Transmission line effects

Do you also use the correct source resistors? What sort of Rise and Fall times are you trying to achieve? Is the PCB trace designed appropriately for your application?

A matched transmission line does not have ringing, thus any impedance discontinuations should be fixed.

The transmission distance and/or bandwidth is limited in a matched system by the signal attenuation, which reduces the signal amplitude at the receiver end. Since the attenuation usually increase with frequency, a transmitted square wave looks in the extreme cases more or less like a sine wave, when the harmonics are attenuated more than the fundamental frequency.

Using intermediate repeaters and frequency equalizers is the standard practice in cable TV networks. In digital systems, pulse regeneration can also be used.

Is your transmitter on one edge of the transmission line, then there is a cable and the receiver is at the end of the cable ?

This should work OK as long as the PCB track characteristic impedance is the same as the cable characteristic impedance and the termination resistance is at the receiver end of the cable. A typical twisted wire cable will have a characteristic impedance around 100 ohms, thus the PCB tracks should have a characteristic impedance about 100 ohms.

The characteristic impedance can be controlled in a microstrip or stripline construction, in which one or two ground planes in addition to the actual conductor is used in a multilayer PCB construction. The signal track width depends on the isolation layer thickness and the isolation material used.

In a transmission line, the signal is actually propagating in the insulation material, thus the losses depend on the board material used. FR-4 will have significant losses even on lower frequencies, G10 is less lossy and if PCB tracks carrying signals above a few GHz, teflon boards may have to be used. Paul

If I understand your post correctly, perhaps you are terminating the transmission line in the middle? i.e.: at the boundary between the pcb and cable.

I assume that the cable, pcb transmission line and source impedance are all equal? Provided that they are and the connection between cable and pcb is done in a manner that preserves that impedance, you should not have any problems entire line is terminated at the end.

'Transmission line' is an ideal concept, not a problem. Within that idealization, all fluctuations in impedance, whether due to amplifier output impedance, or input impedance, or to endpoints of the line (because you can't afford an infinite size cable), result in a second kind of ideal effect, 'reflection'.

Proper termination of a transmission line removes one reflection. Mismatch in the termination introduces controllable reflections. Either or both of these may be desirable. 'intermediate buffer' doesn't mean anything obvious to me, nor will it generally solve a bandwidth problem.

If there is any "ringing" at all, then you've got an impedance mismatch that needs to be diagnosed and repaired before you do anything else.

If you put amplifiers in a line that's already mismatched, you'll only amplify the effects of the mismatch, exacerbating your problem.

Good Luck! Rich

Certainly, but remember that buffers add delay. In certain cases this delay is less than that of the transmission line, so yes, it is done.

I generally try to do that, mainly because it isolates the impedance of the cable and the PCB trace. Each can then be whatever makes sense, terminated by what is appropriate. OTOH, forcing PCB trace impedance isn't impossible either.

That is done on chips where the transmission line is pretty much RC rather than LC. In an RC transmission line the delay does up with the square of the length, so it makes sense to insert buffers on long wires.

Huh? Infinite is infinite. Remember also that BW delay.

Terminate properly and you'll be unlikely to need any extreme measures.

The obvious solution is to get rid of the 18 inches on the lossy pcb and add it to the cable????? If you have more than one load, you have inadequately described your situation.

You have given no clues to the signal frequencies, impedances, noise margin, etc. involved. While the theory is the same, the easiest fix may depend on the signals involved.

A TDR can be helpful figuring out what's needed.

Use proper impedance control and get rid of the ringing before you move on to more drastic measures. Active solutions are often inferior to passive solutions when it comes to cost and reliability.