Re: K3 Receiver Noise Level -- Evaluation and Adjustment

Posted by Julian, G4ILO on Dec 20, 2009; 4:43pm
URL: http://elecraft.85.s1.nabble.com/K3-Receiver-Noise-Level-Evaluation-and-Adjustment-tp4187070p4195131.html

Sverre Holm (LA3ZA) wrote

Concerning the last few days' discussion of noise and artifacts, here is a spectral analysis plot of my K3's output: http://www.qslnet.de/member/la3za/bilder/K3-CW-spectran.jpg

[snip]

It is interesting to note that my spectran plot has discrete line noise as opposed to that of G4ILO in his blog from today (http://www.g4ilo.com/blog.html), plot here: http://www.g4ilo.com/uploaded_images/k3freqresponse-736797.gif . His background noise level is similar to mine.

Could this be due to differences in individual K3's as has been hinted to on this list?

You can see here http://www.g4ilo.com/images/k3-cw-400hz.jpg my attempt to replicate your plot. Note that I could only get half the resolution due to some unexplained jumping of the display when using the 44KHz or higher sample rates - I think my 5 year old shack PC can't keep up.

My original plot was made using SSB. I did note some fixed frequency spikes originally but they disappeared (lost in the noise) when I increased the LIN OUT from the usual value of 4 to 20, so I suspected that they were not being generated by the K3. I think you should try varying LIN OUT too to be sure that you aren't seeing noise from some external source.

On my CW plot there are some small fixed frequency (do not change when moving dial frequency) spikes at 2.3KHz, 3.5KHz and 4.7KHz. These presumably are generated by the K3 because they maintain their presence regardless of the LIN OUT setting.

I changed my AGC settings to the same as yours to take the plot. Changing AGC SLP has no apparent effect but increasing AGC THR increases the signal level while the spikes do not increase, so they are less prominent (particularly the 2.3KHz one) with AGC THR = 8 which is what I used when taking the plot posted on my blog. You can experiment to see this for yourself.

I have no idea what this proves, but hope eventually to learn something. What would be really interesting would be to see plots taken before and after installing a REV D DSP board.