[Elecraft] DSP - 4 questions

[Lyle Johnson]

Hello Bill!

> Now, with the KDSP2, there's no such analog filter.

Correct.  It is a digital filter, not an analog filter.

> Looking at the
> schematics, it appears the audio goes directly through the A/D D/A chip.
> I don't see any way that it is bypassed, unless that happens internally.

When the DSP is commanded to the bypass mode, the CODEC chip bypasses the
A/D and D/A and operates as an amplifier.

> Other than a single pole R/C filter, there doesn't appear to be any other
> low-pass operations to eliminate the hiss, even if the A/D D/A can be
> bypassed.

Correct.  If the filter is on, the hiss is gone.  If the filter is bypassed,
it is, well, bypassed :-)

> So, this implies that the low-pass operation takes place in the DSP
> itself. The bad point about that is that the digitizing operation
> introduces quantization noise, which limits the dynamic range of the
> audio. Unless peak signals are hitting the maximum limits of the A/D
> converter, this dynamic range is going to be substantially less than the
> 90 dB specified.

The dynamic range capability is 90 dB.  The dynamic range of audio presented
to the DSP if AGC is ON is much less than this.

For example, if you have Spectrogram, tune in a loud signal and look at the
noise floor.  On my K2 it is much less than 90 dB below the peak of the
strongest signal.

Most strong signals I have examined have a signal-to-noise ratio of 25 to 45
dB. The 4 MHz "test tone" heard on 80 meters on a K2, for example, is about
40 dB above the noise on my K2.  Giving away 20 dB at the top leaves 70 dB
capability, and in this case we still have 30 dB of margin.

Looked at another way, if you are listening to an SSB signal, the
transmitted IMD products are usually only 30-35 dB or so below PEP, limiting
the useful dynamic range of the signal to about the same 30-35 dB.

I believe the A/D and D/A capability of the KDSP2 far exceed the needs of
good communications quality signals.

> While I like the features of the DSP, I'm worried that having the A/D
> conversion in line all the time will slightly raise the noise floor of
> the rig, especially for weak signals. And I don't like the idea of
> disabling the filter, since that will re-introduce the audio hiss.

The only reason to disable the DSP filter is to save power if you are
running on batteries.

Your ears may be much better than mine. I don't hear any increase in noise
floor, nor do I measure one. If I use the sig gen, the weakest signal I can
discern is no different with DSP or without.  The quality of the signal and
its apparent signal-to-noise ratio does change though, in most cases in
definite favor of having the DSP.

I do hear a marked decrease in noise, though, with virtually any filter
selected.  The narrower the filter, the less noise bandwidth and the less
noise.  And the denoiser drops the background noise I hear by 10-20 dB
depending on how it is set up.

But this sort of thing is highly subjective.

YMMV

73,

Lyle KK7P