[time-nuts] Cheap DMTD for Rb / OCXO locking (was: True Position GPSDP + Rb X72)

[Bob kb8tq]

Hi

> On Sep 22, 2017, at 3:01 PM, Attila Kinali <attila at kinali.ch> wrote:
> 
> On Fri, 22 Sep 2017 11:50:29 -0400
> Bob kb8tq <kb8tq at n1k.org> wrote:
> 
>> Given the noise levels of typical Rb’s - use XOR gates as the mixers. 
> 
> Right. That makes it a lot more robust than using a diode mixer.
> 
>> If you do the filters as R/C’s they are cheap. If you just re-clock, you
>> maybe add a flip flop or two to each channel. Total BOM for the DMTD
>> part of it would be < $1 not counting the OCXO. 
> 
> Recklocking shouldn't be necessary. A simple inverter should raise
> the signal back to an approriate signal level. After that a division
> chain is still needed as otherwise the rate of transitions will be
> too high for the uC to handle.

The output of the XOR will be at F1+ F2 and at F1-F2. The one you want is the
F1-F2 component. You need to reject the F1+F2 somehow. One would be to 
go with a D-FF rather than an XOR in the first place, you still get flaky transition
zones. Next is a filter that rejects the (say) 22MHz and passes the (say) 1 MHz. 
The final one would be some sort of flip flop or two that sub samples (at 10 MHz 
maybe) and thus rejects the 22 MHz. You still have the flakey transition issue. 

> 
> Probably the cheapest way to do this would be to use a 38.4MHz or 38.88MHz
> oscillator (available for 1-2$) and using a 74LVC74 to divide it by 4.
> That gives 400kHz and 280kHz offset frequency, respectively. Dividing
> both beat signals by 256 using two 74LVC161/163 each should get it
> down enough to be workable with a modern uC.

The hope is that you are dealing with something below 1MHz so heroic efforts
are not needed. An L/C filter is probably cheaper than a bunch of dividers and 
this and that.


> 
>> All of the previous comments about “why would you want to do this?” 
>> still apply. Most of the small Rb’s we can get our hands on have really 
>> awful phase noise and spurs. That’s not true of all small Rb’s, but buying
>> them new in small volume at $2K each is not something most of us 
>> are interested in doing ….
> 
> Yeah..
> 
> If anything, I'd do it the other way. Discipline a Rb to GPS using
> a long time constant. Then slave an OCXO onto that using the method above
> and use a loop time constant in the order of 1-100s


And that’s really the point. You now have a GPS -> OCXO -> Rb -> OCXO combo. 
You *could* tune it up to do some cute stuff. Tuning it will not be at all easy. It’s 
going to take good test methods and a *lot* of time. Is that many years or only 
many months of test? Who knows …. If you had to pay somebody to do it all, I’d bet
that buying a Maser would be cheaper than doing this as a one off. Buying a $2K 
“less noisy” Rb would be a slam dunk in that case as well.

This does get back to “what do you want to do? If this is just a reference that is going 
into a 5334 counter and not much else, the stock GPSDO will do that. The stock Rb
with monthly cal will do that. If you want to do direct multiplication to 40 GHz for 
your  3 Hz wide PSK rig … none of it measures up. 

Bob

> 
> 
> 
> 			Attila Kinali
> -- 
> It is upon moral qualities that a society is ultimately founded. All 
> the prosperity and technological sophistication in the world is of no 
> use without that foundation.
>                 -- Miss Matheson, The Diamond Age, Neil Stephenson
> _______________________________________________
> 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.