[time-nuts] My GPSDO project: OCXO Thermal Oscillation?

Sat Aug 17 08:53:15 EDT 2013

Hi

If one of your sources can be offset (it's a DDS) then mixing the two is a good way to increase the resolution. Separate the two by a couple Hz and feed them into a double balanced mixer. Run the beat note into an amp and limiter. Output of the limiter drives the 5335. Resolution goes up by 10^7 if you are at 1 Hz. The gotcha is that you can't use all the resolution you gain due to noise. 

The old style approach was to use a pair of OP-37 op amps, the first as a ~ 10X gain amp. The second as an inverting limiter. Simple R/C filters were used both as high pass and low pass on the signal ahead of the limiter. There is a simple / non-critical  L/C filter between the mixer and the first amp. There are a number of other ways to do it. 

A Mini-Circuits RPD-1 makes a pretty good mixer for a simple setup. The whole thing can be done on perf board including the +/- 18V three terminal regulators. If you have a bit of this and that in your junk box, cost should be < $50. That of course assumes you already have a lab supply to drive the regulators...

Bob

On Aug 16, 2013, at 8:42 PM, Bob Stewart <bob at evoria.net> wrote:

> I thought I could just re-enter the numbers from the last survey I did.  But either something is wrong or I'm missing something (betting on the latter).  I can't seem to enable TRAIM.  I thought I had it setup just like previously, but under WinCore12 it simply won't turn it on.  So, I'm just letting it do a survey again, I guess.  IOW, I dunno.
> 
> 
> And I have a question about MATH functions on the 5335A, if you don't mind.  (I'm an HP newbie.)  Is there away to use Ratio A/B and have it give a greater resolution than a whole number, perhaps by having it average over multiple B ticks?  I think I figured out how to enter an Offset, but I just get the ratio in single digits which doesn't really give me anything.
> 
> 
> Bob
> 
> 
> 
> 
>> ________________________________
>> From: Bob Camp <lists at rtty.us>
>> To: Discussion of precise time and frequency measurement <time-nuts at febo.com> 
>> Sent: Friday, August 16, 2013 7:26 PM
>> Subject: Re: [time-nuts] My GPSDO project: OCXO Thermal Oscillation?
>> 
>> 
>> Hi
>> 
>> Also remember - you need to do a survey on the UT+ and put it into position hold / timing mode. If you don't you can add a bit more to your error budget. 
>> 
>> Bob
>> 
>> On Aug 16, 2013, at 7:58 PM, Bob Stewart <bob at evoria.net> wrote:
>> 
>>> I'll figure out something.  I'm thinking of icing down my DDS device to put it at a known temperature so I can do drift comparisons.  If not that, there'll be something else I can try.
>>> 
>>> On another note: I thought I had destroyed everything when I swapped the UT+ back into the GPSDO.  Fortunately it was just that the UT+ needed to be reset, and I had destroyed a USB-TTL adapter and not my new little Adafruit.  I'm going to have to figure out exactly what command sequence needs to be sent to the UT+ to get the comms working again and write a short program to do it.  For some reason the WinCore12 program wasn't able to bring it up.
>>> 
>>> Bob
>>> 
>>> 
>>> 
>>> 
>>> 
>>>> ________________________________
>>>> From: Angus <not.again at btinternet.com>
>>>> To: Discussion of precise time and frequency measurement <time-nuts at febo.com> 
>>>> Sent: Friday, August 16, 2013 6:14 PM
>>>> Subject: Re: [time-nuts] My GPSDO project: OCXO Thermal Oscillation?
>>>> 
>>>> 
>>>> Hi,
>>>> 
>>>> To be able to test GPSDO's (and GPS in general) was one of the main reasons that I made a temp controlled chassis for an LPRO and gave it air pressure compensation. That was good for tau's of hundreds to tens of thousands of seconds, and even longer with drift compensation. Add in a clean-up oscillator if desired, and you have a pretty good reference.
>>>> 
>>>> Incidentally, I tested 2 FEI5680A's, a Temex LPFRS and 3 LPRO's in this type of setup, but only the LPRO's allowed the air pressure effects to be almost completely cancelled out. For some reason the others did not react to fluctuating air pressure as predictably.
>>>> 
>>>> Angus.
>>>> 
>>>> From: "Bob Camp" 
>>>> To: "Discussion of precise time and frequency measurement" 
>>>> Sent: August 16, 2013 7:45 PM
>>>> Subject: Re: [time-nuts] My GPSDO project: OCXO Thermal Oscillation?
>>>> 
>>>> Hi
>>>> 
>>>> With the 5335 you have a measurement with dead time. That makes things a bit hard to figure out. A much better way to go is to feed a pair of 1 pps signals into the 5335 and measure their time difference. Unless they are quite close, you can go for a while with no ambiguity to the reading. The effective resolution increases linearly with the time length of the observation. There also are a number of very nice programs that will let you collect the data from the 5335 via GPIB.
>>>> 
>>>> Assuming your 5335 works like mine does it's got about a 1 ns resolution at 1 second. It'll give you 1 ppb at a 1 second gate and 1 ppt at a 1,000 second gate. By the time it gets to 1,000 seconds the internal counters have overflowed and the reading is a bit messed up. 
