ECN Forum
Posted By: Frank Cinker Bandwidth - 11/26/01 08:07 PM
What is a "simple" definition of bandwidth? Is this term only used with fiber optic cable?
Posted By: Joe Tedesco Re: Bandwidth - 11/26/01 08:49 PM
http://www.auburn.edu/helpdesk/glossary/bandwidth.html

Bandwidth - Bandwidth (the width of a band of electromagnetic frequencies) is ordinarily means how fast data flows on a given transmission path. Specifically, it is the width of the range of frequencies that an electronic signal occupies on a given transmission medium. Any digital or analog signal has a bandwidth. Bandwidth is measured in Megabits per second.

[This message has been edited by Joe Tedesco (edited 11-28-2001).]
Posted By: Frank Cinker Re: Bandwidth - 11/26/01 09:33 PM
Thank you Joe.
Posted By: pauluk Re: Bandwidth - 11/26/01 09:43 PM
That reference mentions a general definition that bandwidth is the range of spectrum taken by a signal, but otherwise it seems to be looking at it solely from the point of view of digital signal transmission.

I'm not too happy with the "bandwidth is measured in megabits per second" either. The latter is the rate at which data is transferred, and although everything else being equal a higher data rate requires a greater bandwidth, there are other variables in the equation.

In its purest form, bandwidth is the measure of the difference between the lowest and the highest frequencies employed to transmit the signal, be it analog or digital, and is measured in Hertz (or kHz, MHz, etc).

The term was in use long before fiber-optic cable was invented, and can refer to a signal sent down a twisted pair of wires, a coaxial cable, a microwave waveguide, or broadcast through the air.

For example, the signal sent down a regular phone line is normally limited to a range of about 300 to 3400Hz, a bandwidth of some 3.1kHz, while a good hi-fi system might have a bandwidth of 20kHz or more.

For comparison, an AM broadcast radio signal has a bandwith of 9 or 10kHz, an FM radio signal 200kHz, and a TV signal (U.S. system) about 6MHz.
Posted By: Scott35 Re: Bandwidth - 11/26/01 10:31 PM
Paul,

Very good explanations!!!

I also felt the reference was making more of an emphasis on Digital packets, rather than explaining an Analog envelope [which as you know is rather complex as compared to a packet].

The reference from Joe is just what Frank needs to use for FDDI LAN tech, so it's probably better on that angle to keep within the Digital / Binary spectrum as opposed to an explanation of complex Analog envelopes.

Once again - just my $0.02 [Linked Image] [Linked Image] [Linked Image]

Scott SET
Posted By: pauluk Re: Bandwidth - 11/26/01 11:44 PM
Scott,

I guess almost everything these days is now angled toward digital signals, as this definition clearly illustrates.

I won't even start the argument about whether to measure bandwidth at -3dB, -10dB, etc. [Linked Image]

For anyone still pondering this, just consider that today's 56.6kbps dial-up modems operate over the same 3kHz telephone bandwidth as did the 300 bps modems of yester-year.

Which brings to mind another related point to remember: Despite what some advertisement may say, bits-per-second and baud rate are NOT the same thing.
Posted By: Frank Cinker Re: Bandwidth - 11/27/01 02:24 PM
Scott,

The information Joe provided was exactly what I needed. Beyond that, it's well over my head. I do have one more question. In general, doesn't fiber optic cable provide the greatest bandwidth? That is to say compared to Category 5 UTP cable etc.
Posted By: ElectricAL Re: Bandwidth - 11/27/01 03:51 PM
Frank,

You're right. In the most basic sense, light on a fiber optic strand has a greater (much greater) bandwidth than voltage and current on a unshielded twisted pair (UTP) of CAT5e or "CAT6" copper conductors.

How much greater is roughly indicated the cycles per second of the light verses the maximum cycles per second of the wire. On Wire, electrical frequency can go up to 200 to 300 million cycles per second before being distorted beyond recognition by the wire's total impedance. Light's frequency is roughly a MILLION times greater than the frequency of the electricity in common UTP.

The way that information is changed into a signal in the electricity or light is in different stages of development, UTP vs. fiber, as is the hardware that modulates the electricity or light, so the million-fold difference above may be less or greater depending on the individual schemes compared. But basically, fiber is a fat pipe for information compared with UTP.

Al
Posted By: Bill Addiss Re: Bandwidth - 11/27/01 06:10 PM
Good question Frank!

I'm a bit 'stuck' on fathoming this word "Bandwidth" as it is used. As Paul described it I can understand how it is a measurement of the range of frequencies employed in the transmission of a signal.

What I am having a problem with though is how it relates to the transmission speed. It seems to me that the word "Bandwidth" should not be measured in Mps. Or that they are not necessarily that directly related.

