Noise on CW RX with PTT - A solution

Ken - NM9P

I have heard people complain off and on for years about noise on CW receive when PTT is activated, I have done some testing and this is what I have found:

Setting the AGC-T and slice & headphone volumes correctly can make this problem MUCH easier to deal with.  

Symptom....  IF AGC-T is set to the far right, then there is almost as much noise on receive with PTT pressed as there is without PTT.  

BUT....If you adjust AGC-T properly, by beginning to the right and moving it left until the noise just begins to drop.  Then hit PTT you will notice that the noise drops SIGNIFICANTLY.

You can verify this by activating PTT and adjusting AGC-T from about 50 to full right.  Then release PTT.  

It seems that the QSK CW muting may be done by varying the AGC-T setting relative to the control.  The relative difference between the noise you get with PTT pressed vs. PTT open varies significantly with the setting of the AGC-T control.

Those of you who haven't figured out that running AGC-T full bore to the right does NOT give you any more signal than a properly adjusted AGC-T need to experiment a little and learn how to "Drive" this very important control.  With PTT pressed, I can move the AGC-T control from 50 to 100 and the unmuted noise goes from barely audible to unbearably loud (just as loud as on regular Receive with the AGC-T set too high.)  


ADDITIONAL IMPROVEMENTS can be had by changing how you adjust your individual slice volume controls and your speaker & Headphone controls.

I recommend setting individual Slice Volume Controls to 50 to start with.  Then use the main speaker and headphone volume controls to set your local or headphone listening levels.  I run my main speaker control at 100 and then set my Bose Computer speakers to the MAX level I would ever need in my shack.  Then I can turn it down from the rig main speaker control and never touch my Bose speaker controls again..

I find that I have all the headphone audio I need on my Sony broadcast headphones with the rig headphone control at about 50, plus or minus a little.

If I have more than one slice open and need to balance heir relative volume, then it is a simple matter of moving one slice up or down a little from 50%, but I always return it to 50 when finished.

After you readjust your slice volume and main speaker/phone settings, your sidetone will need a one time adjustment to make up for your different settings of the slice and main speaker/phones settings.

I have been running things this way for almost the entire 5 years that I have had the Flex-6500 and have often wondered why some find the RX muting objectionable.  Perhaps this can help others overcome this apparent "flaw" in the 6000 series.

Ken - NM9P

Lawrence Gray

Don't know about other people, but I rarely have AGC-T above 30.  I use headphones exclusively.  When I tested the CW "key-down" noise, AGC-T was around 30.  I don't know how anyone could stand the noise with AGC-T all the way to the right?  I don't think I've ever had it above 50.

K1UO Larry

Excellent post Ken,  you should have mentioned that having the AGC-T cranked all the way to 100 is also "masking" those weak signals with too much gain noise.

Ken - NM9P

Correct, Lawrence. And if you run AGC-T too LOW you will also have a problem because you are reducing the gain too much and then need to compensate for lower RF/AGC gain by increasing the volume. Then when the PTT is activated the background noise is higher because of the increased AF stage gain.

AGC-T adjustment is one of the most powerful receiver controls in the rig.

Ken - NM9P

Correct. When you hit the right sweet spot, weak signals tend to jump out of the noise. Not dramatically, but noticeably.

Lawrence Gray

I'm quite familiar with how to balance the audio gain and AGC-T controls.  The "key down" RX background noise in CW mode is irritating, even with "correct" audio gain and AGC-T settings for the conditions.  While I am sure your work around helps, it isn't a real solution.

Bill -VA3WTB

It is interesting how many big contesters have been using CW on the Flex and have not reported this as not exceptible. A little confusing.

Oxford English

Given that AGC-T is such an important control in SSDR, perhaps Flex could give some thought to the way it is controlled. In many ways it seems to me that AGC-T is similar to, if not the same as the RF gain on an analogue radio. I, like I suspect most operators in days gone by, only used the RF gain control in 1 position - full on, we wanted the most gain possible and most of us used AF gain to reduce of increase volume as we wanted. 

Maybe a sort of secondary control of AGC-T may be possible, not being a programmer, just a user, I do not have the knowledge to know what is and is not possible. But what I do know is that the AGC-T control seems to be the source of much confusion ans misunderstanding. Ken seems to have done a huge amount of investigation which I have certainly found helpful and useful and I thank him for it.

Tim

Rich McCabe

Dumb question. Why PTT on CW?  I am not following

Rich

Ken - NM9P

Rich, some ops use PTT for tx/RX switching with their contest keyers so that it is quiet between elements/characters when sending, but returns to full receive immediately after the message is sent with no breakin Delay. It makes for a bit faster turnaround time.

