Author Topic: SV1AFN GPSDO  (Read 2032 times)

0 Members and 1 Guest are viewing this topic.

Offline WPXS472

  • Contributor
  • Posts: 28
  • Country: us
SV1AFN GPSDO
« on: January 17, 2019, 02:37:41 pm »
I ordered one of these as soon as he announced it. Then, there was a rather long wait. I emailed him about it and he said there had been a redesign that slowed things down. I actually received it last fall. Naturally, the first thing I wanted to do was to compare it to my trusty Thunderbolt that had been on continuously for months. It has an SMA connector for power. It works, but I don't like it. SMAs are for RF, and I just don't like the whole idea of using one for power. But then, I am an old man, so go figure. Anyway, got the thing connected to power, hooked a GPS antenna and compared it to the Tbolt. Lots of sawtooth error. If you don't know what that is, Google it. I attribute that to the cheap uBlox GPS module. I have a couple, and when I set their output to 10 MHz and compare it to any decent GPSDO, I see lots of sawtooth error. All the regular, purpose made GPSDOs I have fooled with, and there have been several, have some way with dealing with sawtooth error. I suspect heavy averaging, though from what I have read about the subject, that isn't the proper way to deal with it. The software used to program the frequencies it outputs looks like a pain to use, so I haven't done that yet. In my disappointment at the performance, I put it away. I wanted to use it as a clock source for an SDR, and in truth, if I can get it to put out 24 MHz, it should work for that. For the price, you can find a decent, used GPSDO on eBay that is much more stable. But, if you need something other than 10 MHz, this isn't a bad deal. The PC board inside looks to be of good quality, and the board is pretty clean looking. There are several sets of jumpers inside, but there isn't anything explaining what they do. If you don't mind the difficulty of changing frequencies, I can see this as a pretty handy way of getting GPS referenced frequencies. Makis seems to be an ethical guy, and he has a lot of interesting stuff for sale. And regarding that sawtooth error, I should have thought about exactly what this is a while before ordering one. I can't really blame my lack of due diligence on Makis.
 
The following users thanked this post: 9aplus

Offline Astralix

  • Contributor
  • Posts: 16
  • Country: de
Re: SV1AFN GPSDO
« Reply #1 on: January 25, 2019, 08:26:16 am »
Hi!
I got one myself and had it sleeping on the shelf for a long time, blinded by the idea that my UCCM / Trimble module is working. Finding out, that the UCCM is not doing it's job, I picked the SV1AFN module from the drawer and thought about giving it it's chance.

Technically I am at the beginning of understanding time in the way of metrology. So from what I understood the uBlox modules have free running oscillators that are then synchronized to the received GPS signal. The granularity of the oscillators is around +/- 10ns. So this are rather big steps compared to a signal that has 100ns of period, as 10MHz has. If a module carries a crystal or a TCXO does not change the fact, that these are free running and can only be controlled in these 10ns steps.

The modules specialized in time that have the -T suffix, are not different in the handling of the XTAL or TCXO, but they can be preset to a fixed position and so they can work precisely with only one satellite in sight, while all other modules need a minimum of 3.

Finally there is the LEA-M8F module that synchronizes the internal oscillator to the GNSS signals and so can keep this frequency stable without any granularity. Unfortunately for us, this module gives a 30.72MHz signal.

As a result, there is no way around a second slow reacting 10MHz source that is carefully and slowly pushed to keep the frequency stable over long time.

For the SMA power connector... I soldered a bridge from the USB connector to the SMA and use a USB charger to power the unit. The SV1AFN module uses a uBlox NEO-7M that internally uses a free running crystal. But there is a second external TCXO that provides basic stability, but it is not temperature controlled in the way that it's temperature is varied by the controller, it is simply put to a single stable frequency. The correction of the frequency itself is made in the SI5342A PLL chip, by combining the 50MHz clock and reference it to a 10MHz clock coming from the re-programmed 1PPS output of the uBlox. So as the 10MHz of the uBlox module are corrected in rather large steps, the PLL reference is stepped in the same way.

However, I still think, that the SV1AFN unit is a good unit for any amateur radio lab...
The stepping is in a range of 0.1Hz at 10MHz and multiplies to ~4Hz if you adjust a 70cm radio. Having already repaired several transceivers, I never came across any service manual, that required more then +/-20Hz at any setting. I know there are units that require more precision but I repair only the transistor units that are rejected by all professional repair shops.

So for anyone who likes to stay in tolerance, this SV1AFN unit should be an option. The UCCM units are more stable and sophisticated, but you have to hack them, sometimes need to buy two or three of them until you get a working one. And you have to remember that the SV1AFN unit is brand new, while the UCCM units had a long, sometimes rough life already.
 

Offline LapTop006

  • Supporter
  • ****
  • Posts: 460
  • Country: au
Re: SV1AFN GPSDO
« Reply #2 on: January 27, 2019, 10:26:26 am »
Finally there is the LEA-M8F module that synchronizes the internal oscillator to the GNSS signals and so can keep this frequency stable without any granularity. Unfortunately for us, this module gives a 30.72MHz signal.

