Flex 6300 CW Audio peak off by 15Hz and audio pops when tuning - Repost

Andreas Junge

Posting to the SmartSDR forum this time...

I have been playing with a loaner Flex 6300 for about a week now. I do a lot of CW and noticed that the signal was getting a bit weaker as I tuned to the correct frequency. 

I noticed that the audio is peaking 15 hz below the supposedly tuned frequency.

Actually I measured not too scientifically:

Using the SPLMeter app on the iPhone 5S for audio level readings. The phone is placed in front of the amplified speakers that are connected to the radio.

The radio is set to 600Hz pitch, AGC off (50) , NB off, NR off, APF off, 50HZ CW

I did run the calibration against the 15Mhz WWV first.

I used a Elecraft XG3 signal generator @ 50uV as constant amplitude signal source:

  Display peak of signal 14.020.000 @ 600 Hz tone - good!

  Audio peak on  14.019.985 @ 585 hz tone - why?

I also get frequent audio pops when tuning around the center frequency. There is no popping when just listening.

H/W 1.2.17.35

SSDR 1.2.17.65

73, Andreas, N6NU

Steve N4LQ

Andrew
Unless I'm missing something, the amplified speakers resonance doesn't seem to be taken into consideration in this test and they could be a factor. How about using a 600 ohm load across the 6300's output and a scope?
I'm using Bose Companion II's and their response certainly isn't flat. 

Andreas Junge

Steve,

I saw as much as 2 DB difference. I doubt that the speakers' frequency response changes much over a span of 15Hz (not KHz).  I also saw the same behavior at a pitch of 650Hz.

I don't think it's the speaker. 

73, Andreas, N6NU
  

Steve N4LQ

OK...I'm going to test mine this afternoon if I have time. I have the SPL meter and the same Elecraft generator. 

Andreas Junge

Steve,

excellent. BTW, I am using an audio frequency counter app to make sure I find the 600Hz point first - An Auto Tune function like on my K3, KX3 and IC7800 would be nice :-)

Then I switch to the audio meter and peak for maximum audio.

Make sure the AGC is set to OFF and gain value of 50.

73, Andreas, N6NU

Tim - W4TME

I have added this topic as a bug in our bug tracker for additional investigation.  Thanks for the report and the detailed information for reproducing it Andreas.

Steve-N5AC

Using a dB meter on my iPhone since it was handier than plugging in the scope, I ran the following tests:

- GPS locked radio (6700 in this case, but the software is the same as the 6300)
- GPS locked HP8565B signal generator
- Generator tuned to 14.100 MHz, signal amplitude set to -106dBm
- Radio tuned to 14.100MHz, APF on, AGC OFF, AGC-T set to reasonable volume level on speakers (59), APF slider MAX to get highest peaking for test

I measured the following volume levels at radio frequencies of:
14.099985  73dB
14.099990  74dB
14.099995  77dB
14.100000  78dB
14.100005  76dB
14.100010  74dB
14.100015  72dB

I don't seem to see an issue here.  Are you certain that your signal source is on frequency?

Andreas Junge

Steve,

thanks for testing.

Why did you turn ON the APF? It was off when I tested. I used a audio frequency counter to tune to 600 hz side tone first, then I peaked for maximum audio. it takes 10 to 15 Hz to get to the peak. 

The audio should peak at the side tone frequency, not at 10 or 15 hz from it. The APF adds ringing and it's very tiring to listen to with weak signals. 

Maybe I can try to re-run the test with K6TU's or K6TD's 6700 and see if they behave the same. 

73, Andreas, N6NU

Steve-N5AC

Well this is how I read your concern -- that there was an issue with the APF.  Without APF the audio level is flat across the filter passband.  It is not designed to make the 600Hz region any louder.  The APF, on the other hand, does exactly this.  I started to re-run the test, but it's not possible to run the test you want when a speaker is involved.  Speakers are going to color different frequencies of audio and will not provide accurate results.  I can see a peak around 455Hz on my audio meter, but this is after getting nicely colored by the speakers.  The filter used is flat in the audio region.

The APF is an IIR filter.  There seems to be confusion about what ringing is.  To most engineers, ringing is what happens when an excitation to a circuit with feedback causes an output to continue after the input is removed. Literally ringing a bell does this -- a single whack on the bell causes a tone to persist long after the input stimulus has been removed.  Because APF is an IIR filter, like most APFs, it can do this if not adjusted well.  

BUT, when the slider is turned up, it also is very sharp and removes all but a small subset of frequencies from the input.  When happens then is that this small number of frequencies are amplified and they become all that you hear.  Since the noise in the band is largely evenly distributed, you get to hear just the noise that occurs on your CW pitch and it sounds like someone trying to send code in the background.  Boy that's irritating.  So turn the APF down to where it adds a few dB to the station you have tuned in and pushes others and the surrounding noise down.  When used effectively, it is an excellent tool in my opinion.

To learn how to use APF best, I recommend fiddling with it quite a bit while listening to signals in the CW band and get an understanding for how it works.  If you try it on the bench in the lab, you're likely to be initially confused -- if you turn on a tone and put it on your carrier frequency and try APF on and off, you will notice that it removes noise when you turn it on and adds noise when it is off, but you have to test with a signal that is only 5-20dB out of the noise to tell.  Any more signal than that and you will not hear the effect.  Why?  Because the AGC in the radio is already pushing the noise out of the way for you because the SNR is so high on the signal.  

If you are testing with a signal 20dB out of the noise and you tune off a 100Hz, you will notice the same thing -- that you don't seem to be able to tell much difference.  Again, what is happening here is that the AGC has a target loudness and it will amplify until it gets this.  When you are on the APF frequency (CW Pitch) APF is providing, say, a 10dB boost.  This causes the AGC to drop it's gain by 10dB to get a uniform loudness for your ears.  When you move off channel, you lose the 10dB from the APF, but then the AGC makes up for it!  The REAL test is when you have to signals in the band that are similar in strength.  The APF will quickly separate the two signals in the SNR domain -- by placing the signal of interest in your CW pitch region of the passband, APF will add, say, 10dB.  AGC will drop by 10dB.  Now your undesired signal has been driven 10dB closer to the noise because it does not get the benefit of the APF.  Give it a whirl.

73,
Steve

Andreas Junge

Steve,

thanks for your explanation. I did re-run this test by piping DAX1 output into SpectrumLab - eliminating the speaker. I think you were right, I was hitting the speakers' resonance/peak. 

Things look better now. I'll play with APF as you suggested. 

This just leaves the issue with the audio popping when tuning around. 

Expect my order for a Flex 6500 coming in soon :-)

73, Andreas, N6NU