The Noisy K3

or: How I Learned to Stop Worrying and Love the RF Gain

The latest kerfuffle currently brewing on the Elecraft reflector concerns the "Noisy K3 receiver" and, predictably, the commenters have broken down into two opposing factions: the "Me Too!" group is posting comparisons with other receivers that supposedly have less background noise and less listening fatigue, and the "Not Here!" group which swears that their K3 is so quiet that they sometimes think it has been damaged.

Whether any or all of the complaints about background noise are valid, and/or whether these people have radios that are somehow defective or simply misadjusted is beyond my ability to discern. I'm not picking sides here, the guys who think their K3s are noisy may have real issues, and if so I'm confident Elecraft will address these issues as they have done with all others in the past. Perhaps we'll all end up with better, quieter K3s in a few weeks as a result of this discussion.

That said, what really fascinates me is that some of my fellow amateurs apparently believe the RF Gain control to be an archaic, vestigial appendage left over from ancient days of vacuum tubes, and that it has no place or purpose in a modern receiver. To wit:

"Bill has linked to and quoted Eric's paper which quite frankly seems way too complicated to me. IMHO, a modern DSP, microprocessor-controlled receiver should figure all of this stuff out automatically and do it for me ... If the receiver has a properly designed AGC system then there are only two variables that are potentially the operator's responsibility: 1) Preamp On/Off and 2) Attenuator On/Off. With the smarts built into modern radios, why can't the radio do, for example upon band switching, a little routine of turning each of these on and off and measuring the resulting SNR and then setting them accordingly?" -- N7WS

And:

"I'm used to leaving the RF gain wide open on the MkV, leaving the audio gain pretty much alone, and maybe switching between SLOW and FAST occasionally. I don't seem to have any trouble hearing the weak ones under the strong ones. Now I have to fiddle with the RF gain (a small knob hidden amongst the others) while running a pileup. Not enough hands (or enough brains)." -- VE7XF

Seriously? Not enough hands? No offense intended to these guys, fine gentlemen both, but we've long suffered complaints about how the K3 doesn't have the all front panel controls one "needs" at his fingertips to tweak a signal to perfection. Now, a single RF Gain knob is too many controls to tweak in order to deliver a good sounding signal? And what N7WS is asking for falls under the general category of "Artificial Intelligence" -- I think we're going to have to wait a few more decades before we start seeing that offered in consumer electronics products!

The fact is, if you run the K3 or any other radio with RF Gain flat out, the result will be a higher level of background noise than if you "ride" the RF Gain. Whether analog or digital, a receiver's AGC cannot magically discriminate a desired signal from noise. Instead, it will adjust the gain of the IF stages in response to the entirety of what it detects -- that is, signal and noise. The purpose of the RF Gain control is to allow the operator to limit the range of the AGC to some degree in order to compensate for this lack of intelligence. RF Gain is like a transmission in a car, and just as an automatic transmission may work well some or perhaps even most of the time, it doesn't always put the car in the right gear for every road condition. Similarly, the AGC doesn't -- and cannot -- always deliver the optimum results under all band conditions. The RF Gain control is the radio's stick shift. Use it.

Advancements in DSP technology have made it somewhat possible for a processor to detect speech or CW from random noise and perform the requisite voodoo to pass the wanted signal and suppress all else, but this technology still isn't perfect. I don't pretend to understand it all, but lot of math is employed to come up with what is still essentially a "best guess" as to what is, or isn't, wanted. In my experience, it doesn't always guess correctly; operator input is still required. The reason there are so many different possible settings for the Noise Reduction (NR) on the K3 is so you can choose what works best for you. But if you can't be bothered riding the RF Gain a little bit, you surely won't want to mess with the NR parameters.

Is my K3 unusually "noisy?" Honestly, I don't know for sure. I don't think it is; when I first got it I did comparisons with the JST-245, a rig which had a pretty damn quiet receiver. At the time I thought the K3 compared quite favorably. However, these tests were not scientific and the antennas used at the time were fairly crappy. Now I have a slightly better antenna... but alas, no more JST-245.