>>>> 
>>>> Without some sort of accurate reference, there's really no way to know for sure what's going on with a GPSDO. One solution is to build two or three of them and watch them fight with each other. Another solution is to pick up a Hydrogen Maser. It's always a "what's in your wallet" sort of decision.
>>>> 
>>>> Bob
>>>> 
>>>> On Aug 16, 2013, at 2:26 PM, Bob Stewart  wrote:
>>>> 
>>>>> Hi again Bob,
>>>>> 
>>>>> D'oh, I think I totally misunderstood your figures in my first response.  The .16ppb is not the frequency accuracy of my GPSDO.  It's the amount that I'm moving the OCXO during a 5 minute timeframe, which is something else entirely.  Like I said I do not have a known good oscillator to compare to.  However, I have a DDS oscillator I made some time ago, and it seems to be pretty stable if I let it be.  So, what I've done is to hook the GPSDO to the clock input of my 5335A.  I've then adjusted the DDS so that it reads near 10.000000 MHz, and watched it over a round-trip 5 minute period several times with a large enough gate that I get 8 decimal points on the counter.  I don't see any relationship between the few milli-Hz movement the counter shows and the changes to the DAC.  During several runs last night, I saw less than 30 mHz of movement, which, if true, would be 3E-9, or 3ppb, right?  Or would that be +/- 1.5ppb?
>>>>> 
>>>>> Bob
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>>> ________________________________
>>>>>> From: Bob Camp 
>>>>>> To: Discussion of precise time and frequency measurement  
>>>>>> Sent: Friday, August 16, 2013 10:47 AM
>>>>>> Subject: Re: [time-nuts] My GPSDO project: OCXO Thermal Oscillation?
>>>>>> 
>>>>>> 
>>>>>> Hi
>>>>>> 
>>>>>> Ok, let's try some math and see if I can do it without blinking this time….
>>>>>> 
>>>>>> +/-4 Hz for 6 volts is 0.66 Hz / V
>>>>>> output is 10 MHz so 1 Hz is 0.1 ppm
>>>>>> your OCXO is running at 0.066 ppm / V 
>>>>>> That's also 66 ppb / V
>>>>>> 
>>>>>> 0.02 V at 66 ppb / V is 0.0132 ppb or 13.2 ppt
>>>>>> 
>>>>>> The UT+ has a sawtooth output that's about 45 ns
>>>>>> That's 45 ppb at one second
>>>>>> 
>>>>>> 5 minutes is 300 seconds
>>>>>> 
>>>>>> so 45 / 300 = 0.15 ppb or 150 ppt
>>>>>> 
>>>>>> If it's the later clone version it might be  about 1/2 of that. 
>>>>>> 
>>>>>> Are you doing sawtooth correction?
>>>>>> 
>>>>>> Bob
>>>>>> 
>>>>>> On Aug 16, 2013, at 11:09 AM, Bob Stewart  wrote:
>>>>>> 
>>>>>>> I'm converting the code for the VE2ZAZ FLL to a PLL.  I'm seeing the phase correction change the EFC up and down about .02V to .03V over a period of 5 minutes or so (it varies).  The full range on the OCXO is about +/- 4Hz varied by 0 to +6V, so at least this is a tiny value.  I feel pretty confident with my code at this point.  I'm using a Trimble 34310-T OCXO for which I've been able to find almost no information.  Could this oscillating phase correction be some sort of thermal oscillation?  I've tried two separate 34310s and both act more or less the same.  My GPS device is normally a UT+, but I just now swapped in an "Adafruit Ultimate GPS Breakout" to the same effect.  Is this good, bad, or indifferent for a GPSDO?  I started this project not knowing what to expect, and I still don't.  Experienced help, speculation, or even just kind words at this point would be appreciated!  =)  I don't have a known good/stable reference to
> compare
>>>>> this
>>>>>>> to.
>>>>>>> 
>>>>>>> Bob - AE6RV
>>>>>>> _______________________________________________
>>>>>>> time-nuts mailing list -- time-nuts at febo.com
>>>>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>>>>>> and follow the instructions there.
>>>>>> 
>>>>>> _______________________________________________
>>>>>> time-nuts mailing list -- time-nuts at febo.com
>>>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>>>>> and follow the instructions there.
>>>>>> 
>>>>>> 
>>>>>> 
>>>>> _______________________________________________
>>>>> time-nuts mailing list -- time-nuts at febo.com
>>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>>>> and follow the instructions there.
>>>> 
>>>> _______________________________________________
>>>> time-nuts mailing list -- time-nuts at febo.com
>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>>> and follow the instructions there.
>>>> _______________________________________________
>>>> time-nuts mailing list -- time-nuts at febo.com
>>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>>> and follow the instructions there.
>>>> 
>>>> 
>>> _______________________________________________
>>> time-nuts mailing list -- time-nuts at febo.com
>>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>>> and follow the instructions there.
>> 
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at febo.com
>> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>> and follow the instructions there.
>> 
>> 
>> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.