Also, the current NB really pinches the audio, seems far more broadbanded than it should be. Don't even get me started on the slider...seems more like a decoration than a setting! :)
Noise Blanking is an area where direct sampling radios can shine because of the wide band sampling capabilities. That may require potentially significant work on the FPGA side so you will have to be patient.
With regard to Noise Reduction (NR), properly setting the AGC-T threshold for optimal signal to noise ratio will in my opinion always beat any NR algorithm ever created. NR algorithms use non-linear functions (LMS) to auto-correlate the desired signal. Given their adaptive non-linear nature, they will always color the desired signal in some way. Different types of noise will affect NR in different ways with varying results. On the other hand, setting the AGC-T gain properly will give you the best SNR without distorting or coloring the signal at all. My personal opinion is that NR will always be inferior to a properly adjusted AGC-T on any rig ever made.
I know you and your staff have a very full plate of things on which to work. We all wish you the best of luck in accomplishing your goals. But your response above sounds like improvements to noise blanking are in the far-distant future, and not on the published roadmap. That's a real surprise and deep disappointment to those who were hoping that the "DSP Enhancements" released in V1.1 in January 2014 would have matched the PowerSDR levels of effectiveness. Since the former level of noise blanking performance has not been carried forward, the "DSP Enhancements" bullet point on the roadmap seems unfulfilled.
I wish Flex would reconsider the apparent decision that the NB performance is "good enough" in its "enhanced" V1.1 form. Particularly for 6-meter and up operators, impulse and power-line noise are very unpleasant facts of life that are in great need of mitigation in the receiver. Flex has a chance to "own" the radio market for noise-mitigation receivers if a superior NB arrangement is developed. As it is, I'll be relying on my Icom 7600 for what relief I can get from the "buzz".
I do not hear him saying that NB & NR are "good enough" as currently implemented.
I hear him saying that NB (Noise Blanker) is something that can be VERY effective on the 6000 series but will take a LOT of effort programming the FPGA (internal processors) to maximize it.
So it won't be in 1.3, but I am hoping that it may start showing up in 1.4???
Regarding NR - (Noise Reduction) I think Gerald is correct in that finding the AGC-T "Sweet Spot" is the first step in noise reduction. It is amazing how the signals can jump out of the noise when you carefully adjust AGC-T!
That being said, however, I think that there is still much that can be done after that in improving the DSP performance of the Noise Reduction routine. Yes, it will ALWAYS color the audio to some degree. I doubt that you can effectively remove large amounts of high frequency noise without sacrificing some desired high frequency response, or adding latency or some distortion.
I am much more interested in effective Noise Blanking performance than NR at this point. I eagerly anticipate additional improvements to the NB in version 1.4 or 1.5, especially on my 6 meter operations..
I use the NR circuit mainly to reduce listening fatigue on noisy bands, and sacrifice the high fidelity to that end. I also make good use of the RX EQ to make the voice articulation frequencies, or CW tones jump out of the noise (after properly adjusting the ACG-T). I have found that this and the APF on CW can be extremely effective. I am confident that the Folks at FRS will deliver. But, I do encourage them to add this as a bullet point on the roadmap as soon as possible so that we will all be confident in the process. If things go as they have in the past, they may put it on for v.1.5 and deliver it in v.1.4!
By the way, I seldom use ANF any more, now that we have TNF. TNF is much more effective and doesn't color the audio nearly as much if you don't make the notch too wide.
"If you want to vote specific features or improvements up, feel free to enter them as an idea and vote them up. We have to weigh these requests against road map commitments that we have already made public and intend to keep."
So if we want NB's status improved do what he said.
My personal preference is to improve any noise abatement we can. I would prefer to be able to hear signals 6dB under the noise far more than I want the ability to operate my radio from the parking lot at the WalMart. I realize that satisfaction of that desire however requires a entirely different kind of engineering commitment than WAN. In addition to noise abatement things like diversity and steering signals using a coherent array of antennas may help. The Flex family has incredible potential in this arena.
I have been playing with BPSK-31 and I do have a noise problem though its not an extreme problem like some have. BPSK=31 does not require any particular fidelity it just has to be readable. The same is true for CW. So for me signal distortion is less significant than actually being able to hear (see) and demodulate the signal either in my digi program or in my brain. Also loosing the buzz very much improves the operating experience in terms of fatigue.
Here is a pic of my digi program with the NB off and then on.
Clearly the ability to pick out weak signals is greatly enhanced with NB on. I always run my radio at best AGC-T.
Gerald you addressed this in an earlier post of mine, however you drifted off into the area of NOISE REDUCTION. This NR feature is useless for me in my EME work, and I see no advantage of it in my CW operation. Perhaps SSB is where it shines, but that is a seldom used mode for me..
Excuse me for being so harsh, however we have waited through several releases with no improvement of NB. I am so passionate about this that I hope you will hold release 1.4 until an improvement is made for noise blanking. WHAT SAY YOU?
Gerald - K5SDR, Official R
As I stated in the last Flex Insider, our goal is to up the game in noise and interference mitigation in v1.5. There are many types and sources of noise so a single algorithm will not address all types of noise and operating modes. We are not going for parity but to exceed what has been done before on any radio. As I also said, this is a "science project." We have a number of ideas that we want to try that have not been used in ham radio radio before. We have a lot of options in the FPGA that are not available in legacy architectures.
Regarding the current noise functions, you may find that you have AGC-T set too high. You should adjust AGC-T for best SNR and then turn NR or NB on. I find that I don't like NR on any manufacturer's radio I have ever used. LMS based noise reduction algorithms always "color" the signal in my ears. I much prefer to adjust AGC-T for best SNR since the signal is not distorted at all with this method.
Regarding remote, LAN in v1.4 is a prerequisite for WAN. We have to get that right first. We remain committed to v2.0 WAN in 2015. Doing WAN right is non trivial.
Others who have noise that is not wideband in nature but more pulses find the noise blanker not effective (plasma, router interference, etc). But for powerline and ignotion type noise, most people are having very good results.
Yes. That would be a good idea. Seems the previous NB was effective for some against electric fence noise. Being able to select either one would cover more types of noise sources out there.
The primary powerline arcing here happened on hot sunny days. Now we are in autumn and it is cooler I do not have the issue (or the utility company FINALLY solved it...)
Haven't heard you on 40 lately. Will keep an ear open.