Re: [SI-LIST] : HIgh-speed interconnect characterization

About this list Date view Thread view Subject view Author view

From: Gary Otonari (garyo@gigatest.com)
Date: Mon Dec 11 2000 - 10:41:34 PST


Hello Ravinder,

We've had some experience characterizing interconnects like the one you are
working on. Here are comments on your three questions:

> 1) How valid is my test setup for the type of measurements I am trying to
> make. Am I seeing an actual response of the interconnects, or will there
> be an impact of long 50 ohm cables, that I have used to connect my setup to
> signal source and oscilloscope.

The 36" cables will have a definite effect on the measurements you take.
Ideally you would verify the desired risetime at the input to your device.
Alternatively, you could use a network analyzer and calibrate the response up
to
the launch point. A second point of concern is the soldered transition from
the PCB trace to the 18" cables to your oscilloscope. This type of transition

will usually create a large discontinuity in your system, causing VSWR
problems. You should use a 50 ohm probe to transfer the signal from the PCB
to
the oscilloscope.

> 2) If my setup is correct then why do I see the dip at 1.3 GHz in my model.

I think you are seeing the effect of the impedance mismatches in your setup.

> 3) Can a 90 degree bend cause the kind of dip I am observing.

Yes, cables have to be bent very carefully. We usually keep the radius of
bends at least 10 times greater than the radius of the test cable itself.

-- Gary Otonari
    GigaTest Labs
    garyo@gigatest.com
==================================================

Ravinder Ajmani wrote:

> Hi SI Gurus,
>
> I am characterizing a differential interconnect comprised of a flexible
> cable attached to a PCB through a connector. I am using the following
> setup:
>
> Two 2" semi-rigid cable are soldered to one end of interconnects on the
> flexible cable. Other ends of the semi-rigid cables are connected through
> 36" long 50 ohm cables to + and - outputs of a pulse generator. The other
> end of the interconnects is connected to an oscilloscope through two 18"
> long cables soldered to the PCB. The pulse generator output rise time is
> less than 100 ps. The purpose of my testing is to ensure that the
> interconnects (flex cable, connector, and PCB traces) can be used to
> transfer data at 1 GHz.
>
> The oscilloscope waveform for a 1 GHz input showed almost no reflection, or
> attenuation or distortion of any kind. As I played with the pulse
> generator frequency, I found that the waveform was getting severely
> distorted at 700 MHz, although I was able to get a reasonably good pulse
> shape upto 1.3 GHz.
>
> Next, I tried to model this interconnect, and frequency-domain simulation
> showed a dip at 1.3 GHz (0.5V signal drops to 0.36V). As I played with the
> model, it became apparent that the dip was due to a 90 degrees bend in the
> interconnect on flexible cable. When I changed the model to eliminate the
> bend, I still observed the dip, but it was very small (about 0.05V).
>
> This experiment has left me with more questions than answers. I will
> really appreciate answers to the following questions:
>
> 1) How valid is my test setup for the type of measurements I am trying to
> make. Am I seeing an actual response of the interconnects, or will there
> be an impact of long 50 ohm cables, that I have used to connect my setup to
> signal source and oscilloscope.
>
> 2) If my setup is correct then why do I see the dip at 1.3 GHz in my model.
>
> 3) Can a 90 degree bend cause the kind of dip I am observing.
>
> Regards, Ravinder
> PCB Development and Design Department
> IBM Corporation
> Email: ajmani@us.ibm.com
> ***************************************************************************
> Always do right. This will gratify some people and astonish the rest.
> .... Mark Twain
>
> **** To unsubscribe from si-list or si-list-digest: send e-mail to
> majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
> si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
> si-list archives are accessible at http://www.qsl.net/wb6tpu
> ****



**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:30:24 PDT