Elecraft K3


The Elecraft K3 transceiver, available as either a kit or factory-built, has received some very favourable reviews for its receiver performance particularly its close-in IMD and dynamic range.  Its functionality is largely controlled by firmware which is being updated on a regular basis in response to user suggestions. This service is both very responsive and free.  Additional hardware options are also available -  a 6m preamp, a 2m transverter that fits inside the K3 enclosure, and a separate panoramic adaptor.

Construction of the kit

My K3 serial 1198 kit was received on 4 July 2008, and was built 7 July, with firmware version 2.10.  It was straightforward.  The stages of construction can be seen in the photos.

K3 rf board

  1 - RF board2 - Filters and panel fitted

rear and side panels fittedrear panel

 3 - Rear and side panels fitted4 - Rear panel view

front panel fittedtransceiver complete

5 - Front panel fitted6 - Transceiver completed

K3 with 4m transverter showing GB3ANG locator

7 - K3 with 4m transverter decoding the locator of the GB3ANG beacon sent on cw

Modifications since first build:-
11 Jan 2009 - KRX3 second receiver, KDVR3 digital voice recorder, AFMODKT, 6kHz filter on main rx, plus treatments for birdies, encoder noise, and RFI. 14 Oct 2010 - VFO noise mod (to cure encoder noise - see K3_vfo_noise_mod_Rev_C.pdf  and MCU3.25 notes);  Speaker output mod to protect  audio chip (see AF_Output_Mod_Rev_C.pdf); K3 IF o/p mod to increase output for P3.

The P3 panadaptor

final view without lidOrdered 5 Sep 2010, delivered 13 Oct 2010 (delayed at my request), built 14 Oct 2010.
[The UK Border agency charged me import duty.  It took over 2 months to get a refund, and still waiting for explanation.]
My P3 is serial no 666 (my lucky number!!)

With the P3, it is easy to spot activity away from the tuned channel, even at very low signal levels (but need to reduce span and scale to get the best sensitivity).

I have also used the P3 to make some measurements (P3 displays signal level to <1dB, although accuracy is only about +/-0.5dB). Measured the height gain  achieved from 31 to 46 feet AGL at 3.5dB on 70MHz; and antenna polar diagram.

NEW! Here are some measurements of the transmit spectrum of the K3 using the P3.

Display boardbox internals
 The display board     A mostly empty box!

Likes, dislikes, problems and solutions

I really do like:-
Here's an example of signal processing in the K3.  In this mp3 recording (1.3MB,1min 24sec, recorded at 1739utc on 28 Feb 2011), you'll hear first the GB3ANG beacon on 70MHz (tropo, 607km) in ssb bandwidth (2.3kHz) with no noise blanking. There's bad power line noise.  At 6.7seconds in, the noise blanker is switched on (set at dsp t2-6, IF narrow 3). At 21.4secs in, the mode is switched to cw - the audio tone remains the same, and the bandwidth changes to 900Hz.  At 30.5secs in, the APF is switched on until 56.4secs.  At 61.5secs we revert to usb mode and at 68.3secs, the noise blanker is turned off.

Issue / dislikeResolved by
Sprogs / birdies - especially on 28 and 50MHzUsing firmware 3.10 approach for fast tuning sprogs - mostly successful
Encoder noise / tuning hash / VFO noise - esp on 28.006 and 28.047Check tightness of all screws and cables. Problem reduced once 2nd rx installed. VFO noise mod done, changed Config:VCO MD to SPI 2 to change bus speed
Problem cured.
Deaf rx on 6m - noise figure is 13dB.  You can measure noise figure with a signal generator and this spreadsheet.
Using a preamp - I have my own but Elecraft sell one also
Unable to simply scan through all frequencies in memoryCan set memories to "channel hopping" mode, but it's rather clumsy
The "noisy" receiver problem - in my case a reduction in S/N on 6m with slight reduction in RF gain control - problem evident with firmware 3.63, not there with 3.27. S/N resolved by performing RF gain calibration (using K3 utility) and using factory defaults rather than an external signal generator.  I have also reduced the receiver gain by 3dB by adjusting RX equalisation values.
Also noted with f/w 3.63 weird behaviour of S meter when RF gain backed off - it increases to S9+25 then falls back to S7.The above gain calibration resolved the weird behaviour.  
S meter reads up to 20dB too high at signal levels above S9+10dB. Was ok with f/w 3.27, but now wrong again with 3.63 This can't be cured by the s meter calibration procedure. Problem outstanding (at 21 Dec 09). And it's still outstanding at 2 Jan 2011 with f/w 4.22
Sub receiver on 6m has spurious responses of the form nLO +/- IF. Only affects the sub receiver - the main has adequate filtering.Problem requires a band pass filter.  Details below.  The subreceiver seems to have a reasonable noise figure (I measured 6.4dB) and doesn't usually need a preamp.

Sub-receiver spurious responses

When tuned to 50.150MHz, the local oscillator is on (50.15-8.215) ie 41.935MHz. The image is at 33.72. Harmonics of the local oscillator produce spurious responses at 75.655, 92.085, 117.59, 134.02 etc.  My measurements of these are:

Frequency MHzRelative response dB

I used Elsie to design a Cauer bandpass filter with parameters as follows:-

Centre frequency50.5MHz     Frequency MHzResponse dB
Passband ripple0.2dB
Stopband depth40dB




The circuit is as follows:

Band pass filter for 6m sub receiver
Measured results are as follows

Original subrx response dBBPF response dBSubrx response with BPF dB
32.92filter null
77.47filter null