[time-nuts] Near-perfect chip for Loran-C frequency receiver

[Bruce Griffiths]

Poul-Henning Kamp wrote:
> In message <1215126999.6104.9.camel at bigdog.icmp.com>, Carl Walker writes:
>   
>> On Fri, 2008-07-04 at 01:02 +0200, Magnus Danielson wrote:
>>     
>
>   
>> There's more than meets the eye initially when you attempt a receiver
>> design of this type - at least as far as the analog section goes.
>>     
>
> I hate to say so, but actual experience proces you wrong on all counts.
>
> Remember we are not talking about fast signals like RTTY or voice,
> if we were, you would be right.
>
> We are talking about measuring one or more frequency/phase signals
> and with an OCXO or Rubidium timebase we have minutes, hours or
> even days of integration time at our disposal.
>
> Currently, my AGC is a manual trimmer and I have not touched it
> for years.  I used to have a 246kHz LW station in visual range
> which set the level for me, and that gave me no trouble at all.
>
> Besides, a loop antenna can null out one strong signal very
> efficiently.
>
> A 12bit A/D gives about 72 dB dynamic range to work with, and and
> as I said, I can detect Loran-C signals as far away as 8830 and
> 7270.
>
> LORAN-C is both spread-spectrum and spread-time, those two signals
> have amplitudes below one bit, but once they're despread, they
> stand about 1 bit taller than the remaining noise.
>
> The entire point about software defined radio is to do away with
> the complicated analog side, and do it in the CPU, and VLF
> time/frequency signals are about the easiest signals you can
> work with for this.
>
> See my earlier work here:
>
> 	http://phk.freebsd.dk/loran-c
>
> Poul-Henning
>
>   
You've misunderstood the problem.
Extrapolating your limited experience ( Loran) to the more difficult 
problem of simultaneously detecting/tracking Loran signals and 
potentially much weaker AM modulated time signals using the same 
wideband receiver with minimal analog filtering is asking for trouble. 
Try detecting decoding a 1uV/m signal in the presence of a strong 1V/m 
signal within the passband and see just how difficult it is without 
suitable analog filters etc.
Not every potential user is conveniently located relatively close to 
DCF77, MSF, WWVB etc.

Just because an ADC with a dynamic range of 72dB is adequate when 
processing Loran signals doesn't mean that it is adequate when the weak 
signal one is attempting to track/decode may be 120dB or below a 
stronger signal within the wide receiver passband.

72dB dynamic range is entirely inadequate if one has a wideband receiver 
located close to a transmitter whilst simultaneously attempting to 
detect/decode a very weak signal.
The solution (other than using an unobtainable high sampling rate ADC 
with 120dB or more dynamic range) is off course to not use a wideband 
receiver.

If one wishes to decode the AM time code broadcast by JJY WWVB, MSF, MST 
etc then the maximum signal averaging time is relatively short.

It is perhaps simpler and easier to split the problem into 2 parts:

1) detecting + decoding Loran signals

2) detecting + decoding potentially weak AM time code broadcasts and 
simultaneously locking to the carrier.

ADCs are relatively noisy devices with limited linearity and dynamic range.
Some analog filtering and gain is required before the ADC when 
processing weak signals.



Bruce