[time-nuts] Quick check of a GPS controller oscillator

[Giuseppe Marullo]

Sorry, images where too big, had to repost.

Please use this link

https://www.iw2jww.it/trimble-and-progrock/

to download them, thanks.

On 5/6/2020 8:57 PM, Giuseppe Marullo wrote:
> Ok,
>
> first of all many thanks for your  help. Yep, not exactly a time-nut 
> piece of engineering, but I need more to be reassured about the 
> Trimble accuracy.
> If the Trimble is good enough, everything will be easier.
>
> @Alex:
>
> >in Europe you should be able to get one DCF77 signal based 10MHz, 
> which you could trust
> I am not planning to add a DCF77 synced 10MHz, I would rather add 
> another Trimble or Rubidium.
>
> The whole point is, given my time nuttery is limited, I could be okay 
> with the Trimble provided it is "good enough" to consider it a 
> suitable reference.
>
> @Dana:
>> Is the ProgRock GPS disciplined or a Rb, or is it just a free-running
>> crystal oscillator?
> The ProgRock in itself is a simple Si5351A based programmable 
> oscillator. It may have a OCXO option (that I built and using) and 
> could use a GPS PPS to be calibrated(that I built and using, as the 
> other pieces from QRP labs).
> The GPS does output NMEA sentences but the ProgRock just use the PPS 
> signal. It is placed in a Clock kit enclosure and the clock does 
> use/display the NMEA sentences but the ProgRock just use the PPS signal.
>
>> And does LH say that the Trimble is locked OK?
> I have included a screenshot of LH(20200506LH.png) as per now.
>
>> Unless the ProgRock is GPS disciplined (or is a Rb), it's a pretty 
>> safe bet
>> that it has simply grown old and the crystal has aged.  And this 
>> notion is much strengthened if LH gives
>> a positive indication that your Trimble is locked OK.
> Don't think it could be aging related, it should use the GPS to self 
> calibrate anyway within minutes. It does/should use the GPS PPS data, 
> saying it is disciplined may be a little optimistic.
>
> @Bob:
>> Hi
>>
>> With any GPSDO (not just the TBolt) you can do some basic checks:
>> (They are in no particular order)
>
>> 1) Are there > 6 sats in view at all times, if not, move the antenna
> Yes they are. Antenna is Simmetricom missile head 28dB amplified 
> (timing antenna) clear view of the sky, on a 2m pole on a balcony 
> facing South.
>
>> 2) Is the unit locked to at least 4 sats at all times, if not, move 
>> the antenna.
> Yes, see LH attachment
>
>> 3) Has the unit completed its survey process? If not, wait for it to 
>> complete.
> Not at the moment I power cycled it this night. It is powered since 
> this night (>12hours).
> Do I have to initiate it?
>
>> 4) Is the EFC > 10% and < 90% of it’s range after an hour on? If not, 
>> repair the unit.
> I don't know what EFC is. If it is the DAC voltage, seems so (looking 
> at its values it never went over the range but I may have not seen it 
> happening)
>
>> 5) Has the unit been on for at least 3 days? If not wait.
> Not yet, waiting...
>
>> 6) Does the output look ok on a ’scope? ( it’s volts not milivolts …) 
>> if not repair the unit
> See attach(20200506_DSO.JPG), it is a vanilla scope nevertheless it 
> seems good to me
>
>> 7) Does the unit show “locked” or something similar? 
> Yes, it seems so. Operation mode: LOCKED / PLL: LOCK
>
>> 8) Is the EFC consistent ( = stable) and not wandering around? If not 
>> repair the unit. 
> Can't say. If EFC is the DAC, well sort of but not exactly stable. 
> Keep in mind I am opening and closing the window now, so there is a 
> change in temperature.
>
>> 9) Do the sat’s show SNR’s above 40 db? If not, replace the antenna / 
>> coax.
> They seem fine, according to the screenshot (C/N values, right?). 
> Cable is 10m RG-8X, Yes it could be better but not shorter. I have 
> some meters of Heliax (.5" or 1+5/8" AKA coke can size) not that time 
> nut yet. They could act as a mast for the antenna, LOL.
>
>> 10) Does it read same / same /same compared to your counter’s 
>> internal time base? 
>  If not, check the counter / repair the GPSDO.
> I would say yes. Counter does have high stability option (04E: Ultra 
> High Frequency Ovened Oscillator <5x10 E-10, not 04R rubidium), so 
> last digits may vary. I have included a photo of what is 
> displayed(20200506_FC.jpg). 10.00000007 almost all the time ending 
> with a 7 once I saw it briefly flickering to an 8.
> I added a second photo with overflowing the FC would squeeze out 
> another digit and it is pretty stable at   [1]0.000000069MHz 
> (20200506_FC_OV.jpg).
>
>> 11) Does the reported ( survey ) location make sense on Google Maps? 
>> If not restart
>> the survey / repair the unit. If the unit passes all those checks, 
>> it’s a really good bet that you can trust the output. It likely is 
>> good to < 1 ppb, almost regardless of who made the device. Not every 
>> check
>> will be possible on every GPSDO. The checks assume you have a ’scope 
>> and a counter.
>> They also assume you can talk to the status port on the device. Some 
>> readings ( EFC range, SNR, may take a bit of research to put into 
>> rational units ….).
> At the moment it is in position holdover. Previous time was able to 
> complete the survey.
> Current Reported position is correct within 5m according to Google Earth.
>
>> ======
>>
>> If all of that sounds like a lot of work (or not complete enough), 
>> buy a second ( or third or fourth ….) GPSDO. Next get a couple Rb’s 
>> …. maybe a Cs or three …. have you looked into Masers?
> I was thinking to play with Aluminum, seems the coolest thing nowadays :)
>
> @Tim:
>> Another way of asking the question to the OP, is he a phase-nut or a
>> frequency-nut?
> I fear phase-nut is EVEN more costly, so I will go for frequency-nut 
> at this time :)
>
> @Bob(take two):
>> Since the ProgRock is not a GPS disciplined unit unless you add this 
>> or that to it, the best bet is that this one is simply the tuned 
>> version. Would the list apply
>> to a gizmo like that? Who knows. A lot depends on just how it’s 
>> cobbled together. 
> What do you mean by tuned version?
> Anyway...I am more questioning the Trimble than the ProgRock. It is 
> clear they are not in the same league, just think that by default 
> ProgRock adjust its frequency only when its 27MHz XTAL is off by 5 or 
> more Hz. The only way to increase the control is to ask the MCU to 
> calibrate every second, so you should see almost every second a 
> frequency change!
> Not exactly a smooth transition (this is the current setting). sadly 
> there is no feedback about corrections, error, lock or whatsover.
>
> but two questions arises:
> 1) is MY specific Trimble accurate enough(I hope Trimble qualify for 
> this ml)? If the answer is a "solid" yes, then the ProgRock is not 
> accurate enough for its claimed performance.
> 2) Why the thing stays always about 2.5Hz below the Trimble? Shouldn't 
> they move in relation to each other? It seems like a coding error, I 
> am tempted to insert 10.000002MHz in the ProgRock to see if it will 
> settle to 9.9999995MHz.
>
> I could try to feed the PPS output of the Trimble to the ProgRock and 
> see if accuracy improves (may not work out of the box, pulse may need 
> to be stretched).
>
> Giusepe Marullo
> CISSP - GCFA - CSSA
>
>