apf gain control needed. 0-10db adjustable would be great.

  • 5
  • Idea
  • Updated 4 years ago
Icom just came out with v2.00 firmware for the IC7600.  One added feature besides the waterfall was the addition of a gain control for the APF 0-6db user adjustable.   On the IC7600 setting the gain at 6db really brings signals out of the noise floor similar to pre v1.2 SmartSDR.  Since SmartSDR on the Flex 6500 early version change around 1.2 the apf gain is almost non-existent / noticable when APF is initiated.  The earlier version had a nice gain peak that really brought weak cw signals out of the noise floor.  Please strongly consider adding a gain control for APF of 0-10db to help optimize weak signal reception.  
Photo of Carl K5HK

Carl K5HK

  • 39 Posts
  • 9 Reply Likes

Posted 4 years ago

  • 5
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1853 Posts
  • 672 Reply Likes
Hi Carl,
Interesting that Icom has that now.   My Yaesu FTDX900 also has a control for the gain.     

Agree with your assessment and here is a link to what you are asking for on the Flex---  it has 23 votes so far.  More votes can't hurt.....

https://community.flexradio.com/flexradio/topics/ssdr_v1_2_apf_could_use_more_gain_it_seems_diminish...

Regards, Al / NN4ZZ  
al (at) nn4zz (dot) com
Photo of km9r.mike

km9r.mike

  • 425 Posts
  • 62 Reply Likes
# 25
Photo of Walt - KZ1F

Walt - KZ1F

  • 3040 Posts
  • 645 Reply Likes
Note to all: Don't click it twice, it will erase your vote. So watch the counter when you 'like' it as it might drop if you've already 'liked' it in the past.
Photo of Steve N4LQ

Steve N4LQ

  • 568 Posts
  • 92 Reply Likes
Interesting that new firmware release for the IC-7600 added a gain increase to their APF. It's probably partly psychological since one could just turn up the volume but somehow it helps. 
Photo of Carl K5HK

Carl K5HK

  • 39 Posts
  • 9 Reply Likes
Steve,  No it is more than psychological.....I can copy better than just turning up the volume because it peaks the signal out of the noise.  Volume/AF alone brings up noise with the tone peak alas.   SmartSDR did a nice job on that briefly before v1.2 but they changed the way it and other functions were implimented at v1.2 and lost that gain peak on APF.   I don't know whether they could just go back to old way of implementing APF and solve the problem or whether a gain control would even help in a APf function but it sure works good on the IC7600.
Photo of W5UN_Dave

W5UN_Dave

  • 316 Posts
  • 30 Reply Likes
I agree with this completely. The APF was a great addition (thanks for listening Gerald). My K3 APF severely out performs the 6700 APF. I know the Flex boys can beat this! I still go back to my K3 when doing very weak signal digging on 160 meters, or on any band for that matter.
Photo of Carl K5HK

Carl K5HK

  • 39 Posts
  • 9 Reply Likes
Dave,

Don't see anything from Gerald in this thread.  Did you contact him via email or phone and please give more detail as to what he said about how they are addressing the APF problems.  Thanks/73, Carl/K5HK
Photo of Steve - N5AC

Steve - N5AC, VP Engineering / CTO

  • 1057 Posts
  • 1097 Reply Likes
There is currently a gain control for APF and it ranges from 3-12dB of gain -- BUT the output of the APF goes to the AGC to prevent overload and level the output.  The net effect is that it lowers the noise around the APF giving you better SNR without raising the volume at the same time.  To us, this seems to make sense.  However if you want to APF to also raise the volume, it's not going to do what you want -- you would have to increase the APF and raise the volume to achieve that.  I would be interested on comments based on this information and your experience with APF.
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1853 Posts
  • 672 Reply Likes
Steve,
I agree with Carl, Paul, Ken and others that before V1.2 the CW signals popped above the noise more.  Turning up the volume does not have the same effect.  Something changed.  

My understanding from some previous posts is the current control affects the Q or shape/narrowness of the peaking.   Other radios offer this capability to make it very narrow down to about 5 hz.   Maybe more control over this would be helpful also. From the manual:




The suggestion to expose the gain control was to let us tailor it to our preference like many of the others implementations of APF.   But if you are saying the existing slider is the gain (not bandwidth), maybe we just need a wider range of gain control.  

https://community.flexradio.com/flexradio/topics/ssdr_v1_2_apf_could_use_more_gain_it_seems_diminish...



