[time-nuts] EFC info on Trimble 34310-T OXCO

[Charles Steinmetz]

Ed wrote:

>I agree with your statement regarding the determination of the 
>optimum time constant, but, as Bob Camp mentioned, temperature 
>change has a significant impact on setting the value.  My 'lab' is a 
>non-airconditioned bedroom.  My Tbolt doesn't have any active 
>temperature control.  If I set the time constant to the point that 
>Lady Heather thinks is optimum, I see large swings in PPS offset 
>when I open the window and the temperature changes by a few degrees 
>C.  If I leave the time constant at the default of 100 seconds, the 
>swimgs are drastically reduced.  Active temperature control is on my 
>'round tuit' list.

Bert wrote:

>As to Ed's and Bob's comments our projects are not able to compete 
>with commercial products and I do not think that should be our 
>goals. Having spend extensive time on temperature control, I limit 
>my self to 10 C and use fans on all Rb's and passive on OCXO's. 
>Concern about vibration induced noise on the  OCXO made me remove 
>the fan on the tbolt. Added a lot of mass and now ordered some foam 
>balls from China to fill the enclosure as some one recommended.

Well, yeah, it goes without saying (or at least I thought it would) 
that one must keep the rate of change of temperature of the OCXO low 
enough that its oven can keep the crystal temperature within design 
bounds at all times.  I just assume that any time nut would do this, 
since it is extremely simple and costs next to nothing (look in the 
archives for my previous posts about "metal boxes," "metal 
enclosures," and "thermal capacitance" in connection with 
OCXOs).  Active temperature control is NOT necessary.  Which is not 
to say it's a bad idea, it's just not necessary to stabilize any OCXO 
worth owning by a time nut.  (I'm not sure the MV-89 qualifies, even 
if you are lucky enough to get a good one.  There has been some 
discussion on this list about the temperature control loop being 
quasi-stable and tending to oscillate or even latch under some conditions.)

I also see no reason why amateur efforts cannot surpass the 
performance of commercial products, particularly if we assume that 
the environmental conditions are limited to those encountered in 
living space, not a radio shelter exposed to the elements at a remote 
tower.  That is why I've been critical of designs that aim only to do 
"the best that can be done for $5," or "the best that can be done 
with a small ARM and 3 transistors."  Given good design, there is no 
reason why an inexpensive DIY GPSDO shouldn't handily outperform a 
Thunderbolt (using the same OCXO), with two conditions: (i) 
environmental conditions are limited to those encountered in living 
space, and (ii) performance during holdover is neglected.

The reasons why most DIY designs do not work as well as commercial 
designs, even if they use OCXOs of equal quality, is that their 
designers evidently cannot design ADPLLs of sufficient performance to 
do justice to the OCXO.  (This includes implementing whatever means 
of phase comparison and sampling are chosen, the DSP loop filter, 
sawtooth correction, and the NCO or DAC/EFC design.)  Doing all of 
this right isn't particularly expensive, it just takes a designer who 
has the skills and is willing to devote the effort.  As a mentor once 
told me, "Good thinking isn't any more expensive than bad thinking."

Some of the performance gain would be in reducing the rate of 
temperature change seen by the OCXO, either passively as I have 
advocated and described before, or actively.  The other main 
improvement would be setting the PLL crossover out where it belongs, 
which becomes possible when the rate of change of temperature is 
controlled.  Avoiding a few common mistakes would provide some 
additional performance gains.

While the foam peanuts, which I mentioned in a previous post, are 
helpful in some circumstances, I have never seen the need for them in 
the case of an OCXO inside a cast aluminum box.  In that post, I 
mentioned my gut feeling that spheres (balls) likely pack too tightly 
to allow sufficient air circulation.  I think irregularly-shaped 
pieces of foam (like packing peanuts), which leave much more air 
space between them, are required.  The intent is NOT to impede air 
flow, but to randomize it.

One point that I think gets lost in many of these discussions:  The 
quality of individual OCXOs, even of the same model, varies rather 
widely, and you often won't know how good a particular OCXO is until 
you have run it continuously for at least 90 days (preferably 180 or 
more).  The job of any GPS discipline is to gently keep the OCXO on 
frequency, without lowering its xDEV performance at tau where the 
OCXO is better than GPS.

The most effective thing you can do to construct a very stable GPSDO 
is to start with a very stable OCXO.

Often, this means buying a bunch of OCXOs (even if you have to do it 
one at a time for budgetary reasons), selecting the best one(s), and 
moving the rest along.  This can take a long time, since you need to 
run each new oscillator continuously for at least 90 days before you 
can know how stable it is.  The odds of finding a good example are 
improved if you stick to models that, after long experience, 
knowledgeable time researchers have found to be consistently 
good.  Alternatively, you can hope that your sample of the $20 ebay 
wonder of the week will live up to the anecdotal report someone 
posted, but the odds do not ride with you.

Best regards,

Charles