AM audio spectrum display not symmetrical

All bug reports here please
RadioSDR1
Posts: 59
Joined: Sun Jun 26, 2022 4:41 pm

Re: AM audio spectrum display not symmetrical

#11

Unread post by RadioSDR1 »

@gregsa, well, the answer to this question is above my pay grade and ability. :) I do not use the audio spectrum display feature very much. However, your post piqued my curiosity and I looked at the function and it does in fact seem to possess some erratic and/or unexpected behavior. Whether it is merely a display only issue or is revealing an underlying audio problem, I cannot say. Though the audio seems fine to my ear (for whatever that is worth) so my guess is that this is a display issue that is not revealing an issue with what is actually heard.

Anyway, you made a very obscure catch in my opinion. Not sure if I would have noticed that or not if I used the feature regularly.

jdow
Posts: 800
Joined: Mon Aug 10, 2020 8:17 pm

Re: AM audio spectrum display not symmetrical

#12

Unread post by jdow »

First, the "DC" carrier present is an inescapable aspect of AM. It is the AM definition.

Second, the 25 Hz HPF is meant to remove the DC offset at frequencies below which the human ear does not work. It is also a frequency at which fading effects won't give you huge very low frequency speaker excursions. Some forms of fading may still give you artifacts.

{o.o} Joanne

Max
Posts: 859
Joined: Thu Aug 13, 2020 9:47 am

Re: AM audio spectrum display not symmetrical

#13

Unread post by Max »

RadioSDR1 wrote: Mon Jul 18, 2022 9:41 pm Anyway, you made a very obscure catch in my opinion. Not sure if I would have noticed that or not if I used the feature regularly.
This has been debated at length many times on here over the last few years and also on the other SDRC group (IO Groups) and both the "issue" and solution are very well documented as referenced by Joanne (jdow) in the preceding post. Also Simon has responded in quite a few of those threads and as far as I recall has deemed it a "non-issue". Personally I see this behaviour all over the place including when handling audio waveforms in many other digital audio applications.

Best

Max

RadioSDR1
Posts: 59
Joined: Sun Jun 26, 2022 4:41 pm

Re: AM audio spectrum display not symmetrical

#14

Unread post by RadioSDR1 »

At the risk of stating the obvious, I am dipping my toe into a subject where both Joanne and Max are orders of magnitude more knowledgeable than me. Be that as it may, I shall further dip my toe further at the risk of fleeing the scene with my tail between my legs. :)

For reference, I paste the following paragraph lifted from the Wikipedia entry for “DC bias”…

Audio
In audio recording, a DC offset is an undesirable characteristic. It occurs in the capturing of sound, before it reaches the recorder, and is normally caused by defective or low-quality equipment. It results in an offset of the center of the recording waveform that can cause two main problems. Either the loudest parts of the signal will be clipped prematurely since the base of the waveform has been moved up, or inaudible low-frequency distortion will occur. Low-frequency distortion may not be audible in the initial recording, but if the waveform is resampled to a compressed or lossy digital format, such as an MP3, those corruptions may become audible.[2]


So based on the above citation, it seems DC offset is undesirable.

Because of my ignorance on this subject, I cannot validate or invalidate any assertion about the commonality of DC offset issues with the function the OP raised in SDR and other digital applications. Assuming that assertion is true, the question is still whether the oscilloscope function should be corrected to show a more intuitive display of the audio wave form.

Now, if the center of the audio waveform is actually offset in terms of what is being delivered to the speakers, then the display is showing the world as things actually are. But if that is true, then it begs the question as to whether that should be fixed (and if so then presumably it would automatically correct the display).

DC rears its ugly head in the form of the “DC spike” in the center of the waterfall that can be successfully suppressed. If DC artifacts can be suppressed there, why not here too?