Al,

I understand your analogy that "basically, fiber is a fat pipe for information compared with UTP." I find it interestingly ironic that the "fastest" Fiber is actually the thinnest.

Bill
Posted By: ElectricAL Re: Bandwidth - 11/28/01 02:14 AM
Bill,

I, too, stand in awe at the information flow that a little hair of glass is capable of.

Speaking to transmission speed. . .when information is crunched by electronics into a binary stream of "on" and "off" and the binary stream is compressed by an exotic algorithm into a shorter, simpler binary stream, there is additional bandwidth gained over the basic analog bandwidth. The better the compression algorithm, the faster a fixed block of info will travel over a connection of fixed analog bandwidth. The actual signal on the wire or fiber is traveling at near light speed, but that is not the speed in consideration. Rather, it is how long it takes to disassemble a piece of info into a binary code, manipulate the code into a minimum of actual bits to be sent down the pipe, sending down the pipe, reversing the manipulation recreating the original binary code (with error detection and correction) and, finally, recreating the info.

So, there are at least two "bandwidths", the analog and the digital, that together work to develop a "transmission speed". Various noise will create errors that slow down the overall transmission by causing retransmission.

Sounds like a transporter. Beam me up Scotty. [Linked Image]
Posted By: pauluk Re: Bandwidth - 11/29/01 12:37 AM
Bill,

I agree with you that bits-per-second is not a measure of bandwidth -- It is a measure of the rate of data transfer.

The link between the two is that when using any given encoding system, a higher data rate requires a wider bandwidth. But that's about as far as you can go without specifics. To accurately determine the required bandwidth for any given data rate you need details of the encoding system - and the relationship between the two can vary considerably.

To me, saying that bandwidth is measured in bps (or Mbps) is like telling someone that wattage is measured in amps. We all know that that's technically incorrect, even though there is a relationship between the two. But without knowledge of the system voltage, power factor, etc. it's impossible to calculate one from the other. Same with bandwidth and data rate.
Posted By: Scott35 Re: Bandwidth - 11/29/01 05:22 AM
Very good discussion thread going on here!!!

Like most everyone else here, I was introduced to the concepts of Bandwidth as an Analog wave format [mine was in Radio, then light, then Telephone, then Audio {sound waves}], way before I was introduced into pure Digital transmission.

I could really start the thread burning by throwing in the Carrier Waves for the transmission!

In the FDDI, we have bands of Infrared or Ultraviolet light, in which the Binary Code is carried and constructed of. These Carriers have bandwidths, so do the pulses themselves.

In normal UTP, or Coaxial LAN cables, the Electricity has bandwidth [let's even look into the multivibrator / clock circuits and the light created from currents flowing through the Semiconductors or even the cables].

This stuff would make for some really in-depth discussions [fun for me!!!]

To answer Frank's latest question;

Yes, in general the Bandwidth between the two items you listed are large!

Fiber Optic LAN Backbones would be in the GHz range [Giga Hertz, or 1,000,000,000 Hz] and could carry many packets on multiple channels within the same Fiber.

UTP LAN cabling would be in several speeds:

* For 10 base T - Tolken Ring [IEEE 802.5], the speeds are either 4 MBPS or 16 MBPS. The slower can accept CAT 3 UTP cable, the later uses CAT 5 UTP cable.
Tolken Ring is a Multi Tolken, limited packet network topology, which can use either Star or Daisy Chain connection schemes between Workstations and a MAU [similar to a Hub]. Between MAUs and Servers, there are in/out "Rings"

* For 10 base Ethernet systems [IEEE 802.3], "Thicknet" was 10 base 5, "Thinnet" was 10 base 2, and of course 10 base T uses Twisted Pairs. 10 base T is most recently of CAT 3 UTP - which is IEEE tested for the maximum speed of 10 MBPS.
10 base [typical Ethernet] is a single tolken - single packet tpoplogy in the basic forms. In the "Full Duplex" modes, there can be multiple packet exchanges [which increases the bandwidth by 2x].

* 100 Base T [or TX]:
Typical to the 10 base system, except has a transfer speed of 100 MBPS, and requires cabling with that rating.
CAT 5 UTP and CAT 5E UTP are usable with 100 Base TX Ethernet LANS.
Typical "10 base" systems only use two of the four pairs in the cable.

Others are Catagory 6 [coax], and Catagory 7 [Fiber]. also there's a few 10 base fiber links, such as 10 base F [the ethernet fiber LAN standard], 10 base FB, FL and FP.

If this is an interesting area, maybe we can collect and complie a cache of LAN Tech information to post in the Reference section.

Let me know if this sounds like an idea - plus if you wish to contribute.

Scott SET
© ECN Electrical Forums