John - K3MA

Bill,maybe it is because when operating in a contest you have so much noise from other stations that your brain acts as a filter to blank out other noises except the specific station your trying to work.

Rich McCabe

Thank you Ken

Douglas Maxwell

Thanks for the idea Ken. I use AGC-T constantly during a contest to tackle large pile-ups whilst running. For me it’s not a set and forget tool that I could use for a bug workaround. My AGC-T is where it needs to be for the receiver at any given moment, using it to subdue false hiss racket during transmit is not an option . The reason I think that this major bug has not been moaned about more, is that there are no longer any serious cw contesters using Flex, if they are, they are either missing a trick for removing needless delay in tx/rx turnaround time (unlikely) or they have sore ears (highly likely). Flex have made the assumption that most casual CW contest users will use semi-breakin and won’t mind missing the first letter of fast return callers callsigns. This logic seems flawed to me when they then go on to produce Multi-flex v3 targeting serious contest users. My advice to them is get the basics fixed first, like PTT (for goodness sake) before offering complex solutions for a non existent user base (from a CW perspective of course).

Douglas Maxwell

Bill, do you have callsigns for any big contesters that use Flexradios for CW contesting? It would be great to ask them how they have been working around this bug.

Douglas Maxwell

Why have QSK indicator lit when breakin is disabled? In every other radio you have full breakin, semi breakin or PTT options. Translated to Flex language, full breakin is breakin enabled with 0 delay. Semi breakin is breakin with a delay to avoid unkeying between sent elements, letters and words (dependant on speed). PTT is the only option to tx when breakin is disabled. The QSK indicator seems to be lit no matter which of these modes is selected, leading to the question why have it at all? Wouldn’t it be better to extinguish it when breakin is disabled? There seems to be a lot of work required in this area to make the Flex intuitive and usable, currently users get the hint Flex doesn’t understand conventional CW usage.

k3Tim

The QSK lite / operation seems a bit odd...  From the "SmartSDR for Windows Software User's Manual", page 100: paraq 26.5:

"The QSK annunciator located to the left of the TX selector button on each flag indicates that QSK is possible in the selected mode."

If ant2 is selected as TX and ant1 as RX, the QSK lite will extinguish as QSK is not supported in this configuration. I take it the user should be able to tell from the operation of the radio, the QSK is possible but not operation in the current use case. Fair enough...

With Regards,
k3Tim

Ken - NM9P

@Douglas - My AGC-T control is linked to the AUX2 button on my FlexControl Knob.  It is by no means a "set it and forget it" control at my shack. 

However, the two problems I have notices most often when using AGC-T are 1) having it set too HIGH, which removes some of the noise reduction advantage of the control and exacerbates the "noise on CW with PTT" problem.  and 2) Having the AGC-T set too LOW, which requires moving the slice of master volume up.  Then you have the problem of louder signals being too loud while trying to copy weaker signals which have been taken out of the effective AGC range.  This ALSO exacerbates the "noise on CW with PTT" problem, because of the increased audio stage volume.

Far too many people are treating the AGC-T control like the RF Gain control on legacy rigs.  It is not.  But once finding the various sweet spots on either side of the transition slope, it is a very powerful thing.  It shouldn't need adjusting every other contact in order to keep from having your ears blown, or to dig out the weak ones.  But it does require small adjustments as the band changes, or as the antenna is rotated and the base noise floor changes. 

I would never suggest over-reducing the AGC-T control in order to reduce the noise during PTT.  That would be silly.  But I have found that the combination of a good sweet spot, combined with the Audio stage adjustments I have proposed, including setting RX EQ, NR, and APF to enhance the CW signals can reduce the irritability of the problem greatly.

That having been said.  I think that Flex can/should tweak the muting routine in order to make it more effective for those who use PTT in high pressure situations.  Hopefully it can be done without introducing additional timing or audio noise generation problems.  This may be a game of careful trade-offs.

My own suggestion, though I am not a highly skilled programmer, just an experienced operator, is that upon keydown of PTT they could take a reading of the current receiver AGC level, mute the audio, send the code.  Then upon PTT unkeying retrieve the AGC value and load it into the AGC register, then unmute the rig audio and let it settle itself upon a new AGC value based upon current signal strength.  Or perhaps a variation of that theme where before it unmutes the audio it samples the signal level and then sets the AGC, then unmutes the receiver audio. 

I am sure it isn't this simple, but that is why the programmers get the "big bucks!"

Here's hoping for a better solution down the pike.  But also hoping that others can find a way to make it work better until then.

