Re: [BP] SDD21 Concern

[Joel Goergen <joel@xxxxxxxxxxxxxxxxxxx>]

Title:

Hi Steve ... inline -joel

Stephen D. Anderson wrote:

 

    Joel, all:

    I think that people focussed on the very high frequency region of my proposed SDD21 magnitude.  This isn't what I intended.  I was focussing mainly on the region from 50 MHz to 5 GHz.  However, when I saw that the equation coefficients I picked created a curve that was below most of the measurements we had, I stopped and didn't go any further.

    John D'Ambrosia pointed out to us that this doesn't allow for an upper-layer ATCA QuadRoute channel, which happens to have a null in the region of 10 GHz.  (My guess is that it probably doesn't allow for any upper-layer channel that doesn't use backdrilling.)

    We (Xilinx) think that most signaling methods are probably not very dependent on what happens at 10 GHz.  So, in the interest of resolving this, Brian Seemann and I are re-evaluating Xilinx' proposed curve.

From what I recall, there was much discussion on the ending frequency earlier on.  15Ghz was agreed upon by many people that believe otherwise.  I want to be careful not to open that up again :)!  Think 50Mhz to 15Ghz :).

    One question I would ask:  If the curve is to be formulated primarily on the basis of whether or not backdrilling is done, then why have we paid so much attention to Dk,Df?  Whether the curve moves up or down by a dB is probably 90% dependent on backdrilling and 10% on material.

I don't know where this came from ... certainly not from me.  Back drilling helps to eliminate effects from resonance structures, reflections, and potential standing wave concerns ... from my perspective.  And I'm sure there are many other perspectives based on individual experience.

Regardless of back drill, and WITH reasonable design practise, the channel can very by up to 10dB in the band above 5Ghz, and up to 5dB below it from mfg process issues with the glass.  I would reverse your percentages.  1dB variance is easy to see from mfg.

The reason we paid so much attention to Dk and Df is to put the effects into perspective and make sure they are included:
- etch width
- XY shrinkage
- drill deflection
- material thickness tolerance
- resin structure variance
- thermal effects
- humidity
- copper geometrical variance
- copper thickness
Including all this gives us the original SDD21 numbers I submitted for use.

I think the model I submitted last gives us the changes at the lower freq with minimal effect on the upper frequencies.

Though, I do agree with Joe that we could create a two piece equation if the higher freq variance really hurts.  I perfer to keep one equation and one set of variables ... but that's just me.

Take care
-joel

    Regards,

    Steve A.
 

Joel Goergen wrote:

Hi Everyone.

I'm really concerned that no one has agreed or disagreed with any of my personal observations on the SDD21 mask.  I'm hoping everyone received what was sent.

Further, I was hoping for discussion and I don't see any at all.

????
-joel