From Wikipedia, the free encyclopedia
Jump to navigation Jump to search
WikiProject Measurement (Rated Start-class, High-importance)
WikiProject iconThis article is within the scope of WikiProject Measurement, a collaborative effort to improve the coverage of Measurement on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Start-Class article Start  This article has been rated as Start-Class on the project's quality scale.
 High  This article has been rated as High-importance on the project's importance scale.
WikiProject Telecommunications (Rated Start-class, High-importance)
WikiProject iconThis article is within the scope of WikiProject Telecommunications, a collaborative effort to improve the coverage of Telecommunications on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Start-Class article Start  This article has been rated as Start-Class on the project's quality scale.
 High  This article has been rated as High-importance on the project's importance scale.

/N888/B/A/ Telegraph Baud /2006/11/21/[edit]

Dear caretaker team of the "Baud" WIKIPEDIA collaborative webpage, I found the following page's "Jean-Maurice-Émile Baudot, and teleprinting" section very useful (from which I quote below):

In the (original) telegraph case, a system's Baud is the electro-mechanical maximum signal rate, and since each signal is one bit and about 1/3 are used-up for overhead (see below), the user sees a bits per second transmission rate that is less than the Baud of the transmission system. Shame on the long history of ignorant (or worse) modem speed sales pitches to consumers for the confusion.

If each signal was set and measured more precisely, say at four levels (0,1,2,3) instead of two (0,1) (as was quickly figured-out for modems but not inside computer processors for some reason) the Baud of the system would stay the same while the theoretical transmission speed of the system would increase by eight times, as each signal could represent one of sixteen states instead of two. AVOID CONFUSION, USE BITS PER SECOND, NOT BAUD. NATE88 BETTERDIFFERENT DOT COM N888

Normally, when current is flowing, the line is said to be “marking”, and the machine is sitting there with the motor running, waiting for a character to arrive. To prepare the machine to receive a character, a special zero bit is sent, called the start bit. This bit starts the wheels of the machine turning, and the machine counts and records each of the five bits as they arrive on their fixed schedule. When all five bits have arrived, the bits are used to position a print head or piece of type, which strikes the paper through an inked ribbon, marking the character on the page. After the last bit has arrived, the line is kept in the one state (marking) long enough for the decoding mechanism to reset. In Baudot teleprinters, this was 1.42 times as long as the bit time. It became known as the stop bit, although it was required at the end of each character and really did not stop anything.

With the start bit, five data bits, and 1.42 stop bits, the Baudot code took 7.42 bit times to print each character. Although the speed of different machines varied, a common standard was for a bit time to be .022 second, for a rate (in Baud) of 45.5 bit times per second. At this rate, a teleprinter could print just over six characters every second, so if you are printing five letter words with spaces in between, that is just over 60 words every minute, a very respectable typing speed, and faster than almost all telegraph operators.

Isn't it true to say that "baud rate" actually refers to the potential number of signal changes/symbols transmitted per second? Since, if one second's worth of 0's are transmitted for some reason, the baud rate of the given channel isn't 0, is it?

done -- guest

"baud rate" is a redundant phrase. "baud" is the commonly accepted word for "symbol rate", in so many symbols per second. a "baud rate", then, is a "symbol rate rate", which can only imply symbol acceleration (equivalently, deceleration).

done -- Tarquin 16:48, 27 Sep 2003 (UTC)


what is 'symbol rate rate' ?

I didn't ask the above.... but, better still ..... what is a 'symbol'? I'm thinking they should first define 'symbol' before moving onto 'symbol rate', eg. a diagram or an actual example of various symbols, rather than just described by words etc. As an example, a sinusoidal signal has a particular peak amplitude, which can be used as a reference signal to begin with, which can be transmitted by the transmitter, and received by the receiver. The receiver uses this signal as the reference. Now, if the transmitter then decides to alter not only the peak amplitude of the sinusoidal signal, but also decides to change the 'angle' (ie. phase) of it, then this can be recognised by the receiver as a change in two parameters .... amplitude, and angle. So each 'symbol' could be considered as a unique combination of amplitude and phase. Each unique combination of sinusoidal amplitude and phase (all relative to a reference sinusoid) is a symbol. And, each symbol can be defined to mean something ... such as '0001' or '1000' or '0011' or '0110' etc. This just means that the successful receiving of one 'symbol' can mean receiving 4 binary bits of data in one shot, which is a lot better than the early 'old-school' communication methods having being able to receive just 1 binary data bit in one shot. KorgBoy (talk) 08:01, 17 May 2018 (UTC)


