Elecraft K3S qsk qrq performance test from 100 wpm to 60 wpm

A qsk qrq cw performance test for the Elecraft K3S from 100 wpm to 60 wpm....K3S set to FULL QSK, qrq cw mode enabled

RESULTS: at 100 wpm several of the dits were way too short causing harsh audio pops...all the dits on this test can vary between 3-6 milliseconds in length...there is also quite a bit of first dit shorting by almost half... however, by only listening, from 90 to 60 , the variance in element length is not really a big deal... only when you analyze it closely does the K3S's erratic element timing come to light

some examples from audio forensic analysis:

at 100 wpm test, many of the dits are uneven

at 100 wpm - here is a good example of the super short dits causing harsh audio pops

the "normal dits are 19ms long, the super short dits are only 9ms long

at 100 wpm here is a good example of the K3S's 1st dit shortening "bug"

the short dit is only 11ms  long, the other 3 dits are 19ms long

at 90 wpm here is a good example of the K3S's 1st dit shortening "bug"

the short dit is only 12 ms long and the other 3 dits are 20-21 ms long

at 90 wpm even the dashes are not consistent in length

1st dash is 46ms long, the other two are 41ms long

at 90 wpm here is the letter "L" showing dit length variance

at 70 wpm these dits vary from 18 to 23 ms long

at 80 wpm on the letter J, the first dash is 52 ms long and the others are 46ms long

at 75 wpm - sending the number "5" several times, with the K3S in [qrq mode +], full qsk, the K3S does not produce even one of the "5"s  with good precision in timing - they are all inconsistent and dit lengths vary widely

at 75 wpm - sending a single dit, with the K3S in [qrq mode +] full qsk, it takes the K3S 180 milliseconds before the receiver audio returns - pictured below - on keydown, the RECEIVER AUDIO stops 3 ms  before the cw note starts - and after keyup, the K3S full qsk takes 180 ms before you can hear receiver audio again - also, the dit is way too short

the K3S sidetone, has a couple of harmonic spikes(11.1khz & 12.5khz) that causes audio clicks and pops - which becomes annoying at qrq speeds...i had to use an audio bandpass filter to get rid of these audio harmonics in the K3S sidetone

If you use the K3S  rear panel "LINE INPUT" for "MIC" input, (or the K3S USB SOUNDCARD LINE INPUT) and turn up the volume of the K3S TRANSMIT AUDIO MONITOR circuit,  there is a lot of audio hiss when PTT is active, before any external audio input shows up,  the K3S AUDIO HISS from closing the PTT is significant ..shown below, and the hissing can be heard without any signal input and with signal input, you can still hear the hiss on top of the actual input audio


Views: 720

Comment by Chuck aa0hw on September 10, 2016 at 9:51am


A qsk qrq cw performance test for the Elecraft K3S from 100 wpm to 60 wpm....K3S set to FULL QSK, with qrq cw mode enabled

- test starts at 100 wpm then 90, 80, 70 and finally 60wpm -

RESULTS: at 100 wpm several of the dits were way too short causing harsh audio pops...many of the dits & dashes from 70 to 100 wpm, on this test, varied at times, between 3-6 milliseconds in length...there is also quite a bit of first dit shorting by almost half... however, by only listening, from 100 to 70 wpm , the variance in element length is not really that big of a deal as far as being able to copy what is being sent by the K3S... However, when you analyze it closely above 60 wpm up to 100 wpm, (THE ADVERTISED LIMIT of the K3S at full qsk with qrq mode enabled) - the K3S's erratic element timing comes to light.

this video also demo's how much receiver back ground audio is getting through between words and/or characters at the speeds of 100, 90, 80,70 and 60 wpm

please see this ongoing article, linked below, for more info and to see the audio forensic analysis with pictures of the test using Audacity

Comment by Chuck aa0hw on September 21, 2016 at 11:40am

HERE IS A VIDEO of how to improve the K3S qrq cw keying using FLdigi

Using FLdigi's qsk controls and 2 individual audio derived cw keying switch circuits(ADCWKS), this video demonstrates the advantages of using the K3S / FLdigi PTT system, to reduce the K3S cw timing / jitter errors at qrq speeds.

