[Elecraft] K4: superhet vs. direct sampling

[W2xj]

Direct sampling has no image issues.

Sent from my iPad

> On Jun 5, 2019, at 3:23 AM, Richard Corfield <richard.corfield at gmail.com> wrote:
> 
> I'd have thought that superhet would always have the advantage of improved
> image rejection due to the IF band filtering. It would also allow ADCs and
> associated processing to run slower or lower down in relation to its
> Nyquist frequency so allowing for more detail in the sampled signal. Also
> can narrow band analogue filtering beat the dynamic range of the ADC? If
> its top 10 bits are taken up handling that huge strong signal in its input
> passband then you've got fewer bits left for your signal of interest. So
> both high bit depth and high sample rate in comparison to signal are harder
> to achieve.
> 
> The opposing view being the cost of achieving a high quality Superhet
> conversion? If direct sampling and high speed signal processing (FPGA?) can
> achieve the results so much more cheaply and simply and reliably? Like the
> Class D amplifier in reverse.
> 
> - Richard (M0RJC)
> 
> 
>> On Wed, 5 Jun 2019 at 03:19, Wayne Burdick <n6kr at elecraft.com> wrote:
>> 
>> The superhet module buys a lot of BDR improvement. But also -- a subtlety
>> I've failed to mention so far -- the superhet module is intended to
>> somewhat improve 2 kHz IMDDR3 *and* make this figure more repeatable.
>> 
>> Q: Say what?
>> 
>> A: As Rob Sherwood noted many times before finally immortalizing this
>> point in his must-read footnotes, A-to-D converters sharing the same part
>> number are not all created equal. The long-time previous occupant of his
>> Top Spot benefitted from a never-corroborated monotonicity in its ADC's
>> LSBs. An act of god. The product of a very good day at the silicon foundry
>> when, serendipitously, all the bunny suits were defect-free, and no one was
>> exhaling molecules of grain alcohol or other substances from the night
>> before.
>> 
>> That said, most ops can get by without the extra BDR and IMDDR3, because
>> they're not situated in the RF equivalent of the Gulf Stream. Hence the
>> different K4 models.
>> 
>> 73,
>