Amplitude modulation on 'constant envelope' modes

User avatar
Conrad_PA5Y
Posts: 101
Joined: Fri Jan 19, 2018 8:11 pm

Amplitude modulation on 'constant envelope' modes

Postby Conrad_PA5Y » Fri Mar 23, 2018 2:32 pm

By constant envelope modes I mean JT65A/B/C, FT8, MSK144 and FSK441.

I noticed that the edge meter of my RF Power meter in TX was flickering as was the indicated output power, also the LCD output meter on my PA. I cannot see these effects on a conventional meter, in my case a Bird Model 43. Monitoring the signal on another radio does not sound strange or wrong in any way.

Upon checking with others they do not see this but instead see a steady state on the output meter when using these modes, as do I when using other radios.

At first I thought that it was a VAC issue but the monitoring tools with PowerSDR mRX PS v3.4.9 disprove that as does my tests below.

Below are the tests that I have performed to disprove VAC as an issue, these tests do not use VAC. However the results when using VAC are the same.

I finally got around to trying the line input with the ANAN-100D - resulting in the same issue of amplitude modulation on MSK144 and FSK441 and to a lesser degree on JT65A and B.

My TX filter bandwidth is from 200 -3100Hz. Test software was WSJT-X v1.8.0

To be clear - I have NO compression, TX EQ or CFC tools active and I am using DIGU, there is no fancy stuff on the soundcard active. I have the correct sample rate of 48000Hz and there are absolutely no audio glitches that I can hear either when monitoring the input audio or demodulated RF on a nother radio. Everything decodes fine - I want this perfect for EME use, for 'normal' use it makes hardly any difference.

I will try to explain what I have tested and observed please ask if anything needs further explanation.

Tests


1. Used UMC202HD on another PC with WSJT-X audio out to IC-746, IC7300, TS-2000 - All modulation modes - there is no AM - checked that there was no ALC action in all cases. This proves the audio source is good for the purposes of these tests.

2. Connected same setup to both line input and then front panel microphone input, set levels so that Mic meter to shows -2dB - this remains, steady - result is still AM, using various single tones 500, 1000, 1500 also shows some variation in amplitude versus frequency both in terms of the mic level and the RF power output. However the RF output from single tones via Line or VAC remains constant, no jumps or AM. Just that there is an output variance with frequency - Before you ask I am 1000% sure that there is no EQ anywhere. I am using DIGU and have checked TX EQ, the leveller and CFC tools. They are all OFF. TX bandwidth has very little effect. Interestingly FT8 exhibits very little AM so bandwidth does seem to play a part . This has a narrow TX bandwidth. The behaviour using Line in and VAC is the same. I do see the AM on the ALC meter.

VAC SRC shows no underruns or overruns for many many hours, there is no problem with VAC I am sure of it now.

3. Any DSP generated tones such as Tune or 2 tone tests show rock steady power output - so the instability not in the PA.

While I do not see any AM on the Mic level meter I do see some on the ALC meter which is nominally -2dB, I have no ALC compression. This really mystifies me - how can this be when all processing occurs in the DSP?

I really hope that someone can shed some light on this. It has me completely baffled.

Regards

Conrad PA5Y
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby w-u-2-o » Fri Mar 23, 2018 10:01 pm

Conrad,

When you say you see AM, how much AM do you see? 1dB? 3dB? 10dB? Perhaps a video of your metering would tell us a better story.

According to this paper here:

https://physics.princeton.edu/pulsar/k1jt/MSK144_Protocol_QEX.pdf


MSK144 has a peak to average power ratio (PAPR, or "crest factor") of 1.28 (straight ratio, not decibels), or about 1dB (10log(1.28)), when measured in a 0-3KHz passband. If you are seeing wiggles of 1dB in modes such as this, it is no big deal and totally expected.

Also, the metering on PowerSDR is not perfect. There will be some slight movement of the meter. Whether this is due to the intrinsic metering process used and/or internal cross-talk is more than I can say. If you really want to know what is going on, do you have an oscilloscope that you can use with an external coupler or sampler to see the signal in the time domain in more detail?

Finally, are you running PureSignal in automatic mode with these modulations? I generally set up PureSignal for digi mode operations using the single calibrate method so that the automatic algorithm does not become confused by constant envelope, or single tone , modes.