'symbol rate rate' == the rate of change of symbol rate. In this context, it was probably used to illustrate why term "baud rate" is, more often than not, nonsensical.

"a 2400-bit/s modem actually transmits at 600 baud, where each quadrature amplitude modulation event carries two bits (four values) of information."

600 events/second * 2 bits/event = 1200 bits/second, right? [drd]

yep 600baud with four values which means log(4)/log(2)=2 bits per value, if there is 600 of this values in one second then it transmits 1200bits/second

No, "symbol rate rate" is a literal interpretation of the phrase "Baud Rate". I chose that description to illustrate my point that "Baud Rate" is a redundant phrase. "Baud" was originally 'symbols per second' or by extension, 'symbol rate'. Some people may think they will seem technically savvy when they use the word Baud, but those who attach 'rate' demonstrate they don't really understand what they're saying. — Preceding unsigned comment added by J 0JFCfrmAyw59oVFk (talkcontribs) 23:20, 29 November 2017 (UTC)


I recommend the addition of a statement to the effect of "Modems have been errounously marketed for years and years using the term 'baud' when in fact the measured speed is in bps". Cause I've bought 9600baud, 19.2Kbaud, 33.6Kbaud and 56Kbaud modems. I've learned a couple times that baud=bps. Bleh. thought the wikipedia way was to acknowledge and refute common misconceptions rather than delete them. TomCerul 18:55, 10 October 2005 (UTC)

Yes. It seems to me that one could determine the baud rate on the serial interface, which is what is important to the user. What goes out on the cable is less important. So, I set the UART clock generator for 9600 bits/second, otherwise known as 9600 baud, though the symbol rate coming out of the modem is much lower. Gah4 (talk) 06:08, 12 May 2017 (UTC)

What about the null state?[edit]

The example of a semaphore flag having 8 possible states and thus conveying 3 bits of infomation each second ignores the situation where there is no information in those 8 states to convey and thus the semaphore flag isn't hoisted or is held straight out; e.g., a space between words, no answer yet, etc. The 3-bit-per-second signal rate only holds when a signal is being sent, but the pauses between signals may be equally significant. This is fine where the signals are totally asynchronous, but where the signals are synchronous, there needs to be a null state, which would require 4 bits.

You are quite right. I have modified the text so that only eight states are required now. 14:42, 23 January 2006 (UTC)

Excellent work[edit]

Excellent analogy with the semaphore, cleared it right up in my head. Have a crisp. -- 17:24, 2 March 2006 (UTC)

Crunch! Thankyou! WLD 23:22, 2 March 2006 (UTC)

I agree. Excellent analogy. Thanks! Sepia tone 10:41, 6 November 2007 (UTC)

Baud for spread spectrum systems[edit]

What is the baud for a system such as CDMA, where the channel is modulated at chip rate, with each data symbol comprised of several chips and each data symbol carrying multiple bits, using PSK or QAM to convert bits to symbols? Does it even make sense to ask what the baud is, or is baud only meaningful for a point-to-point channel carrying a single modulated datastream?

Forgetting about the details of the technology for a moment, what is the rate at which symbols are transferred between the sender and receiver? Each symbol may contain multiple bits of information - in the semaphore example, each symbol carries three bits of information. WLD 08:13, 1 September 2006 (UTC)

Common baud values[edit]

The “common baud values” section is unnecessary and confusing, and should be deleted. It appears to be a list of common RS-232 DTE rates lumped into arbitrary categories, and has absolutely nothing to do with “baud.”

