Re: [SI-LIST] : RF & Digital

S. Weir ([email protected])
Wed, 11 Aug 1999 20:54:32 -0700

I think that the application needs to be understood. The problem that I
see with the proposed stack-ups is what happens when a signal switches
planes. The nearest reflection plane for several of the layers is one or
the other of the PS voltages. This is going to interpose significant loop
area and inductance as the return current will have to traverse some
portion of one of the power planes and then finally return through the
decoupling network. Given the fundamental lower inductance limits on both
vias, and the decoupling capacitors, this is going to really limit
crosstalk attenuation at very high frequency. My vote goes along with the
suggestion of:

S sparse
S
G reflection for L2/L4
S
V
V
S
G reflection for L7/L9
S
S sparse

With respect to the power decoupling and buried capacitance, I think this
is better done by adding two GND layers if the design can possibly afford it :

S
S
G
S
V
G
G
V
S
G
S
S

I am also a disciple of using plane cuts only with the greatest
consideration. It is fundamental that where the digital logic interfaces
to the RF that none of the digital signals cross a plane cut.

I also agree with that consideration must be paid to which if either
voltage plane is powering the RF, and whether it also powers logic. Once
that is known, then a reasonable stack-up and placement rules can be
generated. Without it, I think we are all firing arrows in the dark.

Regards,

Steve.
At 04:35 PM 8/11/1999 -0500, you wrote:
>Depends on what you are looking for, but the most likely effect is the
>possibility of noise coupling directly between the two power planes. With
>the stackup as shown here, 5V couples tightly to L6 and more loosely to L2,
>giving enhanced plane-to-plane capacitance. If 2.5V needs to be the less
>noisy, then switch L9 and L5 layers.
>
>Ron
>
>
>
>
>Shayle Hirschman <[email protected]> on 08/11/99 07:19:24 AM
>
>Please respond to [email protected]
>
>To: [email protected]
>cc: (bcc: Ron F Parson)
>
>Subject: Re: [SI-LIST] : RF & Digital
>
>
>
>
>An excellent suggestion, Mitch.
>
>Any benefit to switching L6 with L9?
>
>Shayle
>
>
>
>
>At 07:59 AM 8/11/99 -0700, you wrote:
>>Hi Lum,
>>
>>As Shayle states this is a rather odd stackup, and I would certainly not
>recommend it unless I was under duress to comply. This will not get you the
>impedance control you're after unless you take some severe steps to control
>it
>>layer by layer (i.e. differential pairings). I'd suggest the following
>stackup design:
>>
>>L1 : signal
>>L2 : GND
>>L3: signal
>>L4: signal
>>L5: +5V
>>L6: GND
>>L7: signal
>>L8: signal
>>L9: +2.5V
>>L10 : signal
>>
>>Good luck,
>>
>>Mitch Morey
>>Sr PCB Designer
>>San Diego, CA, USA
>>
>>Shayle Hirschman wrote:
>>
>>> Why a stack-up of 4 signal layers in a row (L4 to L7)?
>>>
>>> This seems a bit unusual.
>>>
>>> Shayle
>>>
>>> At 08:09 AM 8/11/99 +0800, you wrote:
>>> >If I have a design that have digital circuit and a small portion of RF.
>>> >There will be some signals going from digital into the RF and out from
>>> >the RF to the digital portion. If the digital portion has to be
>>> >impedance control at 50ohms, how should I go about doing such design to
>>> >ensure that both RF and Digital requirements are taken care of, without
>>> >causing any interference between these two circuits? What about the
>>> >ground? This design will be a 10-layer with the following stack-up :
>>> >L1 : signal
>>> >L2 : GND
>>> >L3 : +5V
>>> >L4 : signal
>>> >L5 : signal
>>> >L6 : signal
>>> >L7 : signal
>>> >L8 : +2.5V
>>> >L9 : GND
>>> >L10 : signal
>>> >
>>> >I would appreciate any advise from you experts out there.
>>> >Thanks and regards.
>>> >
>>> >
>>> >**** To unsubscribe from si-list: send e-mail to
>>> [email protected]. In the BODY of message put: UNSUBSCRIBE
>>> si-list, for more help, put HELP. si-list archives are accessible at
>>> http://www.qsl.net/wb6tpu/si-list ****
>>> >
>>> >
>>>
>>> **** To unsubscribe from si-list: send e-mail to
>[email protected]. In the BODY of message put: UNSUBSCRIBE
>si-list, for more help, put HELP. si-list archives are accessible at
>http://www.qsl.net/wb6tpu/si-list ****
>>
>>
>>
>>
>>**** To unsubscribe from si-list: send e-mail to
>[email protected]. In the BODY of message put: UNSUBSCRIBE
>si-list, for more help, put HELP. si-list archives are accessible at
>http://www.qsl.net/wb6tpu/si-list ****
>>
>>
>
>**** To unsubscribe from si-list: send e-mail to
>[email protected]. In the BODY of message put: UNSUBSCRIBE
>si-list, for more help, put HELP. si-list archives are accessible at
>http://www.qsl.net/wb6tpu/si-list ****
>
>
>
>
>
>
>
>
>**** To unsubscribe from si-list: send e-mail to
>[email protected]. In the BODY of message put: UNSUBSCRIBE
>si-list, for more help, put HELP. si-list archives are accessible at
>http://www.qsl.net/wb6tpu/si-list ****
>

**** To unsubscribe from si-list: send e-mail to [email protected]. In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP. si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list ****