73,

Scott
User avatar
Conrad_PA5Y
Posts: 101
Joined: Fri Jan 19, 2018 8:11 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby Conrad_PA5Y » Sun Mar 25, 2018 2:01 pm

Hello Scott.

Sorry for the late reply I have been away.

I see quite large deviations in excess of 2dB.

I see this issue both with and without pure signal active. I set up PS with 2 tone testing prior to using it with data modes.

Regards

Conrad
AB2EZ
Posts: 99
Joined: Sun Apr 09, 2017 2:29 pm
Location: Princeton, NJ

Re: Amplitude modulation on 'constant envelope' modes

Postby AB2EZ » Sun Mar 25, 2018 5:40 pm

Conrad

Since

A) You see no amplude modulation when using WSJT in tune mode ... with a given nominal frequency (e.g. 500 Hz)... but you see a variation in the RF output when you change the nominal frequency.

B) You have looked directly at the audio output of your sound card (digital audio interface), and you have seen no corresponding variation in the amplitude if the analog audio output when you change the WSJT nominal frequency (e.g. observing the sound card’s output waveform using an oscilloscope at the same time that it is driving the audio line input of your ANAN

It appears that there is some frequency dependent gain in the modulation process that is taking place within the HPSDR software application.

You have said that all HPSDR processing features, such as equalization are, turned off... and that you are using transmit profile that has enough bandwidth...but your radio is not behaving that way.

I suggest that you verify, again, that all equalization-related functionality is set to produce a flat frequency response.

Stu
User avatar
Conrad_PA5Y
Posts: 101
Joined: Fri Jan 19, 2018 8:11 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby Conrad_PA5Y » Sun Mar 25, 2018 6:26 pm

I have done this until I am blue in the face Stu. I follow your logic but unfortunately it does not provide a solution. I will leave it a day or two I think.

I have checked the frequency response of the audio from both a UMC404HD, and an UMC202HD as well as using audio tools on the VAC output so that is also not it. I should have mentioned this but of course the fact that it does not happen when using 3 other conventional transceivers it was never likely to be the issue.

Given the facts it does point to some frequency selectivity in the ANAN-100D TX profile but TX filter is 200-3100 so that's not it, There is no TX EQ active so that's not it, There are no CFC tools active so that is not it.

It's not the DAC because Test signals both using tune and 2 tone tests do not have the problem, so I am stumped, which is why I asked here.

Regards

Conrad
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby w-u-2-o » Sun Mar 25, 2018 10:59 pm

I am stumped as well. I cannot reproduce this behavior on an 8000 under any reasonable conditions.
AB2EZ
Posts: 99
Joined: Sun Apr 09, 2017 2:29 pm
Location: Princeton, NJ

Re: Amplitude modulation on 'constant envelope' modes

Postby AB2EZ » Mon Mar 26, 2018 6:45 pm

Conrad

If you change the WSJT-X FT8 tune frequency in 500Hz steps... between 500Hz and 2500Hz... what are the corresponding output power readings... relative to the 500Hz reading?

Also, have you tried toggling the transmit equalization on and off to see if the transmit equalizer is “stuck” in the active state?

Stu
User avatar
Conrad_PA5Y
Posts: 101
Joined: Fri Jan 19, 2018 8:11 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby Conrad_PA5Y » Tue Mar 27, 2018 9:49 am

Hello Stu I will check this tonight, I will do it with both VMB and via the line input. I have toggled the TX EQ many times.
User avatar
Conrad_PA5Y
Posts: 101
Joined: Fri Jan 19, 2018 8:11 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby Conrad_PA5Y » Tue Mar 27, 2018 9:50 am

w-u-2-o wrote:I am stumped as well. I cannot reproduce this behavior on an 8000 under any reasonable conditions.


Well I have a 100D and as you know I am unreasonable. Especially when I am talked to in a condescending manner. No one likes a smart Alec Scott you are making me & others scared of posting. I am really fed up of it, I find your manner intimidating. Maybe it's cultural?

I have lost my temper that's clear.

I asked you a question in private and you forced me on here. I don not like the way you conduct yourself on here which is why I asked you in private.

