[Elecraft] : K3 receiver noise - my test results (long)

[Don Wilhelm]

Joe,

I think you may have hit on some major part of the "problem" here.  I 
measured the passband (and out of passband) response of my K3 and my K2, 
Yaesu FT-900 and Yaesu FT-817.  All measurements were with the SSB 
filter in all transceivers, and the K3 DSP width was set to maximum.  My 
K2 has the 2.4 kHz filter and the Yaesu transceivers have the stock SSB 
filter.

The test setup consisted of an Elecraft broadband noise generator 
feeding that receiver and the output was observed with Spectrogram.  The 
receiver gain controls were adjusted to place the peak in-passband 
response at -30 dB.
I also connected an antenna and listened to several signals 
participating in a 40 meter roundtable, and made my own 'quality of 
sound' assessment.

Several things became apparent as a result of these tests.
1) the K3 in-passband response is almost flat - from 500 Hz to 2500 Hz 
it showed  less than 1 dB variation.
2) All the other passband responses were  down about 5 dB at 500 Hz and 
essentially flat from 1 kHz to 1600 Hz, then began a gradual falloff, 
being down 3 dB at 2 kHz, down 8 dB at 2.5 kHz, the Yaesus were down 17 
dB while the K2 was down 36 dB (this is the effective high end of the 
passband).  So within the passband, there is a rolloff similar to that 
which you suggested for the receive EQ settings on all but the K3.
3) The response outside the passband on the low end was also 
interesting.  The K3 at 200 Hz was only down 3 dB, but dropped steeply 
at lower frequencies (I do not have the low frequency mod on my K3).  
The K2 was down 26 dB at 200 Hz while the Yaesus were down 20 dB.
4) On he high frequency side out of the passband, the K2 and K3 had 
almost no audio response that showed on Spectrogram - in other words 
greater than 60 dB down from the passband peak.  The Yaesu FT-900 had 
audio artifacts that were only 44 dB down at 4 kHz and for the FT-817 
were only 35 dB down.  At 5 kHz the FT-900 had audio content at -50 dB 
and the FT-817 had content only 40 dB down.

Summary of my observations - the in-passband response of both Yaesu 
filters and the K2 filter tapered off above 1800 Hz, similar to a "pink 
noise" response, while the K3 response was remarkably flat with very 
steep filter skirts.  The skirt slope of the other filter passbands were 
more gentle.

On the high frequency end, both Yaesu transceivers had considerable 
audio content while the K2 and K3 content was much lower.

My conclusions: I am not certain what K3 owners are perceiving as noise 
any more than I  had been before these tests.   The flatter passband of 
the K3 may be sub-consciously perceived as being more 'harsh', but I 
could not equate that to 'noise'.  The other filters with a high end 
in-passband response that drops off similar to a "pink noise" response 
*could* be causing a perception of additional noise *if* their local 
noise has significant content in the 2 kHz to 3 kHz range - yes, the K3 
will make this section of the audio spectrum louder than the other 
receiver I measured.  The "audio hiss" is not coming from the high 
frequency spectrum outside the passband because in that area, the K3 and 
the K2 are much more quiet than the others measured.

It would be interesting (at least to me) for someone to make similar 
measurements on an FT-1000 - perhaps with the INRAD filter.  I am 
especially interested to know if the in-passband response shows a 
similar taper at 2 kHz and above, and what is the response for the 
FT-1000 above the filter passband.

For those trying the "pink noise rolloff", my measurements suggest that 
you should set the RX EQ bands from 100 Hz through 1600 Hz at 0 dB, 2400 
at -3 dB, and 3200 at -5 dB.  The 50 Hz band can be set at -16 dB for 
the reasons Joe has stated.

73,
Don W3FPR

Joe Subich, W4TV wrote:
> If one is going to attempt to mimic the "pink noise" behavior, 
> there is no need to bother with the low frequency boost.  It 
> simply puts more strain on the audio amplifier, increasing the 
> amount of IMD products, with little or no aural benefit.  A 
> better RX EQ configuration with essentially the same perceived  
> sound: 
>
>   50:  -16 dB (reduce sub-vocal noise and hum) 
>  100:    0 dB 
>  200:    0 dB 
>  400:    0 dB 
>  800:    0 dB 
> 1600:   -3 dB 
> 2400:   -5 dB 
> 3200:   -6 dB  
>
>   
>