Lastly, in the grand scale of priorities (assuming that there might be consensus that the OP’s issue could be fixed in theory), it has got to be a pretty low priority. If I were Simon and did not feel this was worth chasing, I might add in a future update a link to a dialog box in the oscilloscope function that provides a little explanation of the anomaly. If this issue has come up in the past, then a note in the function might stop people from posting on it in the future. Maybe. :)

jdow
Posts: 800
Joined: Mon Aug 10, 2020 8:17 pm

Re: AM audio spectrum display not symmetrical

#15

Unread post by jdow »

In an AM demodulator DC bias is absolutely necessary for the waveform. About the only reasonable way to remove it effectively is to use a high pass filter. You could fumble around with some form of setting to "balance out" the DC offset; but, signals are not by any means always the same level. The ARRL's :Radio Amateors Handbook" has or at least used to have good discussions including analog demodulators. If this rough level of understanding is interesting to you I highly recommend getting it and going through the first few chapters for a decent technician's understanding of radios and transmitters. It is what I started on back in 1958 or so. It might help decode me, too.)

{o.o}

RadioSDR1
Posts: 59
Joined: Sun Jun 26, 2022 4:41 pm

Re: AM audio spectrum display not symmetrical

#16

Unread post by RadioSDR1 »

Hi Joanne.

At the moment, I am just a lowly shortwave listener and not a ham radio operator. Nonetheless and however, I just blew the dust off of my copy of the 1986 ARRL handbook somebody gave me when I was in college in the 90s. I am getting old. :) Anyway, nothing jumped out at me dealing with this topic. So off to YouTube I went!

I found a great and short little video (link at end) from amateur operator W2AEW titled “Oscilloscope Vertical Position and Offset Explained.” This guy could probably succinctly explain the entirety of quantum mechanics in 5 minutes. :)

The gist I derived from his video is that the default oscilloscope presentation will be the AC audio signal riding along the DC bias voltage of the transistor and this will raise the center line of the audio waveform above the center line of the oscilloscope. And this seems to be what is being shown in SDR Console. However, he says there are two ways to get rid of it.

One way is to use what this guy called the “AC coupling” feature of his oscilloscope. When that is engaged, it eliminates any reference to the DC component and intuitively shows only the audio waveform at the center of the scope.

The other way is to adjust the “DC offset” feature of his oscilloscope. In his example, he had 2.7 volts of DC bias applied to the transistor in his test circuit. So if he offset the DC bias in the scope by exactly 2.7 volts, then the audio waveform would ride along the DC bias in the exact center of the scope.

As a side bar, I am not sure where the DC bias is being measured from that seems to appear in the SDR Console oscilloscope. Is the DC bias from transistors in my SDR radio? Is it from transistors in my computer sound card? Somewhere else? If so, then it seems the exact DC offset may need to vary depending the exact hardware cocktail of the user. And that seems hard to know since it can vary user to user.

Therefore, it seems emulating AC coupling of the waveform is the solution because as was correctly stated earlier, SDR Console is not intended to be a lab instrument. Therefore, nobody should care nor need to see any representation of the DC bias.

That all said and in theory, if Simon wanted to and was not constrained by any other as yet to be discussed obstacles, could he in theory implement AC coupling of the audio waveform in the oscilloscope function? And if so, would that not address the OP’s issue?

P.S. 90% of my knowledge basis of DC bias and offset was just obtained mere minutes ago from watching this aforementioned YouTube video. :)

Here's the video link:


jdow
Posts: 800
Joined: Mon Aug 10, 2020 8:17 pm

Re: AM audio spectrum display not symmetrical

#17

Unread post by jdow »

The feature that AM brings to this picture is that the DC bias will be proportional to the AM signal level as modified by the AGC and it's action. With sine wave signals the 100% modulated AM signal is a sine wave rest its low side on the zero volts line and its maximum exactly twice it's average level. With math the signal is A*(1+sin(f*t)). A is a high frequency carrier sine wave. The Wikipedia article may be overkill or it may be understandable. It is that 1 in the parenthesis, the carrier amplitude term, that gives the inherent DC offset. It is a necessary feature. If the carrier term is less than 1 then you have the spectacle of a "negative" power appearing. Depending on the modulator in use this turns into a flat line. Most modulators do that. A DSB suppressed carrier modulator with a carrier added back in can deliver that "negative" power in some demodulators. A simple diode demodulator or most other common AM demodulators will distort wildly. A SAM (Synchronous AM) demodulator can deal with it as can a standard DSB demodulator locked to the otherwise "huge" pilot carrier. And then we sort of go off into the flammable weeds indigenous to the area where I live.