Agree. WLD 23:00, 25 September 2006 (UTC)
I have removed the section. --Morten LJ 15:40, 19 October 2006 (UTC)

Merge with symbol rate?[edit]

The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.
Majority against merge. -- P 1 9 9 • TALK 17:30, 13 September 2011 (UTC)

Having proposed the merge, it would be a good idea for the merge proposer to put the reason why here on the talk page. In the absence of a good reason, I can see no benefit. Hence:

  • Oppose WLD 12:41, 17 November 2006 (UTC)
    Wait a second: you oppose your own merge proposal? Then why exactly are you proposing it? Snoutwood (talk) 19:25, 19 November 2006 (UTC)
    Oh, I apologise, I see now that you didn't propose it. I'm removing the tags since the proposer has not given any rationale. Snoutwood (talk) 20:45, 19 November 2006 (UTC)
  • Support. And also with modulation rate and symbol (Data). There are too many articles on this topic. The other articles are quite poor, and could basically be redirected to the excellent baud article.
  • Oppose There are articles for Electrical current and for Ampere; for Atmospheric pressure and for Bar (unit); for Electrical potential and for Voltage. I see no reason to merge an article for the unit of measure with the article for what it measures. If the modulation rate and Symbol (data) articles are poor, perhaps they should be merged into Symbol rate. ThreeOfCups (talk) 20:35, 14 September 2009 (UTC)
  • Support The article about symbol rate is more or less the same, let's not put two copies of the same information; that takes double work to keep them up to date and is confusing. Make sure redirects are made for baud rate to symbol rate. -- (talk) 23:07, 12 January 2010 (UTC)
  • Oppose I find this to be a useful quick/ready reference. If merging meant that this page would be removed, that would be a backward step. The term BAUD is still in common use although its ambiguity is not always taken into account. A little more history (eg about the old teletypes that used the 5 bit Baudot code), or a reference, would add clarity).
  • Oppose. This is useful information that can be expanded upon. Symbol rate is archaic. —Preceding unsigned comment added by (talk) 22:09, 13 October 2009 (UTC)

The above discussion is closed. Please do not modify it. Subsequent comments should be made in a new section.


What is that last line of the first paragraph? "this lik3 sucks init" 13:02, 5 December 2006 (UTC)

I have fixed that. XU-engineer 19:00, 11 June 2007 (UTC)

Baud rate[edit]

There's a discussion about the differing meanings of baud and baud rate at User:Dicklyon's talk page. We really should have conducted it here, but I don't think it would be right for me just to copy it. Page moved back to Baud. WLDtalk|edits 11:40, 24 January 2007 (UTC)

I'm OK with the move back, but let's not eliminate the "baud rate" usage entirely. I added it back. Dicklyon 04:08, 25 January 2007 (UTC)
OK. I did refer to it slightly later in the article, but more references aren't a problem. Incidentally, if you move articles, it's normally good practice to check what links to the article (use the "What links here" link on the left) and edit all the linking articles so they point directly to the moved article and don't have to go through redirects. Obviously, in this case, I'm glad you didn't, because it made the task of moving back easier, but I just thought I'd let you know. Regards, WLDtalk|edits 10:52, 25 January 2007 (UTC)
I had fixed up the multiple redirects, but not the singles. There are usually too many of those to be worth wholesale fixing, and they are apparently already acceptable in their contexts, so I usually just fix a few that I care about. Dicklyon 16:03, 25 January 2007 (UTC)

Sampling rate[edit]

In some books ( Tannenbaum, computer networks as example ), I 've found the asseveration that the sample rate should correspond to the boud rate. However, if the boud rate should be interpreted as the symbol rate that is not necessarily true. If you are transmitting by modulated amplitude it is possible that you should have to obtain many samples to obtain one symbol, is not it?. I would appreciate if someone who knows about the subject could explain this.