I have ample test equipment and ample knowledge to work this out for myself, I was looking to save a little time that's all, I want to use the ANAN to work DX, it is a tool to me nothing more.
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby w-u-2-o » Tue Mar 27, 2018 11:39 am

Conrad,

I'm sorry you feel that way.

You might note that I was publicly admitting my ignorance in this matter, which would be considered by some to be a brave and honest thing to do.

Given your obvious displeasure with me it is not surprising that you took the wording of my post to be an attack on your person. I hope you will believe me when I tell you that that was certainly not my intention. By "reasonable" I only meant normal, non-edge case conditions.

73,

Scott
AB2EZ
Posts: 99
Joined: Sun Apr 09, 2017 2:29 pm
Location: Princeton, NJ

Re: Amplitude modulation on 'constant envelope' modes

Postby AB2EZ » Sun Apr 01, 2018 7:41 pm

Conrad

Hi!

Any update? E.g power out v. FT8 tune frequency?

Stu
User avatar
Conrad_PA5Y
Posts: 101
Joined: Fri Jan 19, 2018 8:11 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby Conrad_PA5Y » Sun Apr 01, 2018 9:50 pm

Hello Stu, sorry no. I am moving house and have been on 2m EME then a 4m contest. Its a holiday here tomorrow so I will get chance for sure. Please accept apologies for my tardiness.

I did manage to work ZS6NK on 6m EME today using JT65A so nothing much wrong I think. I have a single yagi.

As Scott pointed out with reference to the MSK144 article in QST as you narrow the BW of the TX filter you can expect some AM from MSK144, With a 3kHz TX filter on my system it is around 2dB.

I will use a diode detector and a scope to measure the peak to RMS variation accurately, I can also try a zero Hz span on my E4406 spectrum analyser (the proper spelling ).

I still think that something is not quite right.

Another thing I can do is use the 2 tone tests in PowerSDR with tones to cover different area of the TX filter.

73

Conrad PA5Y
User avatar
DL2XY
Posts: 116
Joined: Sun Jul 30, 2017 9:47 am

Re: Amplitude modulation on 'constant envelope' modes

Postby DL2XY » Mon Apr 02, 2018 11:41 am

I have checked for this issue on Anan100 on 20m, FT8-Mode via VAC from WSJT-X.

Indeed there is a little residue AM of about 1% peak, but only during frequency changes of modulation.
IMHO this is just an effect of the phase jumps during frequency transitions.


restam.jpg
restam.jpg (68.81 KiB) Viewed 15392 times



73 Walter [DL2XY]
User avatar
Conrad_PA5Y
Posts: 101
Joined: Fri Jan 19, 2018 8:11 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby Conrad_PA5Y » Mon Apr 02, 2018 6:11 pm

Thanks Walter, that is the next stage for me to roll out the analyser.

As promised I have tested WSJT-X static tones and amplitudes, everything behaves as expected and output is stable with static tones but I do see variations in amplitude on the Fwd Pwr meter when using JT65A,B & C which I do not expect. In addition there are amplitude variations when using MSK144 which is expected but seems rather high.

FT8 is stable.

For these tests the TX filter was set from 200-3100Hz.

I used a 1kHz tone as a reference, the input amplitude was set to -1.5dB FS as indicated on the Mic level meter, the output power was set to 30W. Pure signal was off.

I then used tones from 200Hz to 3300Hz in 100Hz increments using the TX frequency and Tune control on WSJT-X 1.8.0. the results may be viewed in the table below:

Image

Next I will look at the results in more detail using an attenuator and E4406A Vector signal analyser.

Walter it would be really helpful if you could repeat your tests with JT65B and MSK144 please.

Regards

Conrad PA5Y
User avatar
Conrad_PA5Y
Posts: 101
Joined: Fri Jan 19, 2018 8:11 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby Conrad_PA5Y » Mon Apr 02, 2018 9:10 pm

I still need to experiment a little but using markers and a combination of RMS and peak hold values I have determined the following.

Peak to RMS with 200-3100Hz filter is 3.4dB for MSK144
Peak to RMS for FT8 is 0dB
Peak to RMS for JT65A and B 0dB

Same conditions as the previous post.

MSK144 still looks too high to me.

