[Elecraft] Different focus

[Wayne Burdick]

> On Jan 25, 2021, at 8:25 AM, Bill Frantz <frantz at pwpconsult.com> wrote:
> 
> IEEE Binary Floating Point assumes (except in case of a very small, de-normalized number), that the high order bit of the mantissa is a one. With this assumption, it doesn't need to actually represent the one in memory. So, it has 8 bits of exponent, 1 bit of sign, and 24 bits of mantissa crammed into 32 bits. It should be able to handle the full precision of a 24 bit AtoD converter.

If only we had 24 bits at RF. "Reasonable cost" for an ADC that can sample at 120+ MHz gets you 16 bits at present.

Wayne
N6KR


> 
> 73 Bill AE6JV
> 
> On 1/25/21 at 8:15 AM, aa4lr at arrl.net (Bill Coleman) wrote:
> 
>> 
>>> On Jan 24, 2021, at 1:33 PM, Wayne Burdick <n6kr at elecraft.com> wrote:
>>> 
>>>> (Or did we have some other definition of “higher-performance”?)
>>> 
>>> Yes: More bits than 16, more monotonicity in the LSBs, general improvement in SFDR, or lower
>> internal noise.
>>> 
>>> Adding more ADC bits does not require more DSP horsepower. The DSP is 32 bit floating point.
>> 
>> Well, that works up to 23-bits! (FP numbers have a 23-bit mantissa)
> 
> -----------------------------------------------------------------------
> Bill Frantz        | There's nothing so clear    | Periwinkle
> (408)348-7900      | as a design you haven't     | 150 Rivermead Rd #235
> www.pwpconsult.com | written down. - Dean Tribble| Peterborough, NH 03458
> 
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:Elecraft at mailman.qth.net
> 
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
> Message delivered to n6kr at elecraft.com