Regards, Al / NN4ZZ  
al (at) nn4zz (dot) com
(Edited)
Photo of W5UN_Dave

W5UN_Dave

  • 316 Posts
  • 30 Reply Likes
Steve, N5AC, with all due respect, the way to know if the Flex APF is working well is to hook up your K3 and practice for a while doing VERY weak signal reception, then use the K3 APF to familiarize yourself of how a very good APF works. Do this under very weak signal conditions with diversity enabled. You will be amazed how much better the K3 APF is at peaking up those signals which weren't even readable before applying  and tuning the K3 APF.

Right now, the K3 APF is my standard.
Photo of Carl K5HK

Carl K5HK

  • 39 Posts
  • 9 Reply Likes
Steve,

Thanks for you response.  On further testing I notice that when I adjust the single APF slider on 100hz cw bandwidth I can hear a distinct peak on weak signals when apf set 80-90%.  I think it would still be useful to have seperate gain control for APF however although operating results are much improved now that I have set apf to 80-90 percent verses lower setttings.  I assume gain is increased with Q by that control.  

What would be the bandwidth/gain range correlated to 80-90% apf setting when CW bandwidth at 100hz?  I appreciate the clarification and hope my input helpful to others.  It actually brings weak signals out of the background now like pre 1.2v did.

73, Carl K5HK
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1853 Posts
  • 672 Reply Likes
Hi Carl,
I also keep the current slider set to about 90.   I don't see a lot of variation as I move it up and down but it does seem better closer to the high end.  I leave APF on since I don't see any negative effects and it does work, but I would like to see it work better.  
IMHO It's not quite as good as some other radios.   But I think with more user control over the gain and Q (or bandwidth) it could be the best.

Regards, Al / NN4ZZ  
al (at) nn4zz (dot) com
Photo of Carl K5HK

Carl K5HK

  • 39 Posts
  • 9 Reply Likes
My feeling too Al and why I have kept reposing this question several times since v1.2 made APF worse.   I am sure Flex will continue to refine it but if we don't keep pointing it out with respect to other rigs apf functions why would they worry.
Photo of Steve N4LQ

Steve N4LQ

  • 568 Posts
  • 92 Reply Likes
Steve

What I do here is feed the audio into an external audio filter (Autek QF-1A). This makes CW stand above the noise plus it makes the sidetone blend beautifully with the received signal for cleaner QSK. I also use a speaker which is near resonant at 600 hz. 

Also this eliminates the need to "ride" the AGC-T as if it were a 1940's receiver. The external filter cleans up the mushy overload. Hey...If AGC-T is "AGC Threshold" then how come it still works with AGC turned off? Or is it really a threshold control?

I can't think of one radio that filters everything like this but it seems far superior to anything I've tried. 

BTW: All audio filters introduce their own noise. I tried many of the SCAF design but all were too noisy. So far this old Autek wins.

Steve N4LQ
Photo of W5UN_Dave

W5UN_Dave

  • 316 Posts
  • 30 Reply Likes
I used an Autek QF-1A analog peaking filter for years during the early part (back in the 1980's and 90's) of my 2 meter CW EME work. I could not have made more than half as many contacts without it.

That being said, there is no reason that a peaking filter just as good or better than the QF1A could not be built into the SDR radio. The secret is to permit  the APF to have a little ringing at it's deepest setting, which tends to cause the signal to "jump out" of the background noise. (one time that ringing actually helps). The other thing is to permit the user some means of easily controlling audio pitch (best done with a knob in one hz increments).
Photo of Carl K5HK

Carl K5HK

  • 39 Posts
  • 9 Reply Likes
Dave,

Yes I could do the same thing.  The reason I keep revisiting this here is that SmartSDR is perfectly capable of doing it all in the rig and anything I had is a ton of external stuff to tweak and all the attended wires, plugs and cables asking for RFI problem.   One of the big reason I choose flex is because they are really trying to do all this in software.  I just feel they have in the past and can do better on this issue.
Photo of Tim - W4TME

Tim - W4TME, Customer Experience Manager

  • 9198 Posts
  • 3558 Reply Likes
If AGC-T is "AGC Threshold" then how come it still works with AGC turned off? Or is it really a threshold control?

A little light reading of the SmartSDR Software User's Guide answers this question:

When AGC is set to OFF, this means that a fixed amount of gain determined by the AGC-T setting will be applied to both fast and slow signals regardless of their level.  The more you adjust the AGC-T to the right, the more gain is applied and the louder the signal and noise will be.  The benefits of increasing SNR with AGC will be lost in this mode.  The other down side of turning AGC off is that the operator must to pay attention to the 'RF Gain' manually to avoid distortion due to overload by strong signals.  
Photo of Steve N4LQ

Steve N4LQ

  • 568 Posts
  • 92 Reply Likes
RE: "The other down side of turning AGC off is that the operator must to pay attention to the 'RF Gain' manually to avoid distortion due to overload by strong signals. "

The same downside applies when the AGC is ON. To me it seems like that unlabeled slider control is actually an RF gain control no matter if AGC is off or on. I think that explains why the so called AGC-T works either way and that's because AGC-T isn't actually AGC-T. One can have AGC turned on and still be BLASTED by strong signals but maybe worse.
Photo of Ken - NM9P

Ken - NM9P

  • 4239 Posts
  • 1351 Reply Likes
The only way I have ever experienced being blasted by strong signals with AGC ON was if I had AGC turned WAY too far to the left (reducing sensitivity on all but the strongest signals) and turned the slice volume too high in order to compensate. Then when a signal was finally strong enough to get over the threshold, it played at the "normal" boosted volume., which was then blasting.

Overdoing it on the AGC-T, very far below the knee or sweet spot, requires excessive audio gain settings.
Photo of Steve N4LQ

Steve N4LQ

  • 568 Posts
  • 92 Reply Likes
Yea but a $4500 rig shouldn't have a "sweet spot" or at least it should find it itself. IMHO
Photo of Steve N4LQ

Steve N4LQ

  • 568 Posts
  • 92 Reply Likes
Someone explain how increasing the slice volume is different than increasing APF volume.
Photo of Steve N4LQ

Steve N4LQ

  • 568 Posts
  • 92 Reply Likes
But is it really adjusting the gain of the filter or of the amplifier? Given the difference between a filter and an amplifier I suspect the latter. 
Photo of Jay / NO5J

Jay / NO5J

  • 1674 Posts
  • 263 Reply Likes
My understanding, and I'm probably wrong, is that the APF function is less a filter and more an equalizer. It boosts or cuts the audio frequency thats been selected. My guess is its similar to a TNF filter as in it only acts on a narrow slice of the AF passband, and within that slice it finds the signal applys whatever gain is set on the gain slider and cuts the gain of everything else in that narrow slice. in other words it filters out everything but the signal and enhances the signal itself. Its a very selective graphic equalizer. It finds the peak signal inside the narrow slice, centers on the peak, trys to follow it and only apply the boost to the peak and cuts everything but the peak.
Sort of a reverse ANF. 
I repeat. My understanding is a guess.

Jay - NO5J    
(Edited)
Photo of Steve N4LQ

Steve N4LQ

  • 568 Posts
  • 92 Reply Likes
There was a time when I understood my radios but these are strictly appliances. 
Photo of Jay / NO5J

Jay / NO5J

  • 1674 Posts
  • 263 Reply Likes
Right. But it's a software defined appliance. It's only doing what it's programmed to do. When it's not working as intended, it might need a better program. You will know it's time to sell, when those better versions stop, They don't have to stop. Figuring out how to emulate an appliance in software is hard and slow. Getting it exactly correct is only going to happen once. If that happens, why even consider selling the perfect appliance, that it's finally become? That might be profitable, others will want a perfect appliance too. A real sellers market, you set the price, and the buyer won't even complain about the shipping cost. Sometimes, I wish I was software defined, and that I could code myself a better life. But I'm stuck with reality.
 
The weird is 10 over S9, I think.
Jay - NO5J        
Photo of Walt - KZ1F

Walt - KZ1F

  • 3040 Posts
  • 645 Reply Likes
@Jay - reality is what you make it. To do otherwise is to exist in someone else's reality. We are all our own self fullfilling prophecy.
(Edited)
Photo of Steve - N5AC

Steve - N5AC, VP Engineering / CTO

  • 1057 Posts
  • 1097 Reply Likes
Let me take another stab at this and see if it helps.  In a digital world, signals are time sampled. The samples exist in a number space that is either floating point or fixed point.  Regardless of the floating/fixed question, there are a fixed number of bits that represent the signal.  In a fixed point system, this is the number of bits in the number, like 16, 24, etc.  In a floating point system, most engineers use the IEEE-754 standard because it is implemented in most modern processors.  In floating point, there is amazing total dynamic range, but lower instantaneous dynamic range.  For example, with a single precision floating point number, we can represent 10^-126 in one sample and 10^127 in the next sample, for a total dynamic range of over 2,500dB.  But in radios, the signals occur at the same time so it is all about the bits in the mantissa (part that holds the numbers and not the exponent).  In a single precision IEEE-754 float there are 23-bits of mantissa. If we figure on about 3dB per bit, this is only 69-72dB of dynamic range.

