[Elecraft] K3 Harmonic Distortion

[Jack Smith]

Brendan:

You might wish to look at the measurements I made, with LIN OUT settings 
from 2 to 100.  There's an animated GIF that steps through the LIN OUT 
settings so you can see  the effect. The data was  taken at 600 Hz.
http://www.cliftonlaboratories.com/elecraft_k3_receive_audio.htm about 
half way down the page

Like you, I see a relatively constant 45 dB or so offset between the 
fundamental and 3rd harmonic for LIN OUT values of 10 and over.

However, setting the LIN OUT at 2 results in quite a bit better odd 
order harmonic suppression, on the order of nearly 70 dB. By the time 
LIN OUT is set to 5, however, the odd order suppression is reduced to a 
bit more than 60 dB. At LIN OUT = 10, it's 45 dB or so where it stays at 
all values of LIN OUT tested between 10 and 100.

(For some reason, the 2nd harmonic pops up at certain settings, but not 
others higher or lower, but that's not of immediate concern.)

There's a  related issue with the AGC. The audio output is proportional 
to RF signal level, as the AGC has a slope to it. Hence, even though the 
LIN OUT might be set to, say, 2, the actual audio level and hence the 
distortion, will vary as the RF signal level changes. The degree of 
audio level variation with RF input changes depends on how the user has 
set the K3's AGC slope and intercept parameters.

I'll make  the measurements on a couple of TTC-108 parts when they 
arrive as a matter of trying to understand where the harmonic distortion 
arises. To duplicate the K3's audio chain, I'll drive the TTC-108s from 
an op-amp so that it will be hit with a low Z source, instead of the 
standard audio generators available to me, with 50 or 600 ohm output Z.

Jack


Brendan Minish wrote:
> Jack 
>
> I have seen postings on the main reflector recommending settings as low
> as 3 to avoid distortion products that are being attributed to
> transformer distortion. This did not really make sense to me, especially
> at such low levels where consumer sound cards tend to get non-linear and
> nasty.
>
> I took a look at the specifications for the TTC-108 audio transformer
> used in the line out of the K3 
>
> Nominally it's specified as having an operating level of -45dBm to
> +7dBm 
> THD is specified at 0.5% @0dBm 300Hz to 3.5 Khz @ 0dBm
> this would make the transformer related harmonics worst case at -46dB
> relative to the fundamental  
>
> I did some measurements with an s9 CW signal zero beat at various pitch
> settings
> I used an old copy of SMAART pro that I have from my Sound engineering
> days. the sound card is a delta44 
>
> Lin = 100       Pitch 300       pitch 550       pitch 800
> DR3             47.3dB          51.4dB          54dB
>
> Lin = 50        Pitch 300       Pitch 550       Pitch 800
> DR3             43.1dB          47.6dB          50.6dB
>
> Lin = 10        Pitch 300       Pitch 550       Pitch 800
> DR3             35dB            39.7dB          43.1dB
>
> Lin = 5         Pitch 300       Pitch 550       Pitch 800
> DR3             31.9dB          37.7dB          42.7dB
>
> Lin = 3         Pitch 300       Pitch 550       Pitch 800
> DR3             31dB            40.3db          47.6dB
>
> DR3 is the Dynamic range between the fundamental and the 3rd harmonic
> (the most dominant harmonic by a wide margin)
>
> I did a quick look at 5th and 7th orders too, again the best performance
> here was found with LIN = 100  
>
> LIN = 100 is an output level of approximately 700mv for an S9 CW sig on
> my K3 
>
> The Harmonic distortion does not increase with higher line out levels,
> it actually improves. This does not appear to be a transformer
> saturation issue 
>
> With lin = 100 users will see a ghost on the waterfall and may risk
> over-driving some consumer sound cards. lowering LIN may get rid of the
> visible ghost by reducing the 3rd Harmonic to below the waterfall
> threshold (& perhaps the sound card input noise floor ) but also reduces
> the overall usable dynamic range  
>
>
> 73
> Brendan EI6IZ 
>
>