the accuracy of GPS timing WILL be degraded. 

For HF eclipse observations, I think this can be ignored, but it is still real.

Don't let perfection be the enemy of good enough. 

73,

Terry     

-- 
Dr. Terry Bullett          WØASP 
NOAA National Centers for Environmental Information (NOAA/NCEI)
Cooperative Institute for Research in Environmental Sciences (CIRES)
Terry....@noaa.gov    720-446-9775 (google voice)  978-337-9092 (cell)   
"Life is Complex.  It has a Real part and an Imaginary part." 

[Sverre Holm]

Some years ago I did some experiments with the Neo-7 GPS module and its 10 MHz output, but as I wrote in the final blogpost of 4-5 on this: "More spurs than I had hoped for": https://la3za.blogspot.com/2019/05/more-spurs-than-i-had-hoped-for.html

73

Sverre 

LA3ZA

[Mike S]

This is a working prototype.

Si5351a module now has a TCXO mounted on the bottom, with the shield soldered to the scratched ground plane foil. The original 25MHz crystal is removed. The TCXO is from eBay, search "Saronix 24MHz TCXO". Seems to be some surplus stock that made its way to sellers. I bought a lot of 10 for 9.99. It is very stable, works very well in this setup. I don't have good test equipment at home, but I can tell the tracking error stays within 2..5ppb, better when the box is closed.

The GPS module is NEO-7M with external active antenna. It is not programmed in any way. The default PPS pulse is 100ms wide, 1Hz rate. I had to remove C2, which connects the builtin antenna. Otherwise it is blinded by Si5351a signal harmonics. Took me some time to figure this out.

The Arduino sketch (attached) is based on F2DC work, with a lot of simplification. The main difference is the measurement interval (1s instead of 40s) and the lower loop gain. Both can be adjusted for better tracking. The servo loop is a simple integrator.

2 outputs are programmed to 14.999 and 14.669MHz. If the second one is set to 9.999MHz, there are combination spurs spaced at 1kHz all over the place.

I hoped to use 1 GPSDO for 2 receivers, but will probably just make more of these. The BOM is under $20.

I have the receiver already. Now figuring out the software. I don't have Raspberry Pi's, and the current price/availability situation is still not good.

How can I get just the modified fldigi, without the RPi image?

73, Mike AF7KR

[Kerry Case]

That’s pretty cool

K

Sent from my iPad

On Oct 2, 2023, at 4:02 PM, Mike S <3ym...@gmail.com> wrote:



Please follow the HamSCI Community Participation Guidelines at http://hamsci.org/hamsci-community-participation-guidelines.

---
You received this message because you are subscribed to the Google Groups "HamSCI" group.

To unsubscribe from this group and stop receiving emails from it, send an email to hamsci+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hamsci/2c15cdd1-69e9-4373-b2a4-6f11f37b9849n%40googlegroups.com.

<gpsdo_fixed.ino>

<gpsdo1.jpg>

[Dave Hinerman]

I have the receiver already. Now figuring out the software. I don't have Raspberry Pi's, and the current price/availability situation is still not good.

 

If it helps, Canakit has Raspberry Pi boards for $35 (1 GB RAM) to $75 (8 GB). If I recall correctly, somebody here told me the custom FLDigi system ran on a 1 GB Pi 3. I haven't tried it though. 

[Mike S]

After some painful learning, I switched to using Erik PD0EK sketch, with minor typo fixes and changes to get it to output 9999kHz rather than 10000kHz.