So why does this matter talking about APF?  When we are placing a signal inside that number space, we shoot for the best SNR which means "price is right rules" -- get as close to the largest number you can, but don't go over the largest number that can be represented in the number space.  Why?  The signal would clip; clipping causes distortion; distortion is bad.  So we have to guess for a normal volume, how much louder might signals get temporarily so we are sure we have the headroom to accommodate them when this happens.  Then we build the system around that nominal level.

Now, we add APF or some other filter.  As has been pointed out, arguably the best way to make filters work in a system is to make them unity gain in the passband.  This way, you can insert or remove filters and the overall volume of the system doesn't change.  But we have an issue in that operators tend not to always recognize the benefit in increased SNR (what you really want) and instead seem to want the volume to increase when filters that are billed as helping you hear a signal are engaged.  So we gave the APF some gain.  If we do not put AGC after APF, then you could raise the level of the APF with its gain past the number space limits and into distortion.  Then we would hear "when I turn APF all the way up it sounds distorted."  

Also there are calls for higher-Q in APF, but the Q is already reasonably large.  We would argue that having some noise in the background is useful as a reference to make adjustments and copy the CW.  If you engage a 50Hz bandpass filter and crank the APF all the way up, I would argue that you're not doing yourself any favors -- you are limiting all sound to a 50Hz region which will make any noise in that region sound exactly like the tone you are listening for.  APF on top of this further exacerbates the problem by limiting the tone to a very narrow region of frequencies.  Any noise that makes it to the AGC will, by definition, only reside in this narrow frequency band and will sound exactly like random CW on the frequency you are listening for.

Sometimes when we hear customer feedback it makes complete sense, we realize how we could do something different, we implement it and everyone is happy.  In this case, it feels to me like there is a gap between perception of operation (how the system feels to you when you operate it -- does it do what you want and expect), how the system actually works and everyone's understanding of what to do next (including ours).  You guys know we're happy to make changes that will result in better operation for most folks.  This is what we're all about.  But if we add more gain to APF, it's going to get squashed in the AGC, as it should.  If you don't like this answer, you need to turn down your AGC and turn up your volume -- you will have lower SNR overall, but APF will make the volume get louder when you turn it on.  If we place APF after AGC, you will be able to overload the digital range of the system or we eliminate the gain in APF and then we're back to "it doesn't seem to do anything."
Photo of Walt - KZ1F

Walt - KZ1F

  • 3040 Posts
  • 645 Reply Likes
Steve, that sounded like a great explanation. I say that because I kinda sorta understood it. However, as I don't have much of a horse in this race, vis-a-vis, the EE of making a signal better, the issue has been mentioned a couple of times in this thread about how it worked better pre-SSDR 1.2. Could I maybe impose upon you to explain what changed and why? I could be wrong and I hate speaking for others but it sounded to me like some were saying it was better then.

Thanks,
Walt
Photo of Steve N4LQ

Steve N4LQ

  • 568 Posts
  • 92 Reply Likes
Steve
Now since we're on the subject. Could you explain why AGC-T has to be manual? To the outsider it would seem easy to have the software decide the optimum setting instead of the user fiddling around looking for "sweet spots" as if this were a Hammarlund HQ-129. The whole thing seems so archaic.  I know other rigs use hardware AGC loops. Is there an aversion to that for some reason?
Steve N4LQ
Photo of W5UN_Dave

W5UN_Dave

  • 316 Posts
  • 30 Reply Likes
N5AC Steve, I can only speak from what I experience by comparison: my K3 APF severely out performs the 6700 APF. I believe the K3 APF is also done in software.

