[Elecraft] K3/K3S receiver and high-precision frequency determination?

Fri Apr 24 19:35:02 EDT 2020

In 1969 -- right out of college and in my first EE job, I was tasked to 
evaluate the technical performance of Sylvania's new R-1414 receiver 
that was designed and manufactured to be the replacement for thousands 
of R-390 and R-390A receivers then very widely used by the U.S. military. 
It seemed like an easy low risk assignment for this newly minted 
2nd Lieutentant under the supervision of my very experienced senior EE 
boss, who coincidentally was a PVRC member (W3GN SK). 

A few hundred R-1414s had been built for evaluation mostly by R-390 
users prior to full scale production. R-390 users used it almost 
exclusively for CW Morse, FSK RTTY and SSB voice. Without 
exception all of their user evaluations came in very positive. My boss 
W3GN was very pleased as were his superiors. 

www.myvintagetv.com/posts/r1414.jpg 

Everything was looking good during my technical evaluations until I 
decided -- on my own initiative -- to set up a bit-error test using a 
12 channel orthogonal frequency-division multiplex (OFDM) modem 
(a high tech HF modem in 1969) modulating an SSB signal generator then 
feeding the antenna input of t he R-1414. The audio output of the R-1414 
then fed back into the OFDM modem. I used a nifty new DEC PDP-8 
minicomputer (remember them?) to evaluate the bit-error performance 
of the OFDM signal passing through the R-1414. 

Much to my surprise I could never get better than a 0.1 percent error rate 
no matter how I configured the R-1414 and no matter how ideal the SNR 
was. 

I managed to borrow a brand new Collins 651S-1 receiver from another 
laboratory and -- much to my surprise -- that receiver easily achieved 
orders of magnitude fewer errors, just as you would expect. W3GN's 
superiors were very, very unhappy with my test report... I'm sure my work 
would have been blown off had a ham not been my boss. 

So why couldn't the R-1414 achieve acceptable error rates with OFDM 
signals? To make a long story short, a design decision had been made to 
use automatic frequency control (AFC) to control the R-1414 master oscillator. 
The AFC operated several times per second, causing phase discontinuities 
in the OFDM signal. On the other hand, the Collins 651S-1 used a 
very well designed phase locked loop that caused no apparent bit errors 
at all. The AFC scheme was so fundamental to the design of the R-1414 
that a fix was impossible. 

To make a long story short, the military's plan to purchase thousands of 
of Sylvania R-1414s was scuttled by the technical work of a barely 
21 year old 2nd Lieutenant. R-390s continued to be used for another 
ten years and several thousand Collins 651S-1s and other commercial 
HF receivers were purchased in the interim until the R-390 and other 
interim receivers were finally replaced by thousands of Racal RA6790/GM 
receivers starting in 1980. 

radioaficion.com/HamNews/images/02-2012/ra6790gm_4-20120212191732.jpg 

Thankfully the frequency jumps in the K3 are much less frequent than in 
the R-1414 and apparently do not cause problems for digital signals. 

73 
Frank 
W3LPL 

----- Original Message -----

From: "Dave Cole" <dave at nk7z.net> 
To: donovanf at starpower.net, elecraft at mailman.qth.net 
Sent: Friday, April 24, 2020 9:23:28 PM 
Subject: Re: [Elecraft] K3/K3S receiver and high-precision frequency determination? 

With the TXO on the order of one every few minutes, without on the order 
two a minute or so. 

73, and thanks, 
Dave (NK7Z) 
https://www.nk7z.net 
ARRL Volunteer Examiner 
ARRL Technical Specialist 
ARRL Asst. Director, NW Division, Technical Resources 