Ken - NM9P

Al_NN4ZZ

Hi Ken, Sounds like another benefit of the Automatic AGC-T IDEA. https://community.flexradio.com/flexradio/topics/option-to-automate-the-agc-t-setting Regards, Al / NN4ZZ

Ken - NM9P

UPDATE:

OK, After doing a lot more testing, here is what I have found.  (in order to save space I will refer to the condition of keying PTT with the rig in CW mode as "CWPTT")

The relative level of background noise with CWPTT is dependent upon the level of the AGC-T control, pure and simple.  The higher the AGC-T level, the more noise remains in the speaker/headphones with CWPTT.  

Lower AGC-T control levels bring a greater difference between keyed and unkeyed PTT.  i.e. when the PTT is keyed, the noise level goes DOWN.  

The relative difference between Normal receive noise and the noise with CWPTT varies greatly with different level settings of the AGC-T control.  You can test this yourself simply by listening to the noise changing as you press and release PTT (I used a footswitch, but it is the same with MOX)  As you do this, vary the AGC-T control between 50 and 100, notice that the closer you get to 100 the less difference there is between normal and CWPTT.  At 100 there is NO difference.  At 50 there is a GREAT difference.  The problem is that sometimes AGC-T requires a setting higher than 50 to be on one side or the other of what I call the "Sweet spot envelope."  

The problem seems to be worse on higher frequency bands (17, 15, 12, 10) which baffled me until I considered that at the present time they are much quieter and require a higher setting of the AGC-T (more to the right) in order to be set at the proper "Sweet Spot." On some bands I found the residual noise with CWPTT very objectionable, (and I think all would agree that I am pretty adept at controlling the AGC-T).

I have not found it as objectionable on the noisy nighttime bands (160-30) because at my location, with the antennas I have, the noise floor is relatively high and requires a lower setting of AGC-T to be in the "Sweet Spot."  This may be different if the operator is using a low noise receiving antenna that requires a higher setting for AGC-T to be in the sweet spot.

I was not able to find any direct correlation between ATU Inline or Bypass and the level of background noise with CWPTT.  EXCEPT for when bypassing the ATU  changes the antenna sensitivity and requires the AGC-T level to be adjusted.

***Things got more complicated when I decided to test with multiple slices on the same or different bands.....

Now the relative noise with CWPTT is dependent not only upon the AGC-T setting and noise floor of the main slice, but upon the same settings of any and all additional slices that are providing audio to the speaker/headphones.  ESPECIALLY if the rig is in Full Duplex Mode.  

Example...if I have everything set so that slice A is perking along with the noise not TOO bad (but perhaps still objectionable to some) But then open slice B, either on another band or as a SPLIT operation, then the noise from Slice B will be added to the noise from Slice A with CWPTT.  I must then carefully readjust slice B's AGC-T, preamp, etc carefully as well to keep that noise from joining the noise from Slice A with CWPTT.

If Slice B is on a quieter band than Slice A, or add a third slice on a quieter band than the first two, and it's AGC-T will need to be higher than for Slice A (or then BOOM, the background noise with CWPTT is now more objectionable as noted in paragraphs above.  At times, VERY objectionable.

THE BOTTOM LINE......

I have come to the conclusion that even though my previous proposal CAN be a temporary and effective workound in certain circumstances, It cannot be a final solution for those who desire to use PTT with CW in order to have more complete control of TX/RX timing and turnaround.

It is now clear to me that another solution needs to be engineered.  In my limited understanding of the way things work in this very complicated machine, I think the problem lies in where the data for setting the residual gain of the receiver while CWPTT is activated.  The linkage between AGC-T control setting and the residual noise in the system with CWPTT seems to be the weak point.

I think there needs to be a better way of muting the receiver with CWPTT.  

BUT, lest anyone start to jump on an "I told you so" bandwagon.  There remains a real, and difficult problem a to how to treat other slices, and Full Duplex situations so that slices that we WANT to hear with CWPTT are not muted unnecessarily or in ways that will reduce the ability of other operators to enjoy THEIR use cases.

I will be filing this report through other venues (alpha team, etc) so that others can test and verify/dispute my results.

I believe that I have been thorough and fair.  And I hope it will be helpful to the engineering team.  Now I make a request of everyone who has been arguing BOTH sides of this issue:

PLEASE....Let's not start another round of arguing about this with more recriminations and name calling and hysteria, or saying "even KEN said it needs to be changed."  (I have no illusion that I or my opinions are that important.)

I have found the folks at FRS and the Alpha team to be great, caring, smart, individuals who work very hard to produce the best product they can.  They must balance a tremendous amount of ideas and workloads to bring about a very difficult software product.  When one thing is changed in one subsystem, it often changes five others in different subsystems.