TEST SETUP: running my K3S in full qsk. [full qrq mode enabled +]
- the k3s by itself and just keying the cw jack, you will see in this video, by sending qrq cw from 70 to 100 wpm, how the k3s shortens many of the dits to where they are almost 1/2 the size that they should be...HOWEVER, if you activate the PTT before keying the cw jack, the K3S TIMING improves substantially.

FLDIGI: qsk on rt channel enabled...
RT CHANNEL audio output goes to the PTT ADCWKS circuit
LEFT CHANNEL audio output goes to the CW ADCWKS circuit
using a USB SOUND CARD with an audio amp like this:
Two audio derived cw keying circuits are used to take the audio output from the left and right channels of fldigi, and convert the FLDIGI audio output into a keying signal for the k3s PTT JACK and CW JACK..
-see here:

Unfortunately, there is built into the K3S, about ~20 milliseconds of delay from CLOSING THE CW JACK until the RF is generated.. see here:
http://i.imgur.com/ITdpA5m.png (blurry photo, but you can see the PTT GRAPH on top of the RF WAVE and get the idea)...that the closest you can get the K3S PTT start- to the RF WAVE START, is about 20 milliseconds(measured on my scope to be about 18ms)...so, you have to live with that as far as how much receive audio you can get from the K3S during full qsk - between characters and words ... etc...

ALSO: my K3S has some phantom harmonics on its sidetone, around 11.1khz and 12.5 khz , which causes audio snaps and poppings on the edges of the K3S SIDETONE....so i use a simple bandpass plugin to get rid of that....demo'd in this video...

ALSO: the video demo shows you how to adjust the VOLUME OF THE LEFT CHANNEL to effect weight and compensation on the CW KEYING JACK/K3S OUTPUT and also you can get even more control of your cw "weight" in both equal time per element and/or equal percentage per element by adjusting the RISE / FALL time on the FLDIGI cw modem controls

SUMMARY: so basically we are using FLdigi to key both the K3S CW jack and the PTT jack. We try to time the PTT to go just before and just after the CW JACK is keyed....but the limiting factor to this is the 20 ms of delay built into the K3S from CW JACK KEYDOWN, until the actual RF OUTPUT WAVE starts. Still there is plenty of receive audio during full qsk between words above 60 wpm.

NOTE: using this method , my K3S can qrq up to about 120 wpm .
NOTE: for the usb sound card used, i set its volume to about 1/2 way up the ladder(just enough to get the ADCWKS to work...and then just a tiny bit more) using the LINUX/WINDOWS OS native sound card volume control...also - i turned off AMPLIFICATION MODE...a sound card option that you could use, with the LINUX or WINDOWS own sound card properties control panel in the PLAYBACK TAB.

Comment by Chuck aa0hw on September 27, 2016 at 1:41pm

Here is a video demo of an experimental setup using AFCW mode to key the ELECRAFT K3S over 100 wpm

This video shows an AFCW A2 MODE transmitter setup for the K3S, to enable it to send perfectly timed cw over 100 WPM. This test runs from 100 wpm to 150 wpm. TWQRQ is the software cw keyboard linux application used to transmit high quality, perfectly shaped cw audio waveforms to the K3S audio input in LSB or USB.

NOTE: you will hear a separate 2nd receiver on the LEFT and the K3S sidetone on the RIGHT...when the A/B switch is moved back and forth in this video demo. The K3S sidetone matches extremely close to what you see and hear from the 2nd receiver(kk7b high performance DC RCVR)...see here: https://www.arrl.org/files/file/Technology/tis/info/pdf/9208019.pdf

Several software apps are needed to make this work as shown...
1. you have to delay the CW TONE from going into the K3S USB SOUNDCARD LINE INPUT by ~50 milliseconds after the PTT is activated. (this delay is hard coded into the K3S firmware)

2. you have to make sure you DO NOT over drive the K3S SOUNDCARD LINE INPUT....so by looking at my OSCILLOSCOPE, I was able to see the first signs of the ALC trying to trim back the RF OUTPUT from too much audio input...then lowered it slightly so there were no signs of RF WAVEFORM distortion on the scope..you can see RF SPIKES on the rise time if you have too much audio input volume....(on my system I had to set the audio attenuator LV2 PLUGIN app to minus -25db)

TWQRQ sends out its beautiful qrq cw audio tones to the input of the software delay line going to the INPUT of the K3S USB SOUNDCARD LINE INPUT pins inside JackRouter's connection bay. At the same starting time TWQRQ CW AUDIO TONES go to the
PTT AUDIO DERIVED CW KEYING 2nd soundcard/SWITCH(no delay)... which activates the PTT CIRCUIT on the K3S and keeps it down for an extra 70 milliseconds after the TWQRQ CW TONE is finished...This is the best semi qsk i could get without chopping off the RF ...so at the sacrifice of full qsk, and going to semi-qsk, you can still hear RECVR AUDIO between words or at the higher speeds....between sentences.