On 4/24/20 2:15 PM, donovanf at starpower.net wrote: 
> Each of these frequency jumps causes phase discontinuity that is likely 
> to cause bit errors in during PSK modulation and demodulation. 
> 
> How often do they occur? 
> 
> 73 
> Frank 
> W3LPL 
> 
> ------------------------------------------------------------------------ 
> *From: *"Dave Cole" <dave at nk7z.net> 
> *To: *elecraft at mailman.qth.net 
> *Sent: *Friday, April 24, 2020 9:09:52 PM 
> *Subject: *Re: [Elecraft] K3/K3S receiver and high-precision frequency 
> determination? 
> 
> Thanks for the explanation, I have always wondered why... I'm good with 
> low phase noise, Low phase noise means less splatter from radios! 
> 
> All the folks close to me have purchased K3s, so even though they are 
> close, it is not too bad. I had a friend that lived 700 feet from me, 
> and had a Collins S-Line, including the KW... It was horrid for being 
> far too wide... He switched to a K3, and for the most part things were 
> far better. Prior to the switch, I could not operate on the same band 
> when he operated, after he switched to the K3, I could! 
> 
> I took P3 screen shots of the change from old to new synthesizer, as I 
> changed to the new one, and as he changed to the new one. It is totally 
> obvious how much better the new synthesizer is than the old... See: 
> 
> https://www.nk7z.net/k3/ 
> 
> for the screen shots. 
> 
> 73, and thanks, 
> Dave (NK7Z) 
> https://www.nk7z.net 
> ARRL Volunteer Examiner 
> ARRL Technical Specialist 
> ARRL Asst. Director, NW Division, Technical Resources 
> 
> On 4/24/20 1:03 PM, Frank O'Donnell wrote: 
> > Dave, 
> > 
> > Thanks, that may have been the thread I was remembering. 
> > 
> > While searching old messages, I also came across a post from Ed Cole 
> > KL7UW from June 2018 in which he stated that, even if an external 
> > reference is connected, the K3 frequency "is established by the 
> > [internal] TCXO and the external reference merely performs a periodic 
> > check. ... the EXREF system does not phase-lock the TCXO. It 
> > periodically corrects the error but the TCXO drifts normally between 
> > freg corrections ... . Now if you want to know why Elecraft chose 
> > periodic freq correction over true phase-locking, the answer is low LO 
> > phase noise. Much harder to get low phase noise in a PLL than in a 
> TCXO." 
> > 
> > 73, 
> > 
> > Frank K6FOD 
> > 
> > 
> > On 4/24/20 12:37 PM, Dave Cole wrote: 
> >> Hi, 
> >> 
> >> That was me... I used to use my Icom 756 Pro, to watch the Doppler on 
> >> changes in the ionosphere, and on Meteors. The K3 uses steps to 
> >> correct frequency, while the Icom did not. Analog vs. Digital 
> >> corrections. 
> >> 
> >> Accuracy is not an issue, it is the steps... They destroy the 
> >> continuity of the tracking... I should have some captures if you are 
> >> interested. Email me off list and I'll send a set. 
> >> 
> >> The issue for me, I am looking for changes in frequency, on the order 
> >> of sub-1 hz., not looking to get an exact reading on the frequency, 
> >> just changes in frequency across minutes, not days or hours. 
> >> 
> >> I don't need to be dead on for frequency, but I do need to have a 
> >> straight line that is affected by Doppler, and path changes. I can't 
> >> get that with my K3 without those annoying steps. 
> >> 
> >> Keep in mind we are talking worst case 6 or 10 Hz., and the steps 
> >> appear to be far, far, smaller than that... 
> >> 
> >> So... I have given this aspect of the hobby up. I did just get the 
> >> TXCO installed along with a lot of extras on my K3, and the steps are 
> >> longer, but still there... I will get some older rig that is analog 
> >> in nature and return to that aspect of the hobby then... 
> >> 
> >> I was able to watch echos from aircraft between Eugene, and Portland 
> >> Oregon, (about 100 miles apart), using the KGW TV, (Ch. 2), aural 
> >> carrier as a source. I got the coolest traces of Doppler shifts 
> >> caused by the aircraft movement. Meteors were far more 
> >> interesting... Anyway, it is not an issue of Frequency accuracy, bit 
> >> one of how the rig makes adjustments. I would love to just turn off 
> >> the steps, and accept the drift over longer times. Oh well... Hope 
> >> that helps... It is NOT the K3 having an issue with measuring 
> >> frequency... 
> >> 
> >> 73, and thanks, 
> >> Dave (NK7Z) 
> >> 
> >> On 4/24/20 10:11 AM, Frank O'Donnell wrote: 
> >>> Sometime (probably last year) I remember seeing a discussion of why 
> >>> the K3/K3S receiver isn't suitable to do high-precision frequency 
> >>> measurement tests at the level of hundredths of a Hertz. As I vaguely 
> >>> recall, the receiver is designed so that its frequency actually 
> >>> varies a bit intentionally, and I remember someone posting a plot 
> >>> demonstrating this. However, I'm striking out on finding any of those 
> >>> posts. Can anyone either help with them, or point to any other 
> >>> discussion/explanation on this? 
> >>> 
> >>> Thanks much, 
> >>> 
> >>> Frank K6FOD 
> > 
> > ______________________________________________________________ 
> > Elecraft mailing list 
> > Home: http://mailman.qth.net/mailman/listinfo/elecraft 
> > Help: http://mailman.qth.net/mmfaq.htm 
> > Post: mailto:Elecraft at mailman.qth.net 
> > 
> > This list hosted by: http://www.qsl.net 
> > Please help support this email list: http://www.qsl.net/donate.html 
> > Message delivered to dave at nk7z.net 
> ______________________________________________________________ 
> Elecraft mailing list 
> Home: http://mailman.qth.net/mailman/listinfo/elecraft 
> Help: http://mailman.qth.net/mmfaq.htm 
> Post: mailto:Elecraft at mailman.qth.net 
> 
> This list hosted by: http://www.qsl.net 
> Please help support this email list: http://www.qsl.net/donate.html 
> Message delivered to donovanf at starpower.net 
> 