{^_-} Joanne - going WAY overboard here, I suspect. It's just that I like to find ways to do things my peers told me could not be done. Used to was if my intuition said (nasty word for fertile bovine poo) it was usually right. That often caused problems.

RadioSDR1
Posts: 59
Joined: Sun Jun 26, 2022 4:41 pm

Re: AM audio spectrum display not symmetrical

#18

Unread post by RadioSDR1 »

jdow wrote: Fri Jul 22, 2022 6:07 am ...and then we sort of go off into the flammable weeds indigenous to the area where I live.

{^_-} Joanne - going WAY overboard here, I suspect. It's just that I like to find ways to do things my peers told me could not be done. Used to was if my intuition said (nasty word for fertile bovine poo) it was usually right. That often caused problems.
LOL! :) Nothing is better than humor and radio with morning coffee!

So riddle me this. I think a possible point of confusion for me is the distinction between the AM carrier & modulation vs. the demodulated audio that arrives at the speakers. Rightly or wrongly, I am and have been operating off the assumption that the oscilloscope function in SDR Console is the demodulated audio and as a result of this assumption, any DC bias characteristics that apply to the AM side of things is moot. But it seems my assumption may not be correct.

If you could elaborate on that, that might make me a little less green with this subject matter. Maybe. :)

jdow
Posts: 800
Joined: Mon Aug 10, 2020 8:17 pm

Re: AM audio spectrum display not symmetrical

#19

Unread post by jdow »

It is showing the AM modulation. It is always in the form of a DC term, 1.0 for convenience) plus the audio at levels from 0 to +/- 1.0 for the peak signal. If there is no high pass filter going into Simon's display, the DC term coming out is the DC term going in as part of the modulation process.

{^_^}

RadioSDR1
Posts: 59
Joined: Sun Jun 26, 2022 4:41 pm

Re: AM audio spectrum display not symmetrical

#20

Unread post by RadioSDR1 »

After mulling all this over and performing a few additional experiments and observations, I believe with all due respect that the general substance of the OP’s observation is 100% spot on and that the behavior in the oscilloscope function for AM is undesirable, unnecessary, and a bug! :)

Now, there are those who know way more about this than I do and may argue there is no problem here. However, one does not need a Masters Degree in Fire Science to know a building is on fire. In turn, one does not need an audio or RF engineering degree to know this oscilloscope function is not right on AM.

What I have learned through this dialogue is that the DC bias from the AM carrier is being brought forward to the audio waveform.

How could that be normal, natural, or expected???

The fact it can be filtered (mostly) out by the 25 Hz HPF in the equalizer function proves it is a problem. If all that is required to fix this is to filter it, then that filtering should be applied automatically behind the scenes in code pertaining to the AM function so then nobody needs to think about it. Furthermore, I observed through experimentation that while the 25 Hz HPF does indeed fix most of the problem, it does not fix all of it (but is ~95% a good work around). Anyway, a filter designed for this exact issue would fix it 100%.

I am puzzled why there would not be a straight forward acknowledgement of the OP’s observation. Since I use SAM for AM most of the time, I never noticed that the audio is actually degraded as a consequence of this problem. Earlier I posted uncertainty as to whether or not this was a display only issue. I am now 100% certain the display is fine because it is showing exactly what is happening to the audio (though that effect is generally subtle in my opinion but it does negatively affect the AM mode audio nonetheless).

SDR Console is free software and those who use it are lucky to have it because, in my opinion, it is the best that’s out there and it’s free! It makes radio fun!!! To the extent either Simon or other users want to know about quirks in the software, well, now you all know about this one. This oscilloscope issue, as minor as it is, is in fact a definite (though minor) problem in my opinion.

And my opinion is always tentative and thus subject to further revision should anyone persuade me otherwise. :)

Cheers!

Post Reply