Other side band not properly suppressed

Sergey R5AU

My friends (F6500 owners) reported with not enough suppression of the other side band in SSB, i have not yet tested on my F6700, but in any way issue exist and need to be investigated, is anybody else can check it ?  at least 2 versions of the SSDR involved (has proved)  1.9.13 and 1.10.16
 Verification was done in FDX mode, RX not overloaded, unsufficient supression in low frequency bandwith. first screeshot with PROC OFF, second with PROC = NOR



Pink Noise  PEP   = -10dB

David Livingston

A low cut of 50 is unusually low. Most people use from 100 to 300. I use 200 and do not have this issue on my 6700. For dx and pile ups I use a low cut of 300 and a high cut of 2400. Put the energy where people can hear it.

I would also like to see what you get without your external audio equipment...!!!

Sergey R5AU

Well David, no external equipment are involved in this test but only Rose Noise generator and Low Cut should not play here, example you running in DIGI and should also cut with 200Hz ? No way, this is not filtering superheterodyne but DDC and opposite side band must be well suppressed from >0Hz

Bill -VA3WTB

I know it is not a Flex problem, I have seen many Flex on the air and the sides of their signal drops perfectly down into the noise floor.

Sergey R5AU

Bill, I keep my finger crossed with all perfect on Flex side , I'll be back tomorrow in my shack and able to look into my 6700, however still open question: 1. What are visible on screen shots above ? 2. Should we see or not artefacts with opposite side band on spectroscope ? 3. Can spectroscope be trustable for measure ? My question to all - pls look into you radio on 2-3 min on such issue to report here. May be that observation incorrect at all and correct statement completely different .

Tim - W4TME

Sergey,

We can’t duplicate this behavior here on a 6300.  The attached pics show both Pink Noise and Individual tones into the rig in Full Duplex mode with essentially the same bandwidth and setup as used by you.  We set the low freq cutoff at 50 Hz.

Observations: 

1.  With Pink Noise input, the opposite sideband suppression was nearly 80 dBm referenced to the peak in-band level

2.  With individual tones into the rig, the opposite sideband was in excess of 80 dBm referenced to the peak in-band level

3.  Your Pink Noise spectrum is suspect as is looks sinusoidal rather than a pseudo-random 1/f distribution.  With the Flex audio chain being essentially flat down to DC, and the 1/f characteristic of Pink Noise,  you may be inadvertently overdriving the radio with lots of low-frequency energy, thus the excessive opposite sideband energy.


Pink Noise:


Tone:

 

Sergey R5AU

So , Tim, just reproduced it on my F6700 with playback tones into player with DAX TX output:

issue in zone below 300hz


1.  2 tones  50hz+150hz, good hearable on slice B




2. to compare (to check measure compatibility) IMD test with playback 1200Hz+1600HZ





Test in the initial post executed by George UA0IHZ 

last post with my radio  F6700, as you can see radio so far from overdriving here
but low freq tones are not suppressed 



SSDR 1.10.16

Don Baughman

Sergey-  I think you are confusing two different things here, and your screenshots demonstrate them.  The first is "opposite sideband rejection" and is primarily determined by the software algorithm used in the Flex.  It produces nearly perfect cancellation of the opposite sideband as demonstrated by Tim's response (80 db or greater).   The second is "transmit intermodulation distortion" and is a result of the mixing action of two tones introduced into the transmitter.  These tones are usually chosen to be non-harmonically related  (700 Hz and 1900 Hz for example)  and are a measure of the linearity of the transmitter's RF chain.  Your last screenshots are examples of this mixing (IMD) action. 

What is important in both of these areas (opposite sideband rejection and Transmit IMD) is that they are both referenced  (in -db) to their peak values in the desired region.  It's not clear what is going on in your original screenshots, but I suspect it is related to your test signal (Pink Noise) or input levels that are not accurately measured.  A good transmitter will show opposite sideband rejection of -60 db or greater.  Any Flex 6000 I have ever seen on the air or owned and tested have been significantly in excess of -60 db..

