Excessive cabling, TR relays, bypassed amps in lines causing IM oscillations.

[W3RSW]

A response or on eHam.net prompted this.
As mentioned on the previous K3 threads:

Some of my first IM spectral scans were affected by external paraphernalia on the way from the K3 to 50 ohm dummy load.  When all this extra switching, cabling, etc. was removed then the scans were much cleaner without extra hash. Here are a couple of those early corrupted scans at 100 watts where serious oscillation of entire spectral display occurred with about a one second interval. This shows up historically in the waterfall from min. to max. width of display intensity in the waterfall. Shots of the display were taken at minimum garbage and at maximum garbage as best as I could time them.

First pix is at minimum oscillation: please observe waterfall too.
Second is at maximum oscillation, a half second or so later.
Oscillations occurred at about 1 sec. periods, sometimes quicker, sometimes longer depending on power level. Never repeatable periods even at the same power level during later scans.

To repeat previous thread, last scan here is at 100 watts with a direct, short LMR400 connection to 50 ohm dummy load. Much cleaner.

[Steve - K4HX]

What is oscillating? Do you mean that your display changes?

[flintstone mop]

The display is of a transmitter modulated with a two-tone frequency? Did you look at the signal path of the K3 with an analyzer to see if it is operating into a pure 50 ohm resistive load? The final might be sensitive to the complexity of an RF load and may affect its IM.........dunno
The garbage disappears after removing inter-connecting cables and TR relays?
That would be an interesting test for me to do with my rig, Kenmore TS850 in SSB and two-tone, with no ALC action and check for purity. Try to copy what you did by removing the hardware in circuit.

There have been some discussions here? and eHam about the K3, and that it is typically not a very clean RF output. There were arguments that Elecraft and many other manufacturers are taking bad shortcuts with PA final design and using a 12 vdc supply. Then, there are others arguing that some manufacturers have clean RF output using a 12vdc supply. Cost-cutting, short cuts, bottom line $$$ profit.

Fred

[W3RSW]

Steve, yes.  I've just shown the min. and max.  Intermediate 'frames' and real time viewing show rising and falling of the crud. You'll note the rising and falling of the freq spread/intensity in the waterfall showing the progression over time.

At any rate a direct connection with short, new coax eliminated virtually all the wah wah (as in music) effect from very bad to not so bad.

In addition I see Elecraft had issued a tech. mod. for dirty KPA3's at one time; not my later serial no. rig which has the fix.  Not related to my findings but makes one wonder if Sherwood duplicated ARRL's earlier review with the unmodified PA.  ARRL later amended test with the fix, but who knows.

At any rate my IM 3, 5 and 7's measurement peaks pretty much track Sherwood's but without all the attendant high grass and crud that his show.

Fred, yes two tones internally generated in the K3 as explained in previous posts today. The two largest twin peaks represent the audio fundamentals, 700 and 1900 Hz. They are 6db down from a phantom point between the two representing PEP.  IM3, 5, etc. clearly show and are accurate relative to the fundamental tones.  
The 50 ohm dummy load is about as pure as you can get at the low frequencies involved, totally encased in metal and oil cooled.

Yes, Your comments mirror tons of discussion on this and other boards. The whole thread on EHam and reason for my testing came about because an Icom marine transmitter with 12 volt finals matched a much more stringent FCC mask than what most ham transmitters produce.  Generally the writers wanted to know why the Icom/Yeasu/...etc. including Elecraft couldn't produce ham rigs just as clean.  A whole host of reasons, some quaint, some cracked, some market related, some disingenuous, ...you name it ensued.

Read the eham thread for far more than you've ever wished, both technically and sophomorically.  

[Steve - K4HX]

Interesting. I wonder what's causing the variation of the crud and/or noise floor?

Thanks for the details. I'll skip the eHam nonsense.