I did a brief A/B test with one of my NRD-515s on 40m SSB switching between both radios with the same antenna, each feeding identical NVA-515 speakers. It was a hands-down win for the K3 even with NR and AFX turned off. Tweaking both radios for best results, the difference in signal quality and intelligibility was pretty significant. Not exactly a fair fight, though... the NRD-515 is a 25 year old design.

Maybe I'd notice this perceived noisiness more if I had a quiet antenna and QTH. Unfortunately, I contend with a constant S7-8 background noise that I attribute to environmental factors (the QRN of suburban hell) and the fact that I have a vertical antenna, by nature more susceptible to electrical noise. In any case, I've never experienced how the K3 behaves on a quiet band. Under my typical conditions I can tweak the RF Gain, AF Gain and NR to maximize the quality of signals at or above the high noise level while reducing the background hash to a very acceptable level, but there is no single setting of controls that works on all signals. If I optimize for a relatively strong signal, I can reduce the background noise to practically nil but then weaker signals then become much less readable. Tweaking to separate the weakest signals from the noise is possible but that brings up the background noise as well -- all the more so the weaker the signal and the closer it is to the noise floor. Every situation is different, so I'm constantly adjusting RF Gain and other controls to match the conditions much the same way I must downshift my Jeep when I climb a hill or drive through the twisties, or upshift when I reach cruising speed on a highway -- no one gear works well all the time.

Once I find the best RF Gain setting for a particular signal, any noise that is still bothering me is handled extremely well by the NR (which, it cannot be overstated, is the whole purpose of having a NR function in the first place!). The K3's NR has been greatly improved since trusty ol' #216 arrived on my doorstep in January 2008. The original NR worked well for me as an SSB op, but the CW guys were not satisfied; so Elecraft changed the DSP voodoo to accommodate them. All of a sudden, I (and many others) found the NR didn't work as well on SSB as it did in previous firmware versions, it made the output sound too hollow. So after we bitched and moaned about this, the boys in Aptos doubled the number of NR settings from 16 to 32, restoring the SSB-optimized NR settings and giving operators enough variety in NR level and aggression to satisfy everyone. I generally prefer very light NR, so I most often use the least aggressive setting in the 8-1 to 8-3 range to make copy comfortable to my ears; rarely do I use 8-4, but occasionally I will try the 7-x range on stronger signals. With the K3's exceptional noise reduction I find myself adjusting NR more and RF and AF Gain less than I did in the pre-NR days with my JST-245, JST-135 and TS-930.

I hadn't touched the AGC characteristic settings since the K3 was delivered, but today I decided to experiment with the AGC Slope and Threshold parameters just to see what effects they have. I found that the threshold (AGC THR) parameter makes a big difference in the amount of background noise amplified by AGC during periods of no signal. After a few hours of playing around the settings I ultimately settled on (for now) are:

• AGC DCY: SOFT
• AGC HLD: 0.20
• AGC PLS: NOR (default)
• AGC SLP: 012 (default)
• AGC THR: 002
• AGF-F: 100
• AGC-S: 020 (default)

Thus configured, and with noise reduction off, AGC set to Fast, and both AF and RF Gain controls set to 12 o'clock position, I've found my sweet spot for tuning around under typical conditions. When find a station, I may switch on NR and/or adjust the AF or RF Gain until what I hear sounds right to my ears.

What it all boils down to is, using the RF Gain isn't a burden, nor is it rocket science -- it's how I've always operated a receiver. Along with NR, Notch, Width and Shift, it's simply another tool at my disposal to recover the most intelligibility out of a signal. Reaching for the RF Gain comes as naturally to me as it does for the AF Gain or VFO. The idea that some ops feel put out by having to tweak the RF Gain control is beyond incredible to me.