"MB/s" / "mb/s"

[evilcow]

alright.. finally a new question has arrived (other then leds!!) hehe. what determines the megabits per sec or megabytes per sec.. in an electrical circuit. there should be some limitations. like lets say i have a 22g copper wire, could i find out it's MAX mb/s or MB/s?
or do i have to know what frequency it's rated at. let's say i have a cable that's rated at 350 mHz, what's the max mb/s or MB/s it can output?

i'm so confused! please can someone clarify this for me! thanks.

 

[Moohooya]

Digital signals are sent as 1s and 0s. In order for the receiver to read the signal, it needs to be able to determine if the line is at a 1 state or a 0 state. (Say Vcc or Gnd)

When you send a signal down a peice of copper wire you run into several issues.
Restistance, capactence, inductance and reflection.

In order for the signal to be clean, the driver must be able to bring the line to a 1 state or a 0 state within the time allowed. The faster this time requirement is, the larger the driving current must be.

Not sure what you mean by a 22g copper wire. 22g would be grams (not very useful) while 22awg would be its guage (more useful), or is this something else?

 

[Superdoopercooper]

Ok... check this....

if yoy have a cable that has a maximum (let's say 3dB breakpoint) flat response of 350MHz. Worst case ditial stream would be 01010101010 becuase of the alternating high and low. So, you can have 350Million 1's or 0's each second. Since a 1 or a 0 is a BIT, then you would have 350 Mega*BITS* per second. Traditionally, a byte is = 8 bits... so, you therefore have a transfer rate of 43.75Mega*BYTES* per second. If you are considering a byte some other length... well.. you can do the math.

Transfer rate over a wire is a function of impedance (R/L/C) and length... (if your receiving end circuitry had stuff in it to decode an encoded sync signal, then you should have no problem over a long lenght, as long as the signal is not ATTENUATED too much... the more length, the more the signal will be attenuated by the time it gets to where its going - that's why there are amps all over the place for the cable system).

 

[AnoTech]

Are you implying that dB directly relates to transmission speed in copper wire?

Maybe I mis-read, but if I didn't, I clearly don't understand how that works...



EDIT: After re-reading the above post, i have realized it is way too late for me to be thinking... I'll go to sleep now and oh,. hey,
forget you ever read this post.

PS: everything makes perfect sense.

 

[thorin]

"if yoy have a cable that has a maximum (let's say 3dB breakpoint) flat response of 350MHz. Worst case ditial stream would be 01010101010 becuase of the alternating high and low. So, you can have 350Million 1's or 0's each second. Since a 1 or a 0 is a BIT, then you would have 350 Mega*BITS* per second. Traditionally, a byte is = 8 bits... so, you therefore have a transfer rate of 43.75Mega*BYTES* per second. If you are considering a byte some other length... well.. you can do the math."

Wouldn't that assume that you could only tranfer data on the leading (rising) edge of the clock signal? What if it's a DDR solution (transferring on both the rising and falling edge). Or a QDR/QBR solution transferring on the rising edge, the peak, the falling edge, and the gully?

Thorin

 

[SCSIRAID]

Super didnt say anything about the clock. His discussion relates to the raw bandwidth of the 'channel'. Any encoding which may be required for clock recovery at the receiving end or error checking would reduce the actual data rate to below the channel rate (serial busses for example). Running the channel to the 3db point will result in sinusoids which may not be very useful depending on application. Ultra 320 SCSI actually 'precompensates' the signal by overdriving the amplitude of the leading edges to result in higher amplitude of higher frequency harmonics to compensate for lack of channel high frequency response.