[time-nuts] Soundcard sampling Re: Picking a good HP 10811

[Bruce Griffiths]

Lux, James P wrote:
>   
>> Real AD cards with adequate performance are usually far more
>> expensive than $300.
>> In principle, one could build an ADC card with adequate
>> performance using 4 AD7760 ADCs.
>>
>> Bruce
>>
>> ____
>>     
>
> The AD7760 eval board is $150, and I think you need another board to hook it up to a computer.  However, if you have a couple kilobuck budget (i.e. you're doing this for work), then stacking up 4 of these eval boards might be a decent way to get started.
>
> The AD7760s themselves are $36 each (Qty:1000).. The eval board is 4 layers, so that's within the realm of inexpensive prototyping, if you were to lay out a board.  Scrounging, etc., you could probably put something together for a few hundred bucks with pretty decent performance.
>
> If it were me, though, I'd look at the high quality pro widgets already assembled, and spend my precious hours writing software to do the data reduction from the captured WAV files.
>
> ( I note that the DMTD box they have here at JPL (Greenhall, Kirk, and Tjoelker) is reported as having a noise floor Allan dev of <1E-17 at 1000s, driving the commercial synthesizer offset generator and two inputs with the same H-maser.. And it is a straight counter style measurement (123Hz))
>
>
>   
The mixer RF input frequency is 100MHz for that system.
The beat frequency slope amplifier/limiters in previous incarnations had
a performance some way from that of a more optimum design.
Achieving an equivalent performance with 10MHz inputs would be somewhat
challenging.
I wouldn't expect to do quite as well with a sound card based system
perhaps an order of magnitude or so worse (at least for Tau< 100s).
When comparing such systems it is necessary to know the system noise
bandwidth.

Bruce