It would be more correct to say "at least". If you start with many samples, and compute amplitude from that, you can then reduce to one sample per symbol. If the amplitude detector is analog, you can sample its output at one sample per symbol; more typically, with coherent PLL-based detectors in analog, sampling one sample per symbol is very typical, with a symbol-sync PLL to keep the sample phase right where you want it. At least, that's how I used to build them back in the 1970s; nowadays starting with more samples and doing all the synchronization digitally is probably more common. Dicklyon 15:17, 29 March 2007 (UTC)

Baud (comment)[edit]

I have just been reading your article on the telegraph term "Baud"

When I worked on radio telegraph in the 50s & 60s we were taught the definition of Baud was "the speed of the shortest element expressed in bits/second." We were taught that the term was only to be used where elements of a signal were of different lengths - Bits/Sec was to be used where the elements were all the same length. In those days machine telegraph in the ICAO network used the Baudo code in which the start and codes elements were the same length and the stop element was 1.42 times the length of the others. With codes like this BPS was not an appropriate term so Baud was used.

When was the definition of Baud changed to allow it to be used for signals with all elements of the same length?????

Peter Davidson peterd at

In my opinion, any discussion about "symbol rate" should begin with an agreement on the definition of "symbol". So it is a case-by-case thing. So, at the beginning, in a particular discussion, if the symbol is defined as 1 binary bit, then transmission of 1 bit per second will amount to 1 baud. And, at the beginning, of another discussion, if the symbol is defined as 10 binary bits (per serial transmission codeword including start bit, stop bit, parity bit etc), then transmission of 10 binary bits for each second will amount to 1 baud. KorgBoy (talk) 03:44, 22 June 2017 (UTC)


  • How do calculate the baud rate? XU-engineer 19:01, 11 June 2007 (UTC)


From International System of Units

Symbols for units are written in lower case, except for symbols derived from the name of a person. For example, the unit of pressure is named after Blaise Pascal, so its symbol is written "Pa" whereas the unit itself is written "pascal". Jim.henderson 00:28, 26 June 2007 (UTC)


In the article, it is misleading to discuss chips and spread spectrum in the same context as baud and data rate. In these contexts, the symbols do not increase the data rate, they actually lower the data rate since a larger number of code bits are used to represent a smaller number of data bits. This is different than quadrature amplitude modulation where the symbols represent specific data bits. It is true that the same principles of baud and bit rate apply, but then it would be helpful to discuss the difference between code rate and actual data rate which seems beyond the scope of this article. (talk) 19:29, 20 November 2007 (UTC)


Bits per second and baud are not interchangable. (Nor, for that matter, are bit and baud.) It happened that the Bell 103 modem transmitted one bit with each change of signal state (its baud), which occurred 300 times per second. Unsophisticated computer users in the late 1970s--the overwhelming majority of whom had 300-bps modems--confused the two and adapted "baud." (It's also likely that "baud" was preferred because it's easier to say than "bits per second" or "bps.")

Of course, once you move up from Bell 103's amplitude modulation technique to other modulation techniques (like QAM, or quadrature amplitude modulation) that can transmit more than one bit per baud (change in signal state), bps and baud don't match. 2400 modems used a technique that transmitted four bits per baud, which meant that they operated at 600 BAUD while transmitting 2400 BITS PER SECOND.

As another contributor here mentions, ths confusion was the basis for a lot of uninformed sales pitches.

The misuse of the term "baud" will never end, but the true meaning of the term, plus its relationship to bps must be preserved somewhere for reference. —Preceding unsigned comment added by (talk) 05:54, 5 April 2008 (UTC)

Sure it will end. Users of DSL modems and Cable modems are not told of a "baud" thus the term will die (except among people who design such equipment) with the last user of a voiceband modem, sometime late in the century. Jim.henderson (talk) 00:37, 17 April 2008 (UTC)
It is more complicated than that. Yes baud is used for the symbol rate on the modulated carrier, such as the audio output of a modem, but it is also used on the serial port of a computer, and the cable that connects (in the external case) to the modem. So, for the Bell 212 case, with two bits/symbol, the 1200 baud serial (usually RS232) port runs at 1200 bits/s, and 1200 symbols/s. Even more, Bell 212 is then called a 1200 baud modem, based on the serial port symbol rate, though the modulated carrier runs at 600 symbols/s. I suspect, as noted above, this is from the earlier days of 1 bit/symbol modems. It gets even more complicated when you have buffering and compression. Note also that in information theory, bits are used based on information transfer, and not on the symbol rate of a binary signal. Baud is pretty much never used for information theory bits/s, though. Gah4 (talk) 01:46, 30 November 2017 (UTC)