Regards

Conrad PA5Y
User avatar
DL2XY
Posts: 116
Joined: Sun Jul 30, 2017 9:47 am

Re: Amplitude modulation on 'constant envelope' modes

Postby DL2XY » Mon Apr 02, 2018 9:59 pm

Here are the results of the requested measurements.
Analyser is Rohde&Schwarz FSAS in Receiver-Mode (AM-Demodulation).


JT65 shows the usual phase jumps at frequency transitions but a straight line between them:

JT65 AM.jpg
JT65 AM.jpg (57.41 KiB) Viewed 15358 times


MSK144 has AM components of about 16% as expected.
Note the changed scaling and timebase.

MSK144.jpg
MSK144.jpg (62.32 KiB) Viewed 15358 times


MSK144 on logaritmic scale, this is only 1.4 dB peak.

MSK144_log.jpg
MSK144_log.jpg (94.03 KiB) Viewed 15358 times


Regards
Walter [DL2XY]
User avatar
Conrad_PA5Y
Posts: 101
Joined: Fri Jan 19, 2018 8:11 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby Conrad_PA5Y » Tue Apr 03, 2018 5:32 am

Thanks very much Walter.

I don't have an AM demodulation option fitted to my Agilent E4406A however it does have Vector Signal analysis tools in the time domain so I should be able to do something similar. I hardly ever use the time domain tools and need to spend a little time experimenting. I also have Linrad and a Perseus which I can use to make demodulated measurements. This will also take a while to figure out.

In order to measure the RMS to peak ratios in dB I merely used spectrum analysis on the E4406A with markers and rms and peak hold. That seems like a reasonable approach.

What TX filter bandwidth did you use out of interest?

I am beginning to think that I have no problem with WSJT-X V1.8.0 on JT65 and FT8 but something is going on with MSK144.

Can you look at the same modes using MSHV please, there seems to be something wrong with MSHV as I see quite some variation on static tones and the harmonics of single tones are much worse than WSJT-X.

I really appreciate that you have taken the time to make these measurements.

Regards

Conrad PA5Y
User avatar
DL2XY
Posts: 116
Joined: Sun Jul 30, 2017 9:47 am

Re: Amplitude modulation on 'constant envelope' modes

Postby DL2XY » Tue Apr 03, 2018 9:36 am

Hello Conrad,

there is virtually no difference in PAPR of MSK144 between WSJT-X and MSHV.

But there are much differences in signal quality.

Second harmonic of 1kHz Tone (Tune Mode) @ -1dBfs:

WSJTX: -98 dBc
MSHV: -45 dBc


Frequency response (JT65 @ 1kHz AF) is worse too.

WSJT-X: 0.06 dB peak
MSHV: 0.24 dB peak

Below zero spans (JT65) over 10s with high amplitude resolution:

JT65-WSJTX.jpg
JT65-WSJTX.jpg (75.85 KiB) Viewed 15332 times


JT65-MSHV.jpg
JT65-MSHV.jpg (78.28 KiB) Viewed 15332 times



Hint: You may use PowerSDR for precise PAPR measurement:
Activate RX2 on predistortion-sampler.
Set VFO-B to VFO-A.
Set RX2 bandpass filter for resolution bandwidth.
Transmit on VFO-A.
Use RX2 Multimeter dB readout to compare Signal (peak) with SigAvg (RMS)

Regards

Walter [DL2XY]
User avatar
Conrad_PA5Y
Posts: 101
Joined: Fri Jan 19, 2018 8:11 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby Conrad_PA5Y » Tue Apr 03, 2018 9:45 pm

Walter thanks so much for the testing I am really grateful. I see the harmonics in RX2 with static tones on MSHV, I have posted to the MSHV email reflector about it. Thanks for the tip about the peak and average metering. I never thought of that!

I won't get chance to do any more testing until late tomorrow evening.

Once again thanks a lot!

Regards

Conrad PA5Y
User avatar
Conrad_PA5Y
Posts: 101
Joined: Fri Jan 19, 2018 8:11 pm

Re: Amplitude modulation on 'constant envelope' modes

Postby Conrad_PA5Y » Fri Apr 06, 2018 9:25 pm

Walter what time constant for the metering is appropriate for this measurement?

73

Conrad PA5Y

Return to “Digital Mode, Rig Control & Logging Software”