About Matched Lengths Between Pairs

[mulfycrowh]

Hi everyone,

I defined 3 rules.
Each rule has a different priority.
As shown on the attached screenshot, each rule applies to several differential pairs.
I chose not to use differential pair class because I already experienced some trouble when I have to reimport the schematic.

The trouble here is that it randomly runs: sometimes I succeed with the tuning, sometimes not.

Is something wrong in the query syntax ?

Thanks.

[mulfycrowh]

I tried to tune eight differential pairs.
I wrote the following query syntax: InDifferentialPair(Pair0) OR InDifferential(Pair1) OR ... InDifferential(Pair7).

I tuned 2 pairs (In and Between), I've got the following signal lengths:

Pair0_P (94.856)
Pair0_N (94.685)

Pair1_P (95.247)
Pair1_N (95.247)

I set a tolerance of 0.254 mm for both rules.
Everything is green for all pairs when I drag the boxes.

But as you can see, it's wrong between Pair0 and Pair1 because of the tolerance.

Any idea ?

[mulfycrowh]

Does the problem come from the Query InDifferentialPair instead using a differential pair class ?

[mulfycrowh]

I made the same test with differential pair class: I have the same result.

[mulfycrowh]

As you can see on the attached screenshot, the length tuning is not done at 100%.
If I drag one side of the box, I can't do better.
I tried full stop and comma without success.

Could you help please ?

[mulfycrowh]

Another example to show that something is wrong.
I tuned PEG_RX0 ... PEG_RX3.
The tolerance is not OK between PEG_RX0 and the other ones.

[qdrives]

As nobody else is commenting...
Could it be that your query "InDifferentialPair" is causing your issue?
What is the effect when you use a differntial pair class?
I must admit, I rarely ever use differential pairs so this is just a suggestion.

[mulfycrowh]

qdrivesThank you very much for reply.
I made the test with differential pair class and it's the same issue.

I have one question: what is the difference between signal length and routed length in PCB Editor (Nets) ?

[mulfycrowh]

Another issue !
I defined another rule to tune length between pairs.
As I select Interactive Diff Pair Length Tuning and drag on one track of one differential pair involved in the group, the dedicated box doesn't show up, only a small white point shows up.

My motherboard is really huge.
Maybe 250 differential pairs to tune !

Very difficult to work with AD not doing what I want.

[mulfycrowh]

When I face an issue, I always check what I've done before getting the issue.
This morning, before going on with the tuning, I was checking the schematic again and noticed (refer to screenshot) that several differential pairs have 2 different names.
That's because I had capacitors on those differential pairs on a previous version.
So I removed the differential pair directives and the names as well near the chip.
I validated the project: no error !

From this time, it is impossible to tune a differential pair (between pairs) and apparently whatever it is.
If I rewrite the directives and the names, it is OK again with the tuning.

I don't understand but that's what I state.

[mulfycrowh]

Altium (21) is completely out of mind !

I tried to tune (BetweenPairs) another group of differential pairs.
Each pair has a length of about 158 mm.
As I drag on a differential pair to tune its length, the bargraphe on its right side shows 900 mm !

For sure it's impossible to have it green.

Do you already meet such problem ?

[robertferanec]

Maybe I would keep diff pair rule for tolerance within pairs, but I would create a standard signal length matching rule for all the signals what needs to have the same / similar length.

It's not really important they are differential pair signals, if they need to have the same/similar length you can handle them as single ended signals. That is what I would try.

[mulfycrowh]

robertferanecI think you have read what I wrote.
It is really awful.
Same with AD22.
When it seems to run (bargraphes green), in fact I get matched length between pairs violations, for example 1 or 2 mm (instead 0.254 mm).

AD doesn't do what I want.
Is it a huge bug or is it on my side ?
Is it bound to the size of the project ?

Matched length InPairs seems to be OK but that's all.

It puts me in a very difficult situation.

[mulfycrowh]

Everything went wrong since what I wrote yesterday at 03:40AM

[robertferanec]

But are you still using this? "InDifferentialPair(Pair0) OR InDifferential(Pair1) OR ... InDifferential(Pair7)" I would delete that rule and replaced it by one using just simple signals.