Survey vote: Should Baud rate/Baud, Symbol rate, Symbol (data) and Modulation rate be merged? Article name?[edit]

The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.
No consensus at Talk:Symbol rate. -- P 1 9 9 • TALK 17:30, 13 September 2011 (UTC)

Please discuss and vote at Talk:Symbol rate#Discussion of Suggested Merges. Mange01 (talk) 08:37, 6 May 2008 (UTC)

Baud, not "baud rate"

"Baud rate" is a redundant term. Baud is a rate, whether bit rate (bits per second), symbol rate (symbols per second), or whatever, but the term "baud rate" is a redundant use of rate. To use "baud rate" is to say "bit rate rate" or "symbol rate rate". It should be baud and only baud. No one uses kHz per second or MPH per hour or RPM per minute, yet that is what people who use "baud rate" are doing.

The above discussion is closed. Please do not modify it. Subsequent comments should be made in a new section.

The author of the article missed the point that the modems use serial communication[edit]

In the case of serial communication, the two bits "0" and "1" can be differentiated only by their duration. The duration of the "1" is double of the duration of the "0" (or viceversa, depending on how the signals are coded). Because of that, the duration of the transmission of a fixed number of binary digits is not constant, it depends on how many of them are "0" and how many are "1". An example, transmitting 100 digits of "1" will take double amount of time than transmitting 100 digits of "0". The bauds per secons is the speed calculated considering an equal number of "0" and "1" digits. From the 100% percent time, 66.6% will be spent sending "1" and 33.3% will be spent sending "0". — Preceding unsigned comment added by (talk) 02:06, 23 January 2012 (UTC)

No. There are lots of ways to encode bits that do not rely on a difference between durations, for example, quadrature amplitude modulation. It's a while since I've thought about this topic, but I imagine all practical systems use equal durations for each symbol (where baud = symbols per second) because the receiver can then synchronize with the transmitter and know when to look for another symbol, and because there is no reason to have some periods carry less information than others. Johnuniq (talk) 02:34, 23 January 2012 (UTC)
The audio tape recording format used by many microcomputers before floppies became popular often used the variable rate form described above. That is, FM with one cycle at either F or 2F. It would not work well for disks, where you need to be able to write over a block of the same length, but it works for tape. This is easy for software encoding/decoding, not so easy in hardware. So, yes, the symbol rate can be variable, in which case one normally gives the average. Gah4 (talk) 02:06, 30 November 2017 (UTC)

Removed pedantic paragraph[edit]

The common term baud rate is a misnomer since by definition baud is pulses or switches per second and baud rate would be pulses or switches per second per second or acceleration.

  • "Baud rate" means "[symbol] rate, measured in baud". - Mike Rosoft (talk) 05:09, 29 March 2012 (UTC)

What is the correct unit symbol for gigabaud?[edit]

See discussion at MOSNUM. Dondervogel 2 (talk) 15:00, 8 December 2014 (UTC)

Manchester coding and modified frequency modulation have a bit rate equal to 1/2 the baud.[edit]

Is there a reference for this statement? I know the meaning, but I don't believe that it is usually counted this way. Specifically, symbol and transition don't have to be equal. In Manchester coding, a symbol is either low followed by high, or high followed by low. Yes the bandwidth is twice what other coding methods would need, but it is self clocking, unlike some others. Gah4 (talk) 06:05, 12 May 2017 (UTC)

I now removed the incorrect statement regarding Manchester and MFM. For Manchester, a symbol is a full cycle at either 0 or 180 degrees. That is, high followed by low, or low followed by high. It is even more complicated for MFM, but symbol rate is not equal to transition rate in either case. Gah4 (talk) 02:01, 30 November 2017 (UTC)