NOTE: for more info on this beta version of TWQRQ 0.9, send an inquiry to WA0EIR ...http://wa0eir.bcts.info/twcw.html

NOTE: one of the advantages of keying qrq cw this way, is that you can choose your own RISE/FALL TIME, your own weight by percentage, your own weight by TIME, and you can choose which kind of raised cosine edges you want to transmit. HANNING, HAMMING, BLACKMAN, NUTALL ...each raised cosine window has its own unique qrq cw note sound.

the software apps were all free LINUX LV2 plugins:
1. Calf Filter
2. Simple Delay Line = linear interpolation
3. Simple amplifier

NOTE: the audio conversion process by YOUTUBE added some of its own noise to the cw audio edges that are not on the original...but still i think this is a good representation of what can be accomplished by using AFCW mode for qrq'n the K3S over 100wpm.




JACKROUTER CONNECTION BAY WITH WIRING DIAGRAM (click on picture for full size image)

Here are the LV2 PLUGINS apps and their settings: (click on picture for full size image)

Comment by Marc Veeneman on October 29, 2016 at 5:41pm
I guess it's time to ask what currently available transceiver would be better than the K3S for QRQ, both transmitting and receiving. I'd like to avoid using a separate computer to drive the rig, if that's possible.
Comment by Chuck aa0hw on November 1, 2016 at 6:49am

Tom, W4BQF, has made some comments about this TOPIC for best QRQ CW RIGS,  on his blog: 


A quote from his article:

"The radios that I know of that can run full QSK at speeds over 100 wpm are the Ten Tec Corsair II and the Icom IC-781, and they do it flawless. In my opinion, full QSK is not at what speed you can hear another signal between dots, but at what speed can you hear your fellow ham trying to break you! Hearing between dots is a fine criterion for speeds below about 40 wpm, but is inconsequential over 60 wpm. Unfortunately the more digital circuitry that is added to modern transceivers, the less high speed QRQ capable they become."

and He also made reference to great qrq qsk full break in with the ICOM 7700...i think he tested and measured on his SCOPE that the cw elements were in perfect form up to 150 wpm or maybe up to 200 wpm ...

Comment by Chuck aa0hw on December 16, 2016 at 1:57pm

Here is a RAPSBERRY PI Desktop & LIVE OSCILLOSCOPE VIEW about how to properly adjust the FLdigi RT CHANNEL QSK controls for best possible QRQ QSK FULL BREAK-IN cw keying on an Elecraft K3

Using an oscilloscope you can dial in the FLdigi RT QSK control so you can get better cw keying on an Elecraft K3 by activating the PTT on the K3 just before until just after the CW JACK is keyed.

A Raspberry Pi 2b(Raspbian Jessie) is demo'd in this video keying the ELECRAFT K3 by using the PI'z Raspbian version of FLdigi and with the FLdigi's audio tones output - connected to the input of 2 home brewed audio derived cw keying circuits..FLdigi cw audio output will activate and key the K3 PTT jack - and - the K3 CW jack - in sequence - in order to provide the most possible QSK HF BACKGROUND AUDIO between the K3's CW KEYING EVENTS...and also make things better to be able to key cw on the K3 at the highest CW WPM speeds(100 to 120 wpm)(the K3 is advertised as being able to only key up to 100 when using the QRQ + feature)

using just the Pi'z own native analogue sound card, we are going to tell FLdigi to use that sound card for its audio output which will activate 2 individual audio derived cw keying switch circuit circuits(ADCWKS): see here: http://qrqcwnet.ning.com/forum/topics/what-circuit-for-computer
- one of the ADCWKS will key the PTT jack on the K3
- the other ADCSKS will key the CW jack on the K3
and precision timing of when the PTT and CW jacks are keyed are achieved by visual inspection by observing the SCOPE K3 RF OUTPUT and by listening to audio artifacts that represent errors in K3 cw keying...

so while sending a dit stream....you adjust the FLdigi RT CHANNEL QSK CONTROL timing by watching the scope AND by listening to any clicks or pops or artifacts that would suggest the timing is wrong....then make more adjustments to PRE and POST qsk FLdigi timing in the FLdigi MODEM for that feature.

