Talk:Baseband

Does ethernet use baseband signaling?

Yes, it does. Although we don't have that piece of information in an article so far as I can tell. See Ethernet and IEEE 802.3. -Splash^talk 17:54, 14 February 2006 (UTC)

Not necessarily. The orignal Ethernet used "Manchester coding", which can be viewed as shifting the theoretical 5 MHz baseband of 10 Mbps data up to a carrier frequency of 10 MHz, so the band occupied can be as narrow as 5 to 15 MHz. Due to AC coupling with capacitors and/or transformers at various points, moving the peak of the data spectrum up from 0 to 10 MHz is helpful. This, however, is but one physical implementation of Ethernet. Others such as 802.11 shift it up to RF, of course. Dicklyon 16:30, 19 August 2006 (UTC)

CISCO state the the "base" in 10Base-T refers to baseband and by inference surey this means Ethernet uses baseband.

By definition, Ethernet uses baseband signalling. It might be that some statements in the article are wrong. The important one being that the frequency spectrum goes close to zero. Gah4 (talk) 22:12, 7 January 2019 (UTC)

Intro

Let's work on a good opening sentence and definition (current version uses "comprises" in the typical incorrect way, and doesn't get to the point very specifically). It would be good to at least admit that any kind of Fourier analysis gives negative frequencies, too. For real signals, these are just reflections of the positive, not new information, so they are conventionally ignored. And the "RF" spectrum needs to reference double sideband and/or single-sideband modulation to explain where this comes from. I'll see if I can help. Please comment on my attempts. Dicklyon 16:30, 19 August 2006 (UTC)

In typical fashion, this article seems to be geared toward those who would have no reason to look it up on wikipedia in the first place, having already earned their PhDs in engineering. Perhaps it could be made a little less intimidating for the uninitiated.--72.255.35.11 (KartoumHero) 02:25, 16 July 2009 (UTC)

In the Intro, it says "from zero" to the highest frequency. Later on in the article, it doesn't say "from zero" but only to the highest frequency. This article is frustrating for that and other reasons. If you don't know what baseband is to begin with, this article doesn't help. 70.166.123.49 (talk) 03:37, 9 February 2010 (UTC)

ῶ —Preceding unsigned comment added by 119.153.61.167 (talk) 01:01, 17 May 2010 (UTC)

Useless Article

What combination of idiots wrote this article? It makes no sense! To be clear, I understand the topic now, but only because I went back to the VERY FIRST draft of this article from 2002 (http://en.wikipedia.org/w/index.php?title=Baseband&oldid=71326)!!! That at least was a clear explanation, especially if read alongside the wikipedia articles for Band and Modulation. I'm afraid to help edit the article though, because I have no idea what the hell the useless drivel on the current version means. Mmurfin (talk) 20:14, 15 August 2010 (UTC)

This article seems to be textbook stuff for students of signal processing theory. There is a specific and widespread usage of "baseband" in mobile telephony (and in phrases such as "baseband modem", "baseband IC") - this article really could benefit from explaining that. Wikipedia's much more useful when it's making sense of fast-moving, real-world concepts than when it's just a substitute for buying a textbook. —Preceding unsigned comment added by 217.140.96.21 (talk) 12:48, 18 November 2010 (UTC)

I second the opinion. This article is not only useless, but also MISLEADING. The Opening paragraph on the Baseband is totally misleading.

Aursani (talk) 12:56, 12 June 2015 (UTC)

Not So Useless, and btw, what's the frequency bandwidth of Ethernet?

I disagree, I see some value here. Go around the web and search for something that connects "Bandwidth" and e.g. "Ethernet". There is so little out there that even many engineers get so confused about something so basic as to claim that an Ethernet signal has no frequency spectrum because it's digital. Granted, it's something engineered once per decade or so, but not digging into this is like not explaining how things work because we're not supposed to understand the guts of a Tamagochi like they used to understand the guts of a spring-driven toy car. And, by the way, can anybody add the real-world (as opposed to information-theory or whatever) spectrum bandwidths of the various Ethernet signals? It' a bit like not mentioning that swimming is a form of locomotion (go see! :-P ) Spamhog (talk) 22:54, 1 December 2010 (UTC)

In the case of many signals, it isn't useful to give one number as bandwidth. You really need a power spectrum, how much power is transmitted at which frequency. In the case of coaxial ethernet, it gets more complicated. While the data signal might fit in a 5MHz to 15MHz band, collision detection is done it DC, that is, close to 0Hz. The signal on the cable is not AC coupled, and the average DC voltage, after filtering, is used for collision detection. It might be that between 1kHz and 5kHz that you could squeeze in some other signal. Twisted-pair ethernet is transformer coupled, and so can't go down, or even close, to 0Hz. Gah4 (talk) 21:55, 28 March 2017 (UTC)

Definitions

What is baseband?