I appreciate and commend what you guys at Flex are doing. Keep up the good work.
Photo of Steve N4LQ

Steve N4LQ

  • 568 Posts
  • 92 Reply Likes
As for the K3's APF. Having owned a K3 for 7 years I'll make the claim that its APF sounds better because without it the radio is noisier and APF has more to work with. It's just a matter of contrast.
 Face it....If there were no noise getting past the normal filters, APF would be redundant a fact proven by the Flex. The much used term "signals seem to pop out of the noise" really applies when there is noise to "pop out of". 
Photo of Al / NN4ZZ

Al / NN4ZZ

  • 1853 Posts
  • 672 Reply Likes
N5AC Steve,
Does the FLEX  APF operate in the audio domain where I assume the RX equalizer operates?   My assumption is that APF is essentially a super narrow audio equalizer.  It would boost the audio signal at the sidetone audio frequency and within the bandwidth selected.  It would attenuate the audio frequencies outside the selected audio bandwidth.  And providing more control over the boosted bandwidth and gain would allow us to tailor the performance to our liking.   (essentially like the Autek QF-1A and other external audio filters).   Is this a correct assumption or does the APF operate at a different stage in the FLEX 6000?

Dave,
Do you know where in the signal chain the APF is located in the K3?


I guess the big question is:  Can the APF be better or is the current performance the best we can expect to see on the FLEX? 


Regards, Al / NN4ZZ  
al (at) nn4zz (dot) com
Photo of W5UN_Dave

W5UN_Dave

  • 316 Posts
  • 30 Reply Likes
Steve, my experience with the K3 APF was over the last several years on 160 meter DXing where signals can be challenging. I would try with the 6700, and then always go back to the K3, where I could dig the signals out better. The two radios are side by side with antenna switching and separate keys. Maybe it's just me, but I know what works best here.

Let me finish by saying I want my 6700 APF to work as well or better than the K3 APF. As far as noise, well yes, 160 has plenty of that.
Photo of Steve - N5AC

Steve - N5AC, VP Engineering / CTO

  • 1057 Posts
  • 1097 Reply Likes
Would you be willing to do a video recording of both on the same program material and show us what the differences are?  We'd be happy to do an analysis and see if we can't make things better.
Photo of Steve N4LQ

Steve N4LQ

  • 568 Posts
  • 92 Reply Likes
That could end up like my video comparing the K3 to the 2B. https://www.youtube.com/watch?v=Z9mAzxkuk18
Photo of Lee

Lee, Elmer

  • 680 Posts
  • 287 Reply Likes
I did a recording and I think the APF works fine.  If you want the post AGC effect simply turn off the AGC and set it to a level consistent with the volume you want.   OMG run without AGC!!!  Yea you get blasted once in a while... so what.  There are many ways to make this radio do what ever you want.

https://www.youtube.com/watch?v=cuitoTHd8Vg

73  W9OY
Photo of W5UN_Dave

W5UN_Dave

  • 316 Posts
  • 30 Reply Likes
Steve, I just again sat down to do a side by side comparison of APF's. and found by using W9OY's method, the Flex APF does a pretty decent job. I now see where my real issue is, ease of adjustment. Finding an extremely weak 40 meter signal (no waterfall detection) out of Europe in mid afternoon here I again did the comparison. It was really easy to peak the signal on the K3 with the APF engaged, by using the shift knob to find the sweet spot. What is happening on the K3 is that the knob changes the AUDIO pitch, not the frequency. This is what is lacking on the Flex: the ability to change the audio pitch. Am I wrong here?  I think the question was asked earlier by someone about whether the Flex APF is an audio filter or is it something else. If it is an audio filter, and if a button on the Flex Control were to be programmed to allow the Flex Knob to adjust that pitch in 5hz increments, I think that would be perfection for me!
Photo of Steve N4LQ

Steve N4LQ

  • 568 Posts
  • 92 Reply Likes
To aid in understanding Dave's post here is a snip from the K3 manual:

APF and Dual-Passband CW Filtering In CW mode, DUAL PB is used to turn on either an Audio Peaking Filter (APF) or Dual-Passband Filtering. Both are described below. CONFIG: DUAL PB selects the desired function. APF can make a dramatic difference in copy of weak signals buried in noise. When APF is turned on, SHIFT fine-tunes the center pitch, and NOR M resets the pitch to that of your sidetone. The filter graphic will be similar to that shown below. Dual-Passband Filtering (Dual PB) sets up a 150- Hz-wide focus filter, set within a wider context bandwidth that is attenuated by about 20 dB. This lets you hear off-frequency CW signals, which may be useful if you’re calling CQ, running a net, etc. WIDT H varies the width of the context filter. The filter graphic appears as above.