The idea is to key the PTT on the K3 first - before keying the K3 CW Jack. When the PTT is activated, the K3 performs much better at qrq cw full qsk keying(especially with any cw speed over 50 wpm). Otherwise, when just keying the CW JACK, the K3 puts out shortened dits and erratic dit lengths...and on my K3, the top transmit cw speed is limited to just over 90 wpm, whereas if the PTT is activated, the K3 will key well all the way up to about 120 wpm...AND when the PTT is activated the dit length variance is about half of what it normally is when just keying the K3 CW JACK.
- see this article for more info:

FLdigi and the Raspberry PI work well together with its native analogue sound card to make it fairly easy to get GREAT QRQcw keying on the K3...while preserving much of the "in between" keying... HF AUDIO BACKGROUND - for good break-in ability for the usual reasons for choosing to operate QSK.

NOTE: JACK AUDIO CONNECTION KIT is used in this demo...using the jack apps from the AutoStatic Raspbian repo's. See here:

Comment by Chuck aa0hw on April 7, 2017 at 9:57am

Here is a video demo'N the K3S chopping off CW ELEMENT lengths... and sending short dits when just using the CW JACK in full QSK mode and FULL QRQ + mode activation in the K3S software options.......then the PTT is activated and you can see and hear the cw elements(dit stream from a cw keyer keying the K3S) return to their normal length

using a CW KEYER to send a stream of dits at a speed where this old analogue oscilloscope can keep it in the center to view the waveform - THEN - the PTT JACK is keyed to activate for 3 seconds, then the PTT is released for 3 seconds ...cycle repeats a few times - kind of an - A / B test - PTT vs NO PTT - with normal cw keying of the CW JACK on the K3S at full QSK, with the K3S QRQ + OPTION selected and activated

- RF WAVE on TOP SCOPE is from the K3S RF output
- rectangular waveform cutting through the RF ENVELOPE =
voltage from the K3S CW JACK KEY input (SCOPE PROBE)

SCOPE on the BOTTOM is a software lv2 plugin that is measuring the K3S SIDETONE MONITOR's audio output...(which closely follows the RF output as far as length(timing) of the cw elements)

RF OUTPUT starts ~10ms after the CW JACK is closed...
RF output goes to ZERO about 10 ms after the CW JACK is OPENed

RF OUTPUT starts ~3 to 4 ms after the CW JACK is closed...
RF OUTPUT goes to ZERO about 10 ms after the CW jack is OPENed

MEASUREMENT RESULTS of the LENGTH CHANGE of the CW ELEMENTS when using or not using the PTT with the CW KEYING...

- each CW ELEMENT is shortened by about 6 - 7 ms
- the dit lengths are "normal" in length - more accurately following the original cw timing output from the CW KEYER

if you activate the PTT just before and release the PTT just after the CW JACK is keyed...(FLdigi can do this with its RT CHANNEL QSK control...as well as most CW KEYERs in their software options) you will get more accurately produced CW ELEMENTS -that closely follow the CW KEYER's original output timing...
ALSO - with PTT /CW, you will get less variance in the K3S element length in general..since the K3(S) does have CW ELEMENT variance...up to PLUS or MINUS 3 to 4 ms...
HOWEVER, when using the PTT with the CW keying, the CW ELEMENT variance is cut in half...
ALSO - on this K3S, when using the PTT with the CW keying, the QRQ KEYING will go higher...before it starts glitching... (on this K3S, when just keying the CW JACK - No PTT - with CW VOX "ON"...the TOP qrq speed is about 94 wpm(qrq+engaged)(full QSK)...where it starts to break down and starts sending erratic CW ELEMENTS etc.... HOWEVER, when using the PTT - with - the CW keying, this K3S can then QRQ up to about 120 wpm without any CW ELEMENT transmitting glitches ...

Comment by Chuck aa0hw on April 7, 2017 at 12:04pm

Here is another short video demo of the K3S transmitting excellent FULL QSK CW from 20 wpm to 120 wpm when using the K3S PTT JACK keying technique with the usual K3S CW JACK keying...both JACKS being keyed by an audio derived cw keying switch...from the STEREO AUDIO OUTPUT of FLdigi using FLdigi's QSK RT CONTROL option

With the K3S setings of
1. NO CW VOX(PTT has to be activated before CW is Xmitted)
2. full QSK,
3. K3S qrq+ option selected in MENU and activated...
4. using FLdigi to "key" the PTT JACK before the CW JACK is keyed...while sending a text file...

a CW TEXT file is sent from FLdigi from 20 wpm to 120 wpm to
demo the FULL QSK receive audio performance ...and to demo how well the Xmit CW does in general at the highest qrq cw speeds possible on the Elecraft K3S(advertised to max out at 100wpm)

when using the FLdigi's RT CHANNEL QSK feature, you can key both the CW JACK and the PTT JACK by using the STEREO AUDIO OUTPUT of FLdigi and have both the LEFT and RIGHT audio output channels of FLdigi going to 2 individual AUDIO DERIVED CW KEYING CIRCUITS receiving the FLdigi cw tones, from the computer''s sound card... using circuits like this:

without the FLdigi PTT technique for keying the K3S, the top speed on my K3S is @ 95 wpm before it starts glitching ...
HOWEVER, when using the FLdigi's RT CHANNEL QSK feature, and setting the RT QSK FLdigi options to
4 ms PRE and -4 ms POST for this test...
- the K3S does remarkably better at QRQ CW and was able to increase its top qrq speed by 25 wpm ...(went from a max of 95 wpm... to a max of 120 wpm when using this PTT technique)

NOTE: using the PTT technique also helps prevent the K3S from chopping off tiny bits of each CW ELEMENT...previously measured to be shortening each cw element by 6 to 7 ms(when not using the PTT) AND it also reduces the K3S's CW element timing variance from plus OR minus 3 to 4 ms - (when NOT using PTT)... down to plus OR minus 1 to 3 ms cw element variance - (when USING PTT technique)

NOTE: the oscilloscope usb camera did not function well for this video...but may still be useful for inspection of the cw elements when in view...and also to view the PTT PROBE waveform on the scope to see when it is activated and released in relation to the actual RF waveform...
NOTE: DUMMY LOAD was used and RF GAIN turned up to increase the background receive audio static, so you can hear how much of it remains between XMIT & RX during FULL QSK
NOTE: the lower SCOPE view is the sidetone monitor audio of the K3S, which closely resembles and "tracks" the RF output of the K3S...so if you hear glitches in your K3S SIDETONE MONITOR audio.....you most likely have something going awry with the K3S RF output too...
NOTE: FLdigi was sending cw from a Raspberry PI2b using its own analogue sound card speaker jack...audio cables connected from the pi's speaker jack to the input of the the 2 audio derived cw keying circuit switches ...LEFT AUDIO CHANNEL to the CW JACK keying circuit and the RT AUDIO CHANNEL to the PTT JACK keying circuit

Comment by Chuck aa0hw on April 7, 2017 at 2:20pm

Here is another video demo of the K3S being keyed at Extreme QRQ CW speeds using a Rapsberry Pi

This is a follow up video of the previous video(showing FLdigi using its RT CHANNEL stereo QSK feature to CW KEY the K3S up to 120 wpm)...this video shows a Raspberry Pi2b keying the same ELECRAFT K3S(k3 with all the k3s updates//upgrades) but not using FLdigi...instead TWQRQ is used to provide the software CW KEYBOARD, and some free LINUX LV2 APPS are used to provide the function of keying the PTT before and PTT "HOLD" for a certain amount of time AFTER the CW jack is released,...all controlled by using the Pi's own stereo audio output analogue sound card speaker jack going to a PTT and CW audio derived keying circuit that connects to the PTT JACK and the CW JACK of the K3S.
The idea is the same as the previous video...key the K3S PTT first, before the K3S CW JACK is keyed...and then HOLD the PTT until just after the K3S has finished transmitted the CW NOTE. 2 AUDIO DERIVED CW KEYING CIRCUITS are utilized ...see this article for more info on these circuits:


WA0EIR's Linux qrq cw keyboard, TWQRQ, is used to send out the morse code audio tones... http://wa0eir.bcts.info/twcw.html

TO setup the PTT pathway...we are going to use a steady output audio TONE at 3.4 khz or so...from the free LV2 PLUGIN called INVADA TEST TONES...it first goes to the RT CHANNEL on an LV2 GATE plugin, called "CALF GATE", but the CALF GATE threshold is raised "just above" the INVADA TEST TONES output volume level, so that it will not actually go through to the sound card.... just yet....
TWQRQ's audio output is routed to channel 1 of the CALF GATE
and when it arrives, its volume is louder than INVADA TEST TONES volume and thus triggers the CALF GATE to turn "ON" and remain ON for the setting values shown on the video...you have to adjust this to your own preferences ...to get a certain amount of receive audio between SENDING....so the audio from INVADA TEST TONES gets passed through to the PI's analogue sound card output speaker jack where a stereo audio cable goes from there to 2 AUDIO DERIVED CW KEYING CIRCUIT SWITCHES ...1 for the PTT jack and 1 for the CW jack....the RIGHT CHANNEL of the STEREO AUDIO OUTPUT of the PI goes to the PTT CIRCUIT and the LEFT CHANNEL of the STEREO AUDIO OUTPUT of the PI goes to the CW CIRCUIT

NOTE: the LEFT STEREO CHANNEL audio going to the CW JACK's audio derived cw keying circuit must be delayed - to activate a certain amount of time after the PTT has finished activating....on this system it turned out to be about 12 milliseconds of delay that was needed...provided by another FREE LV2 Linear Delay Line plugin. so TWQRQ's cw audio goes through the CALF GATE and then first to the PTT keying circuit and then 2nd to the CW keying circuit due to this delay line....then the CALF GATE's release time is adjusted ...first to make sure it is not too short of a time that allows chopping of the ends of the cw elements and then second...it is set for the amount of time before releasing the GATE HOLD...according to how much QSK RECIEVE AUDIO you want to hear between sending

Comment by Chuck aa0hw on March 1, 2018 at 5:37pm

Testing and measuring DIT VARIANCE in the ELECRAFT K3S at 40 wpm at FULL QSK(qrq+ enabled)

TEST SETUP:  send 100 dits at 40 wpm keying the K3S CW KEYING JACK with a cw keyer...

The calculated DIT LENGTH is 30 milliseconds at 40 wpm, which the keyer perfectly sends out for every dit...here is a scope view of that keying signal that is used to the key the CW JACK on the K3S...and...as you can see...the cw keyer's keying signal dit is exactly 30 milliseconds long...

The calculated perfectly constructed dit at 40 wpm according to FLdigi is 37 ms long at 7 ms rise & fall time...here is a scope view of the FLdigi dit at 40 wpm showing the measured 37 ms:

NOW, when using the 30 ms(40 wpm keying signal) to key the ELECRAFT K3S,  in FULL QSK(qrq+ enabled), the dits were measured to be erratic and random in length from 35 milliseconds all the way to 40 milliseconds...

So the ELECRAFT K3S is throwing out random lengths of dits, anywhere from 35 ms all the way up to 40 ms...

******************************HOWEVER ************************************

A way to get the ELECRAFT K3S to key cw in perfect timing for every dit is to use AFCW MODE with a specialized CW KEYING system made for the AFCW MODE....like the ZamGate software CW KEYER

When the K3S is keyed by ZamGate using AFCW mode,  all dits measured to be 37 ms...there was not a single dit that varied from the 37 millisecond length....also, when using the ZamGate CW KEYER, you can also vary the rise and fall time edges making the keying harder or software(which is not an available option on the K3S)...you can also increase or decrease the dit length(by time) to make the keying lighter or heavier...

here is what the new ZamGate dit looks like compared to FLdigi:

the ZamGate  CW waveshape has a smoother curvature to its rise and fall time and its spectrum is cleaner


You need to be a member of QRQcw to add comments!

Join QRQcw

© 2022   Created by Chuck aa0hw.   Powered by

Badges  |  Report an Issue  |  Terms of Service