Only if you use the internal oscillator, it also has provisions for an external disciplined oscillator.
 
The following users thanked this post: Astralix

Offline WPXS472

  • Contributor
  • Posts: 28
  • Country: us
Re: SV1AFN GPSDO
« Reply #3 on: January 29, 2019, 08:54:07 pm »
Astralix, thanks for the comments. I suppose I have become spoiled by having some rally good GPSDOs. I am now wanting to get my hands on a uBlox ZED-F9P. They are a good bit more expensive than any of the other uBlox modules I have had.
I am anxious to see if they can produce the position accuracy they claim. Here, I don't have a reference I can trust, however. I am thinking of just seeing how much their reported position varies over time. If the ZED-F9P is really as good as they claim, it should be useful in testing antennas. I want to see if a choke ring is actually better. I have a Leica choke ring antenna, and it is a patch, just like my Trimble survey grade antenna as well as a Chinese made one I got from Alibaba. You are quite correct that the SV1AFN GPSDO should be suitable for most amateur uses. I haven't tried to change frequencies on mine yet, but it looks complicated. If that proves not to be the case, I suspect I will find a lot of uses for it.
 

Offline eb4fbz

  • Regular Contributor
  • *
  • Posts: 122
  • Country: es
Re: SV1AFN GPSDO
« Reply #4 on: January 31, 2019, 10:09:40 am »
 

Online texaspyro

  • Super Contributor
  • ***
  • Posts: 1395
Re: SV1AFN GPSDO
« Reply #5 on: February 09, 2019, 04:10:23 am »
I am anxious to see if they can produce the position accuracy they claim. Here, I don't have a reference I can trust, however. I am thinking of just seeing how much their reported position varies over time. If the ZED-F9P is really as good as they claim, it should be useful in testing antennas.

I got in one from Sparkfun last week (they are now currently sold out).   I am currently adding support for the F9 to Lady Heather.   I sent a 16 hour RINEX v3.03 file to CSRS-PPP and they returned a position with reported lat/lon/alt error ellipses of 5/10/22 mm.  (my antenna has a very poor location).

In 3D navigation mode the positioning data tends to wander around +/- 1 meter.

Speking of antennas... to take advantage of the F9 you will need an antenna that supports the L1 and L2 bands for GPS, GLONASS, GALILEO, and BEIDOU.   I am using a $100 antenna from China.  It seems to work very well.

https://www.ebay.com/itm/CORS-RTK-GNSS-Survey-Antenna-high-gain-measurement-GNSS-GPS-GLONASS-BDS/253786590956?hash=item3b16dc0eec:g:p8sAAOSwu7hbNoe0:rk:5:pf:0

Search Ebay for CORS antenna for more sellers.

For precise positioning post processing with CSRS-PPP you only need GPS and GLONASS... they don't yet handle GALILEO and BEIDOU data.

I going to try and characterize the 1PPS and 10 MHz outputs this weekend.  The reported sawtooth error spans around +/- 3.5 nsec.
« Last Edit: February 09, 2019, 04:17:17 am by texaspyro »
 

Offline WPXS472

  • Contributor
  • Posts: 28
  • Country: us
Re: SV1AFN GPSDO
« Reply #6 on: May 20, 2020, 12:50:32 pm »
I sort of forgot about this. Sorry. I ended up getting an F9T module. When I compare its 10 MHz output to either a Thunderbolt, or RFTG, it doesn't seem to have sawtooth error, but instead kind of wanders around a good bit. I have noticed 2 GPSDOs doing this, but the F9T seems to be wandering much more rapidly. I tried it using all the different nationalities, but I think it works better just using GPS. I have an L1/L2 antenna, but I can't tell that it uses anything but L1. When I look at the position plot in the uBlox software, it is wandering around quite a bit. Perhaps my antenna location isn't as good as I thought it was. Is there a way to set mask angle? I don't recall seeing that. That SV1AFN unit is still on the shelf.
 

Offline pbreed

  • Newbie
  • Posts: 1
  • Country: us
Re: SV1AFN GPSDO
« Reply #7 on: June 22, 2020, 04:11:45 pm »
Yes the ublox 1PPS has clock granularity issues, but the offset of this value is available over the serial interface via the UBX-TIM-TP message.
This message gives you the picoseconds offset of the time pulse.
Note that the ublox sends the TIM-TP message for the NEXT time pulse.
So if you want a software disciplined oscillator:
Set up to measure the offset between the disciplined oscillator and the PPS, read the serial port message, adjust and operate a high resolution DAC
feeding the TCXO frequency adjust pin.

If you want to precisely measure the clock offset of a single time pulse this probably requires a flip flop, an XOR gate and sample and hold...
Or one can just count longer periods.. IE 10 seconds, 100 seconds...

Even with just a straight counter and no phase measurement:

   Measurement interval in Seconds         
Ref Clk   1   10   100   1000
10 Mhz   100   10   1   0.1
100Mhz   10   1   0.1   0.01
            
Measurement error in ppb            

So a 1000 second measurement of a 100Mhz reference is 10E-11 or in the realm of Rubidium standards.












 
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf