Interactive length tuning for LVDS signal

[senseye]

Hi,

I am trying to use the command interactive length tuning for a differential pair. The lengths of the signals are 53.994mm and 55.12mm. The tolerance is set to 0.1mm. When I use the command with the shorter line, I get an empty bar with three numbers: 53.994mm,  55.12mm and 55.02mm. I move the cursor but the bar status does not change. When I press the bar command, I have:

From Rules checked.
Target length 55.12mm
Clip to target checked.
Mitered with lines style
Max Amplitude 0.5mm
Gap 0.5mm
Amplitude Increment 0.05mm
Gap increment 0.05mm

Thank in advance,
S

[T3sl4co1l]

Hit TAB and play with the measurements; or even better, hit ~ to see what keys are active (uh, miter size 1/2, gap size 3/4, amplitude ,/. and so on) and play around with them to see if anything happens.

There should be a length rule as well.  You'll see the rule on the TAB dialog, showing which trace(s)/bus the length matching is referenced to.  It sounds like this is working, because you got the right numbers in the tuning gauge.

Tim

[senseye]

Thank very much for your help Tim. The reason why the command wasn't working, was that the differential pair was routed inside a polygon. I test the command without the polygon and it worked. After tuning the lengths, I'll put the polygon back to its place.

[Ice-Tea]

Just checking: did you take that polygon pour into account when you calculated your diff impedance?

[senseye]

Thank you for the comment Ice-Tea. After reading it and some information on the Internet, I will leave a clearance between the LVDS signals and the ground polygon. I've read something about at least 3x the trace-to-trace spacing of the differential pair, still need to investigate more.

[T3sl4co1l]

Alternately, you can keep normal clearance, and reduce the trace width slightly.

How much?  Depends.  Differential CPW calculators are hard to find (unless Saturn does it, I forget).  You could estimate the change by comparing microstrip to CPW, using the diff pair width (outside trace edge to outside trace edge) as the trace width.  (The impedance resulting from this calculation will be an underestimate of the common mode impedance, by the way.  But for differential impedance, just use the ratio as adjustment.)

And if you're really desperate, you can create arbitrary shapes in ATLC and calculate the impedance that way.

The change is pretty small; bringing in ground usually only decreases the impedance by ~20%.  You could probably leave it alone and ignore it.  Digital comms have wide tolerances, and you can always adjust the termination resistor after fabrication.

Tim