[time-nuts] WWVB SDR discussion

[Bob kb8tq]

Hi

The problem with the crystal is that it has a temperature coefficient. As a 
narrow band filter, it will have a *lot* of delay. Crystal resonance moves 
(with temperature) and the delay changes. 

How much delay depends a lot on a bunch of fiddly details. A 10 to 100 Hz wide
bandpass could easily have delay in the > 100 ms range. 

How much change?

The crystal could be in the 10’s of ppm / C range (might be lower, could be higher).
At 60K Hz 10 ppm is 0.6 Hz. With a modest sort of basement temperature swing
you *could* get 10% of the delay changing around. 

Yes, that’s a bunch of guesses.

Net result would be delay variation in the 1 to 10 ms range. 

Bob

> On Aug 11, 2020, at 3:37 PM, paul swed <paulswedb at gmail.com> wrote:
> 
> Mark
> His antenna hit a preamp as I recall about 20 db of gain. To see something
> on a scope add 40 more db approx. Unfortunately a purely broadband solution
> will show 40 db of pure garbage these days. Using the 60 KHz watch crystals
> $2.00 for 20 out of China you can most likely find a reasonable match.
> Thats what I did. It is hi Z so it feeds one side of an opamp. Look at the
> spectracom schematics to get a sense of what to do. I made a small socket
> to plug them in and found the one that worked. As an alternative you can
> build a bandpass filter with opamps lots of variations. Anything to get the
> received bandwidth reduced. Look at Johns front end also.
> Regards
> Paul
> 
> On Tue, Aug 11, 2020 at 3:20 PM Mark Haun <mark at hau.nz> wrote:
> 
>> Hi Detlef,
>> 
>> On 11-Aug-20 3:46 AM, Detlef Schuecker via time-nuts wrote:
>>> you do not necessarily need a variable ( physical ) oscillator. You mix
>>> the signal down in the digital domain. A 'digital local oscilator' is a
>>> mere complex value, which is rotated and the power is adjusted:
>> My proposed block diagram does actually have a digital LO, only mine is
>> 1, 0, -1, 0... (in-phase) and 0, -1, 0, 1...  (quadrature).  You could
>> of course use an variable-frequency NCO, but I need a physical
>> oscillator in any case to clock the MCU.  I am also thinking in terms of
>> a WWVB-DO where I want a stable local reference to steer.  (Although in
>> fairness, for WWVB I think you probably want stability over the diurnal
>> propagation variation, and my crappy OCXO has no chance at that.)
>>>> part, unfortunately.  My tuned loop seems still too broadband, even
>>>> after a couple more poles of op-amp filter.  I have a bunch of 60-kHz
>>>> tuning-fork crystals and wanted to try a crystal filter like the "pros"
>>> Good point.
>>> Firstly I tried a tuned resonant LC circuit with a BF245 preamplifier to
>>> keep a high Q.
>> Besides having minimal analog design experience, I think what is
>> confusing me is these crystals have a really high impedance far away
>> from their parallel resonance.  Even at series resonance the motional
>> resistance is in the 10s of kohms, IIRC.  So I'm not exactly sure how to
>> deploy one in place of an LC tank circuit, e.g. in a collector or drain
>> circuit.
>>> Secondly an accurate ADC is an option. With 24Bit you get more than
>> 100dB
>>> dynamic range, so you dont care about a 60dB stronger nearby interferer.
>> 
>> Fancier ADC shouldn't make any difference.  Even though the STM32L4 ADC
>> has an SNR of (IIRC) ~ 69 dB, you can sample at several MHz.  After
>> downsampling to something like 100 Hz (post LO), the SNR is well over
>> 100 dB, which should be plenty.
>> 
>> (Of course, this presupposes that there are not strong interferers or
>> SMPS noise spurs, etc. within a 100-Hz BW centered on 60 kHz.  But if
>> there are, a fancy ADC wouldn't help you anyway.  The main thing is to
>> make sure the interference isn't causing any clipping or nonlinearity
>> before you sample.)
>> 
>> I should note that my big tuned air loop and preamp, which is modeled
>> after Joe Magliacane's, may be adequate even though I can't see WWVB on
>> the oscilloscope.  I was hoping to build confidence there before hooking
>> it up to the MCU and doing SDR stuff.  I also want a small loopstick
>> version of this so I can embed it in my own nixie wall clock, hence the
>> interest in crystal filters, etc.
>> 
>> Mark
>> 
>> 
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com
>> To unsubscribe, go to
>> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
>> and follow the instructions there.
>> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.