If or when they decide that they are going to take this issue on, it will take a while to accomplish, because I don't think this is going to be just a simple timing change in the programming.  It may require reworking multiple sections of the audio and TX/RX chain, (and multiple alpha versions as part of OTHER updates, additions, changes, etc.)   So please be patient and courteous.

I have done my best.  Now let's do our best to make this the best radio system it can be, and the best radio system on the market for ALL users.

73.

Ken - NM9P

Douglas Maxwell

Hi Ken, thanks for the practical thoughts. I don't think AGC-T has anything to do with a fix for this, in fact I believe playing with a receiver control for a quick fix to lack of PTT will cause all sorts of undesireable effects in fast QSK CW. It would also complicate future developments of AGC-T. I think latency in audio compared to the application of PTT is the technical problem here. If PTT were a sideband signal to the main receiver data flow and applied in parallel to all data presently in flight throughout the platform data pipeline, this would have the effect of early onset and delayed removal of PTT which would mess up QSK. I think that the complexity lies in the inherent latency of audio against the immediacy requirement of PTT. I think PTT was removed to simplify fixes for performance in QSK functionality approximately 4 years ago.

Ken - NM9P

Douglas,  Just to be clear, I am not proposing AGC-T as a long term solution to the problem.  I was merely identifying the muting linkage to AGC-T as a possible SOURCE of the problem, based upon my observations, and a possible hint to a differently engineered solution.  That is as precise as I can get without knowing exactly how the muting algorithm works. 

It may be that muting with PTT should be done at the audio stages, or it could be that muting or reducing the RF gain via the AGC feedback loop value is the way to go -- whichever doesn't cause popping and clicking and allows the fastest return to the proper gain needed when PTT is removed.  With SDR, It is often all the same, only different numbers on a different line, but not necessarily so.

It is my hope that they will revisit this and engineer a more elegant solution, whatever that may be.

Indeed, one major obstacle, as you have mentioned, is that they will need to find a way to do this without fouling up the non-PTT Fast QSK break-in characteristics.

I have reached the limit of my ability to contribute.  I hope it was helpful.

Ken - NM9P

Lasse Moell

Ken, read my comments on the issue in the other thread. My take is that PTT could set AGC-T to 0... hence muting the used slice. Interesting is that having two slices (split) and muting the audio on the transmit slice will remove audio on the receive only slice during PTT. Would be interesting to hear from developers on why they have it like this.

Ken - NM9P

It doesn't mute it just on the active slice, but the same relationship between AGC-T vs. Muting noise exists on all slices, as far as I have been able to determine.

Douglas Maxwell

Hi Ken,
AGC-T is an algorithm which takes time to take effect. It should work on receive data originating from the ADC. PTT is an immediate clinical on/off non-algorithmic switch. It should work on data originating from the ppt or mox inputs. I would have thought the designers actually have to avoid any affect PTT has on the input to AGC-T, never mind using it to mask audio that shouldn't be there in the first place. Anyway we will never know for sure due to lack of contribution from Flex.

TF4M

The Flex 6600M running stand-alone in CW is unusable because of this issue.  PTT does not mute the receiver on CW at all, it only seems to disconnect the antenna from the radio.

I discovered that setting the ATU on the front panel of the 6600M to ON (lit up green) seems to solve the issue for the moment at least.   

If the ATU is set to BYPASS, the issue returns.

73 Thor, TF4M

LZ2BE

Isn't it easier and logical just to immediately mute RX audio during TX and let the Side tone / Monitor in earphones (if enabled) instead to adjust AGC-T for each QSO? Something similar to what other radios usually do.

0.5-1 sec delay of Audio from PC/Maestro, Side tone / Monitor available only on radio, missing 1-st letter from CW sent text - is it what I have to accept from a high end contest radio?

73 Boyan LZ2BE / LZ8E

TF4M

It would be nice to get an official response on the CW issues.   

As I operate nearly 100% CW, this is a deal breaker for me and I will get rid of the radio if it is not solved.   

The method of only solving issues that attract the highest number of complaints is flawed.

This is a High End radio and as such, this is unacceptable for a basic issue.

73 Thor, TF4M

TF4M

In fairness to FlexRadio, I have to add that my external WinKeyer was compounding the problems I was experiencing on CW with my 6600M.   I had upgraded the WinKeyer to the latest Firmware (3.01.02) and it eventually stopped working altogether.  I downgraded to Firmware 3.00.10 and the WinKeyer is back to normal.  I was getting erratic PTT and missing CW elements and the keyer was losing the USB connection to the computer.  So my comments about the radio being unusable may be ignored.  I apologize for the confusion.