Isn't the definition of baseband 'the signal you are trying to transmit'? Maybe add a quick explanation for the lay person, or an example, like: "In AM radio, the baseband includes the announcer's voice. The baseband is then modulated with the carrier frequency, which is the frequency on the radio dial, and then demodulated so that a voice is heard at the receiver." 71.139.176.47 (talk) 19:35, 12 December 2012 (UTC)

I cannot agree more with this statement. Although replacing the entire article with this talk entry would be a large net improvement, simply offering the above explanation as the first paragraph would at least allay the offensive nature of this article to date. 5000+ people a month are being victimized by this academic fappery. I try not to attribute malice to that which can be explained by the innocent bloviation that comes naturally to experts but articles this bad cause me to wonder if I'm being naive. Allenc28 (talk) 05:56, 13 July 2015 (UTC)

Meaning of Baseband

Transmission of signal with infinite bandwidth is called the basband transmission. A single signal transmitted over the transmission line occupies all the bandwidth of the transmission line. A digital pulse transmitted without modulation occupies infinite bandwidth.

Aursani (talk) 13:03, 12 June 2015 (UTC)

While that sounds very intelligent, it means absolutely nothing to a person with no prior knowledge in this subject. I came here for a refresher and after attempting to decipher this article, I am now absolutely confused. The average non-engineer is not going to be able to make sense of "infinite bandwidth". I cannot get much out of this article as it is currently written. Why is it that I can go on YouTube and watch 1 or 2 videos and understand this topic completely, yet I cannot do anything with this source of "information"? That is an important question that should be address if WP is going to have any value to casual readers. And yes, it does read like a low-quality textbook - which explains why students have such a difficult time with this topic. Perhaps a re-write with better diagrams would help? Good writing should inform, not confuse the reader. 73.6.96.168 (talk) 04:25, 4 March 2021 (UTC)

Definition of Baseband

In my experience and opinion, a baseband signal is a signal that is centered at zero (f = 0) when viewing the two-sided (positive and negative) spectrum. It doesn't really have anything to do with bandwidth, as stated in the article.

It is true that when one needs to modulate a signal, the carrier frequency must be chosen based on the bandwidth of the baseband signal and is usually much larger than the bandwidth, but this is a consequence of modulation.

Yates (talk) 16:27, 8 October 2016 (UTC)

Just reading the first sentence" Baseband is a signal that has a very narrow and near-zero frequency range, raises questions to me. First of all, very narrow relative to what? 1000baseT, that is, gigabit ethernet, is baseband, though with a bandwidth over 100MHz. And how close does it get to 0Hz? Not very. Seems to me that in actual use, a signal is baseband if its bandwidth is much wider than the frequency of the lower end of the band. That is true for wide (say 10MHz or more) and narrow band signals. The result is that there isn't spectral space for another (frequency shifted) version of the signal at lower frequencies. There might be room for other signals, though. Gah4 (talk) 21:44, 28 March 2017 (UTC)

I was about to say something about this, but it seems that I already did. But "narrow" is always relative to something. The available frequency spectral space increases as faster transistors become available for low prices. Narrow isn't as narrow as it used to be. Gah4 (talk) 22:17, 7 January 2019 (UTC)

Carrier

Baseband ethernet is, by definition, baseband, yet it does have a carrier. The distinction, in using Manchester coding, (aka. phase modulation) is that the carrier frequency is close to the modulation frequency (data rate). Actually, the carrier frequency is equal to the data rate. The carrier is obviously necessary for CSMA/CD to work. I think this means that some of the definitions in the article need to be redone. Gah4 (talk) 22:22, 7 January 2019 (UTC)

It might be better to think of Manchester encoding as a pulse position modulation; depending on whether the "DC pulse" is in the first half or the second half of the bit cell, you get a one or a zero. That's pretty much how it's realized, at least in systems I've looked at many years ago, though concentually it might be easier to treat it as modulation of a square-wave carrier at the bit rate. Dicklyon (talk) 04:47, 8 January 2019 (UTC)

OK, but in spectral space, it looks like a 10MHz carrier modulated by (multiplied by) a signal that is either +1 or -1. The modulating signal goes from 0Hz to 5MHz, the latter alternating 1's and 0's. With appropriate filtering, which I believe Ethernet requires, the spectrum is from 5MHz to 15MHz. As usual for AM signals, the spectral space is twice the modulating spectral space. On the other hand, for broadband Ethernet, the carrier frequency is much higher. Baseband telephone signals are considered 300Hz to 3400Hz, again not down to zero. I think we need a definition that allows for signals that don't go all the way to zero, but close enough. Gah4 (talk) 07:02, 9 January 2019 (UTC)

Actually the "main lobe" of the spectrum is from 0 to 20 MHz, which is where the first nulls are, since the modulating signal's first null is at 10 MHz (it's not filtered to 5 MHz bandwidth, though it theorertically could be). The definitions in the book I link in the next subsection, which allow for "line codes", seems like what we need (not sure if he states a suitable def explicitly). Dicklyon (talk) 16:10, 9 January 2019 (UTC)

OK, back to the "very narrow" in the first sentence. I suppose "narrow" is always relative, but baseband 10G Ethernet will not be "narrow" in just about any sense. Next, how close to zero the spectrum gets also depends on its bandwidth,as it should. 1G Ethernet might only come within 10MHz of 0Hz, but still be baseband. (I actually don't know what 10G spectrum looks like at all.) Baseband telephone signals might only go down to 300Hz, but I suspect people will complain if you add in a 100Hz signal. Gah4 (talk) 23:39, 9 January 2019 (UTC)

"Narrow" is relative to carrier, if there is one, or to upper band edge; baseband Ethernet is wide relative to carrier or upper edge. Dicklyon (talk) 06:09, 10 January 2019 (UTC)

As well as I know it, and I well mention somewhere else, broadband came with the need to make an IF strip for TV receivers to amplify the wider bandwidth with appropriate frequency response. (There are complications from vestigial sideband modulation.) Radio signals don't require the complicated IF strip, and so qualify as narrowband. Gah4 (talk) 04:28, 2 May 2023 (UTC)

A possible resolution

To me, baseband can mean centered at zero, or modulated such that the bandwidth extends essentially to zero (like baseband Ethernet). I'll look for sources... This book might be useful, discussing various "line codes" including Manchester encoding, which don't quite go down to DC. Dicklyon (talk) 00:00, 8 January 2019 (UTC)

I think that sounds closer than the current article. It now seems, even with removal of "narrow" to suggest narrowness. It says non-zero near f=0Hz, but 10G Ethernet is going to be non-zero up to somewhere in the GHz range. If there are significant frequency components near, relatively, to zero it is baseband. OK, the bottom of the traditional US TV band is channel 2 at 54 to 60MHz, a nice representative non-baseban signal. The bottom of the spectral range is 9 times the bandwidth. Is that a reasonable transition point? Gah4 (talk) 09:03, 10 January 2019 (UTC)

Yet another attempted definition (Sept 2021)

I followed a link to this article to check that baseband meant what I thought, and having read the article and this discussion, I'm still not sure! Here's what I expected it to mean:

• The information being transmitted has a certain frequency band intrinsic to it: for example ~0 to 20 kHz for audible sound.
• The information can be represented by a signal having the same frequency content as the information, plus a suitable margin for good reproduction. This is a baseband signal.
• However, this typically isn't a suitable band for transmission (good luck trying to use audio frequency radio waves, for example . . . ), so the signal is converted to a higher frequency band for transmission, and then needs converting back to baseband.

If this is correct, I think it gives us a couple of possible definitions:

• A baseband signal is one whose frequency spectrum is unshifted relative to that of the information that it carries.
• A baseband signal is one which doesn't require any frequency shift before being used for its final purpose.

In other words it's the basic frequency range of the thing we're trying to transmit, and the frequency range wanted by the subsequent electronics. Frequencies in the baseband signal represent themselves. It's "narrow" in the sense of being no wider than it reasonably needs to be to encode the information, and if it's close to zero then that's because the frequency band of the encoded information is close to zero. And we can define it in terms of its purpose, rather than how many kHz or MHz counts as "narrow" or "close to zero". Musiconeologist (talk) 20:00, 14 September 2021 (UTC)

I pretty much go along with your thoughts. Here are some actual sources backing that up,

• "Baseband: A method of communication in which a signal is transmitted at its original frequency rather than being impressed upon a carrier frequency." [1]
• "Baseband: Definition: Refers to a basic set of frequencies of an RF signal prior to any modulation: essentially, an unmodulated signal." [2]

Furthermore, many sources contrast baseband, the unmodulated signal, with passband, the modulated signal,

• "One of the most fundamental operations in radio communications is the act of modulating a carrier wave by a band-limited data signal. Modulation converts a baseband signal to a passband signal." [3]
• "Modem is a contraction of modulator–demodulator, and expresses the fundamental purpose of the modem as a device to convert a baseband digital signal into a passband analog signal that can be reliably transmitted over a passband channel." [4]

Like others above, I also take issue with the claim baseband is narrow. That is at best meaningless and at worst utter nonsense. In terms of fractional bandwidth, baseband is actually wider than the modulated passband.

Another issue with the lead is its talk of a cutoff frequency. Baseband signals rarely have a definite cutoff. This usually has to be artificially imposed by the processing circuitry in order to get the signal through the transmission channel. SpinningSpark 23:57, 14 September 2021 (UTC)

That makes some sense for audio, which has a natural frequency range. It makes less sense for others, like most digital data. Since we can't hear below about 20Hz, someone might argue that audio isn't baseband. I believe some radio systems use subsonic signalling, so making use of that small bandwidth. There used to be a WWV radio transmitting at 20kHz. I believe the lowest one now is 60kHz. There also used to be (in the cold war days) radio transmitters to submarines at 7Hz. (It is the resonant frequency of the earth for EM waves. And the data rate is very low.) The problem is that not all signals, especially digital ones, have an intrinsic frequency band. It is not so obvious how to assign intrinsic frequencies to video. Gah4 (talk) 00:10, 15 September 2021 (UTC)

I'm not clear what you think makes sense for audio and not for digital data. Is it my claim that baseband is not narrow? If so I disagree and will expand on that if asked. It is also incorrect that baseband audio signals do not go below 20 Hz. Air pressure vibrations can get arbitrarily close to zero. The fact that we process that signal to remove the parts that are not useful to make transmission easier is neither here nor there. Using the consequent LF gap for signalling (as was done on POTS phones]] is even less relevant. My relevant point was that we should go with what the sources say and they all pretty firmly define baseband as the unmodulated signal. SpinningSpark 08:25, 15 September 2021 (UTC)

In the case of audio, we know what an unmodulated signal is. We have a voltage proportional to acoustic pressure, or one of its derivatives, possibly with some pre-emphasis. (I ignore any derivatives or filtering that might be done.) But what is unmodulated in the case of data? We can't stuff physical 0's and 1's down a wire, we need a voltage. Is unmodulated some voltage for 0, a different voltage for one? In that case, baseband Ethernet, with Manchester coding (which is actually synchronous phase modulation) is not baseband! Many data transmission systems use modulation to get a clock through to the receiver, along with the data, not because they need a better frequency for (often) radio transmission. As for baseband audio, I was considering audio as only what we can actually hear, as that is (usually) the point of audio. Gah4 (talk) 20:35, 15 September 2021 (UTC)

I think maybe I muddied things by thinking in terms of information content—the real point is that there's a signal in some useful form, which gets transmitted in a different form, and we need it back in the useful form. Details of how it relates to the information represented aren't relevant, but helped me see why "a narrow band near zero" doesn't make sense. "Unmodulated signal" seems accurate to me, and also gives scope for making the lead section more explanatory for non-technical readers. Musiconeologist (talk) 12:29, 15 September 2021 (UTC)

In the case of audio, it is somewhat obvious what the unmodulated signal is. It is less obvious in other cases. Especially since, as above, baseband Ethernet isn't unmodulated. Gah4 (talk) 20:35, 15 September 2021 (UTC)

I've rewritten the lead to get rid of, or at least be more exact about, close to zero. I've also taken out a lot of other dubious stuff. SpinningSpark 21:05, 15 September 2021 (UTC)

Close enough for me. Gah4 (talk) 23:21, 15 September 2021 (UTC)

A vast improvement! Many thanks Musiconeologist (talk) 00:17, 16 September 2021 (UTC)

Spinningspark, thanks for the improvements ~Kvng (talk) 15:36, 19 September 2021 (UTC)

Signal that has a very narrow frequency range near zero

It seems to me that Signal that has a very narrow frequency range near zero leaves out what very narrow is relative to, and also near zero leaves out how near. This question comes up in real baseband signals. Start with traditional phones, which are spec'ed as 300Hz to 3400Hz. OH, 3100Hz is pretty narrow, but 300Hz is a ways from zero. For an actual real case, traditional CATV put broadcast TV frequencies into a coaxial cable, so down to 54MHz. Then later there was need for an uplink for pay-TV cable boxes, so they found space below 54MHz, to add it. The cable system amplifiers have to amplify both directions, so have to separate out the two directions first. Because filters are imperfect, you can't get all that close. In any case, broadband CATV has plenty of room below. In any case, baseband rarely gets close to zero, and rarely is narrow. Gah4 (talk) 20:11, 11 August 2020 (UTC)

A way forward from all of the questions about the rigor or clarity and accuracy of this article

To attempt to answer some of the questionss about this sort of slippery topic.

I have looked at the web for a better definition and so far haven't found one so I'm going to try for a 'more rigorous' definition of baseband and maybe an additional section should be added to appeal to engineers, physicists, and mathematicians. I don't think a perfectly rigorous definition is possible so we resort to some logic and an admittedly heuristic definition.

OK so the first problem is the very first figure in the article as it is, has no figure designator and it needs to be referred to in what should be added to this page. Lets call it Fig. 1 here.

OK so Fig. 1 is actually a spectrum for an analytic signal for which a good wiki article exists. So then to start this new 'more rigorous' heuristic section we must identify Fig 1 as the spectrum of an imaginary (complex or analytic) signal and redraw an additional version as the amplitude spectrum of a real signal, such spectrum as an even function adding its mirror image in a new figure. Then explain that the purpose of an analytic signal is to ignore negative frequencies of real signals for arriving at simpler depictions and analyses as in Fig 1.

OK so then we have an amplitude spectrum which is an even function which is a necessary but not sufficient attribute of a baseband signal. In addition to being an even function, every frequency point in the supported spectral domain has one and exactly one nonzero alias at its negative frequency, of identical magnitude. This works for all forms of modulation except SSB.

So then to try to deal with the SSB case we can add that for a baseband as distinct from its SSB form the spectral third moment about the second moment, is of similar magnitude as the second moment about the mean frequency (zero). This works except for extremely narrowband SSB signals e.g. sinusoids which carry no information beyond steady state frequency and phase. So if we have a spectral standard deviation $$\omega _{\sigma }=\left[\int _{-\infty }^{\infty }\omega ^{2}\,F(\omega )\,\mathrm {d} \omega .\right]^{1/2}$$

and define

$$\omega _{\lambda }=\left[\int _{-\infty }^{\infty }{\frac {(\omega -\omega _{\sigma })^{2}}{2}}\,F(\omega )\,\mathrm {d} \omega .\right]^{1/2}$$

then if

$${\frac {1}{5}}<{\frac {\omega _{\lambda }}{\omega _{\sigma }}}<5$$

this would be an arbitrary exclusion of narrowband SSB signals

So then we verbally exclude narrowband SSB signals such as sinusoids. This final condition being strictly logical regrettably.

The 'regrettability' arises from the fact that a SSB modulation of a sinusoid is another pure sinusoid which has no feature described as baseband. So we have to apply this last attempt using the above equations to actual information bearing signals and any of their modulated representations, which as quasi-stochastic would only be characterised in the frequency domain with spectral density as opposed to a Fourier transform amplitude.

Comments welcome.

— Preceding unsigned comment added by Groovamos (talk • contribs) 23:48, 7 January 2022 (UTC) Groovamos (talk) 03:10, 8 January 2022 (UTC)

We're writing an encyclopaedia for a general readership here, not a textbook. Why do you even think that a formal mathematical definition is required? Baseband is a straightforward, simple, basic concept that can be stated quite adequately in words. There's no way that any of that is going in the article, especially the very arbitrary last bit, in the absence of reliable sources backing it up. Even then, I would probably be against it – introducing difficult mathematics for our readers when it is not necessary to do so. Even my textbook from university (Goodyear, Signals and Information), which is chocked full of integrals of spectral densities and the like throughout, does not feel the need for such a formal definition of baseband. SpinningSpark 10:02, 8 January 2022 (UTC)

Who on here said that a formal definition should be required? It seems yours truly just posted a section that indicated a formal definition is not possible, if one cares to read it. I have a Ph.D friend, physics, who read the article yesterday and was confused by it. So would it be worthwhile to tell the reader that a formal definition is not possible, made moot in the SSB case? I think I just explained why.

From comments above may I quote:

"What combination of idiots wrote this article? It makes no sense!"

"I second the opinion. This article is not only useless, but also MISLEADING. The Opening paragraph on the Baseband is totally misleading."

"Just reading the first sentence" Baseband is a signal that has a very narrow and near-zero frequency range, raises questions to me. First of all, very narrow relative to what?"

"Let's work on a good opening sentence and definition (current version uses "comprises" in the typical incorrect way, and doesn't get to the point very specifically). It would be good to at least admit that any kind of Fourier analysis gives negative frequencies, too. For real signals, these are just reflections of the positive, not new information...."

Maybe one has a constructive comment to add to all this. There are people in the sciences and others who might appreciate that.

Groovamos (talk) 18:11, 8 January 2022 (UTC)

That has not been the opening sentence of the article for five months. The earlier discussions are now completely out of date. SpinningSpark 22:14, 8 January 2022 (UTC)

Oh OK so not clear if the opening sentence change revamped the article and made all ssimilar comments out of date. A sentence not out of date would be any of the ones heard on Friday from one friend with a physics Ph.D., if one cares to reread why I came on here. Groovamos (talk) 21:19, 9 January 2022 (UTC)

Short description

The current Short description range of frequencies occupied by an unmodulated signal is too long and needs to be shorted per WP:SDSHORT and formatted correctly per WP:SDFORMAT. Two attempts to resolve this have just been reverted rather than improved. Please attempt to improve the SD — GhostInTheMachine ^{talk to me} 22:06, 16 June 2022 (UTC)

Too long is preferable to wrong. This is an encyclopaedia, not online scrabble. Also, there is an interesting conversation going on at Talk:Waveguide/Archive 1#Short description – we cannot find views or examples of where an over-long SD has actually caused a display problem. SpinningSpark 22:14, 16 June 2022 (UTC)

Trying to improve an article is generally the best way forward. See WP:SDSHORT, WP:SDFORMAT, WP:SDNOTDEF — GhostInTheMachine ^{talk to me} 22:25, 16 June 2022 (UTC)

As I teach the subject, to electronics engineers and technicians, I did some editing of the article for clarification. The opening line is fine I feel. The rest of the article seems a bit broad with the inclusion of RF modulation using baseband signals, however, it just passes for relevance. I agree that as soon as you put mathematics into what can be described using words, it alienates most of the reading population, so during my lessons in RF Modulation, transmission and reception, I rarely use mathematics and certainly not for an encyclopaedic article, but that's just me. --Read-write-services (talk) 11:37, 17 June 2022 (UTC)

I've started a central discussion at Wikipedia_talk:Short_description#Too-short_descriptions ~Kvng (talk) 15:45, 20 June 2022 (UTC)

Broadband

It seems that the article Broadband is now only about high-speed internet, and lost is the application to radio (well, TV) transmission and reception, and especially the IF strip required for such receivers. When the bandwidth is large compared to the frequency, and that would be IF for many receivers, it complicates the design. That term then continued on into data communications, but now the original use is lost! Gah4 (talk) 04:34, 2 May 2023 (UTC)