As far as the IMD results that your last screenshots demonstrate, an average Amateur transceiver will range from the mid -20 db to -35db range for the 3rd order products.  Higher-order products are even further reduced  to the -50 to -70 db range.  The Flex 6000s I have owned (6300 and 6700) have both been in the range or -40 db for the 3rd order products.

In summary, I can't say for certain, but I think your testing method is providing you with erroneous results.

73
Don K7MX

Ken - NM9P

What I have noticed , Sergei, is that in your first pair of graphs, you have a low frequency bass spike somewhere between 50 and 100 Hz that is about 20 dB louder than the rest of your signal. My guess is that even though your meter is not reading in the "red" that the hyper bass boost is overdriving your audio circuits and causing some IMD that is showing up below your USB signal. I would suggest reducing your bass boost, perhaps even moving your low-cut up to 65-100, and retesting. Whenever I see a SSB signal with that much of a low frequency spike, I can tell just by looking at the panadapter that he is going to be difficult to understand unless he is very strong, because the low bass note drones on like a bagpipe in my ears. Many times I am forced to shift my RX filter up to about 150 before he is intelligible at all.

Sergey R5AU

Don and Ken,
i am appreciate you notes and opinions however results can be easily duplicated.
  In my last post above i mix 2 measures because i also not believed at beginning such pity results what send by my friend from far east, and second picture with IMD test just show that PAN of my F6700 suitable for measures and trustable - at least RF detector and SW in this part works well !

So i repeat test with 100 Hz and 1000 Hz tones  to understand suppression of this frequencies.
NO overdrive, NO additional HW or SW in use both test are very clean. You can repeat this test so easily and in case missing Generator SW you can make you tone or pairs here :  http://www.audiocheck.net/audiofrequencysignalgenerator_dual.php for playback with you windows audio player, so results looks like:

1. Transmitter power = 1 watt




2. Transimmer power 40watt (slider on 40 watt)



Here we can see more harmonics and subharmonics but result so repeatble


I will very happy if my measure not repeatable at all and so appreciate you notes with your measures.



P.S. To compare and measure purpose your can download WAV file with pairs of the my test tones 100hz +1000hz here:
https://www.dropbox.com/s/ilfkef12271j6pb/audiocheck.net_100Hz_-10dBFS_1000Hz_-10dBFS_10s.wav?dl=0

Sergey R5AU

So Tim, jut reviewed original message from my friend , so you are right this is not a Pink Noise but a Human Voice with aaaaa- , nevertheless result clear enough. So on you screen could you make more wider (4-5 time)  passband of filter  and you can see artifacts in opposite side band

Sergey R5AU

So, just received corrected update from George UA0IHZ with pink noise as screen shots:


1. Flex-6500 Low Cut 100Hz_Pink Noise PROC OFF



2.  Flex-6500 Low Cut 100Hz_Pink Noise PROC ON




3. Flex-6500 Pink Noise PROC OFF



4. Flex-6500 Pink Noise PROC ON






--------------------------------------------------------------------------------------------------------

IMHO :  in my experience since July 2013 issues with opposite side band suppression 
appeared/disappeared periodically

K2CB Eric Dobrowansky

I have been running 0-3000hz , and occasionally 0-4000Hz (low end of 80m rag new) bandwidth, for some time now. I would regularly get good reports from the Flex users, and even the Anan ESSB guys. However, after the second to last upgrade, I also received reports that my low end was blowing out and causing opposite sideband issues. Similar to the above reports, if I raised the low end to 100Hz, things got better. But something changed in the last two updates...... in the past running down to 0Hz was not an issue.

Bill W2PKY

In the Radio Setup panel clicking Filters there is a setting for Voice; Has anyone tried different settings to see if the problem can be solved with a different setting?

Sergey R5AU

Bill, all tests with Auto=Off and Filters=Sharp, I love my radio but like Erik mentioned above , I received not so pretty reports from other SDR owners . BTW that can be a problem in DIGI with wide bandwidth and others negative impacts.

DH1RK

@All,
this phenomenon was visible for a longer time, too!
First remark from other SDR user was 19.12.2016 => so version 1.10.8
So, it not an issue of 1.10.16 only.

Attached some pictures...

Flex 6500 with 10.16.xx:


vs. ANAN-100D (different station,more bassy sound):


Best regards,
Ruediger

Sergey R5AU

HI Ruediger !  you are completely right and really strange to listerning missing of issue by other hams. I do not wonna to talk concerning core of issue - i am not SW designer here but it should be investigated overwise it looks like classic superheterodyne with carrier on -10db on the low filter slop 

DH1RK

Hi Sergey, I forgot to mention, the zero beat on the Anan was not from the Anan it was 3rd station carrier as you can see on the bottom part... Okay, the 100Hz frequency parts in the other channel are for practical activities are not so highly important, they are down 25dB but for our competitor users visible. I will only mention the missing pre-distortion feature. For many hams I think it's the main topic not to buy a Flex! Are there any news in the right direction? Will follow the development. Kind regards, Ruediger, DH1RK

Chris DL5NAM

LOL

>> For many hams I think it's the main topic not to buy a Flex!<<

Sergey R5AU

Chris   I wonna to make our radio better !
from here :  http://www.flexradio.com/downloads/flex-6000-family-datasheet-pdf/

Sergey R5AU

4 years ago - https://community.flexradio.com/flexradio/topics/opposite_sideband_suppression?topic-reply-list%5Bsettings%5D%5Bfilter_by%5D=all

Chris DL5NAM

Sergey, yes - making the Flex better is always great. There is always some space for that.
I am with you!

But not with KILLER phrase like Rüdiger. That's not helpful - it's for me attempt at blackmail.

Problem is now on table - FRS will give us a statement and if problem real - i trust in FRS that they bring a solution for us. (no, i am not a ELMER :-)

73 Chris

Sergey R5AU

So , with other syntaxis  i also like to be focused on constructive  behaiviour.
Problem recognised and adressed and i hope will be solved by FRS in upcoming SW release, however issue not yet confirmed by FRS (  :-(  )

K2CB Eric Dobrowansky

Has Flex acknowledged this issue yet?

Sergey R5AU

Not yet unfortunately

K2CB Eric Dobrowansky

I'm surprised more users have not commented on this issue. Tim, is Flex aware of this issue, or should one of us open a formal trouble report?

Bill -VA3WTB

The reason is because Flex has always had one of the cleanest TX anywhere. I have seen many 6000 Flex radios with amazing TX,,just like the Anan shown above.

If you are still having trouble, open a help ticket.

K2CB Eric Dobrowansky

Huh?


One of the cleanest? Compared to what?


Other 12v PA radios? Maybe.


But like the ANAN with APD shown above?

I don't think so! Not even close.


Or compared to the 50v PA in the 7700/7800/7851 radios? Also not even close.


Anyway, back to the topic at hand - something has changed in the last release or two. This issue was corrected once before, early on in SSDR, but it has evidently resurfaced.


Tim - have there been any other additional reports of this issue recently?


Eric

K2CB

Sergey R5AU

Bill, I am talking concerning very dedicated frequencies --> as you can see on pictures above all question regarding freq. below 300hz, in case you have cut off on you TX filter like 200 or 300 hz, reports will be more clean, however from my side question still open, example : why we have in SSDR EQ with 63 or 125 hz if that freq are not suppressed enough ? In this thread above I placed links for WAV samples what can use by everyone to duplicate this issue. Here in RU I got reports at least from 5 hams who are involved in ESSB or got reports from other SDR users based on PowerSDR SW.

Sergey R5AU

You are right Eric, that issue was before already and then solved and before enter the formal ticket I ask hams to repeat my test from above with report- 5 min effort.

DH1RK

Hello,
I think we are talking about the TX filter shaping factor / filter parameters.
It´s digital and perhaps something with the calculation for the coefficients is not optimal...?

Ruediger, DH1RK