RE: [SI-LIST] : Interesting theoretical questions

About this list Date view Thread view Subject view Author view

From: Balaji Lakshminarayanan ([email protected])
Date: Tue Nov 02 1999 - 06:21:37 PST


Hi,

I would like to give a different answer to 1 and 3.

1. I partially agree that the frequeny is limited by substrate thickness.
But there are published papers in MTT and other journals that talks about
generating frequency upto several hundred tera hertz using GaAs or InP
devices. The frequency is limited predominantly by the size of the device
(gate) than the substrate thickness.

3. An asynchronous machine has been developed at MIT which is called the
Data Flow Machine (i think proposed by Arvind). The instructions such as
addition or subtraction is triggered only when the signal arrives and it
is not operational otherwise. The effenciency of the machine is very low
and people are researching on improving the effenciency and commercializing
it.

Thanks

 --Bala

On Mon, 1 Nov 1999, Jian Zheng wrote:

> Following are my answers to the questions (check keyword ANSWER in the
> following attached context):
>
> ----------------------------------------------------------------------------
> ------------
> Jian-X. Zheng, Ph.D
> Zeland Software, Inc., 39676 Mission Blvd., Fremont, CA 94539, U.S.A.
> Tel: 510-797-8109, Fax: 510-797-8241, Web: http://www.zeland.com
> ----------------------------------------------------------------------------
> ------------
> Special Announcement: (1) We will be offering a training class on using the
> IE3D/FIDELITY on December 2 and 3 of 1999. Please check our web site for
> more information. (2) IE3D 6.0 is released. The IE3D 6.0 will be about
> 50-300% faster than the IE3D 5.X. If you have the ZELAND.EXE 5.2, you will
> be able to use the IE3D 6.0. (3) IE3D 7.0 beta should be released before
> December 1999. It is a big step from IE3D 6.0 according to its optimization
> capability. It has the implementation of symbolic optimization goals. It
> also features integrated IE3D+CURVIEW+PATTERNVIEW for automated pattern
> calculation and pattern optimization. Interested IE3D users can request the
> IE3D 7.0B for a try after it is released.
> > -----Original Message-----
> > From: [email protected]
> > [mailto:[email protected]]On Behalf Of subas
> > Sent: Monday, November 01, 1999 4:24 PM
> > To: '[email protected]'
> > Subject: [SI-LIST] : Interesting theoretical questions
> >
> >
> > Hi all,
> >
> > I have some theoretical questions regarding high frequency signals:
> >
> > 1) What is the maximum frequency you can generate electronically by our
> > current technology (GaAs???)? Can we have an oscillator that can oscillate
> > in the frequency of the visible light, for example?
>
> ANSWER: For planar circuits (PCB or millimeter wave IC), from what I know,
> people can produce circuits as high as 100-400 GHz. It is not easy to go to
> much higher. The reason for it is that the frequency is limited by the
> substrate thickness for the GaAs circuits. Some people may be using
> substrate as thin as 25 microns. For 25 micron thick substrate, the maximum
> frequency you can go is about 300 GHz. When the substrate gets thin, the
> GaAs circuits become fragile and it is very hard to process. The 2nd problem
> involved is the high metallic loss.
>
> To go to higher frequency, people have to use optical waveguide such as
> optical fibers.
>
> >
> > 2) The way light gets reflected from a metal (silver in a mirror)
> > is that it
> > first induces current (as a receiving antenna) and again radiates (as a
> > transmitting antenna) the EM wave instantly so that we get the impression
> > that the light is being reflected by the metal surface. Then how come
> > different metals can have different colors (gold - yellow) when we expect
> > them all to reflect light of all colors?
>
> ANSWER: At low frequency (compared to the frequency of light), plane waves
> hits the metal and will get total reflection from the metal. At high
> frequency, it is not true. How big a reflection is depends upon the complex
> permittivity (EPS):
>
> EPS = EPSr - j SIGMA / ( OMEGA * EPS0 )
>
> EPSr is the relative permittivity of the material. SIGMA is the conductivity
> of the material. OMEGA is the angular frequency. EPS0 is the permittivity in
> free space: (8.86e-12).
>
> Typical copper's EPSr is about 1 and SIGMA = 5.8e7 s/m. At 1 MHz, we have,
>
> EPS = 1 - j 5.8e7 / ( 2.0 * 3.14159 * 1.0e6 * 8.86e-12 )=1-j1.042e+12
>
> The refraction coefficient N is:
>
> N = SQRT(EPS) = 7.2e+5 - j 7.2e+5
>
> When light hit the copper, the reflection coefficient is:
>
> (1-N)/(1+N)and it is almost -1.
>
> At light frequency, the wavelength is about 7.0e-6 m. The frequency should
> be 4.0e+5 GHz (or 4.0e+14 Hz). Assuming the EPSr and SIGMA do not change, we
> still have,
>
> EPS = 1 - j 5.8e+7 / ( 2.0 * 3.14159 * 4.0e+14 * 8.86e-12) = 1-j2600.
>
> N = SQRT(EPS) = 36 - j 36.
>
> The reflection coefficient is still high. However, it is quite different
> from -1. With increased frequency, normally the SIGMA should decrease. The
> reflection from metal should be significantly lower than the total
> reflection, and it will differ with different frequencies. I think that is
> the reason why we will have differnt colors, at least from the
> electromagnetic point of view.
>
> >
> > 3) Has there been any computer made without any clocks (asynchronous)? How
> > fast is this computer?
>
> ANSWER: I have no idea about it. However, it seems to me it is impossible.
> For a computer without a clock, it has to be analog. When I was in school, I
> learnt that there were analog computers. However, you can not guarantee the
> accuracy because analog signals are affected by noise. In order to guarantee
> the accuracy, we have to digitize the signals. When we digitize it, we
> should need the clocks for synchronization.
>
> >
> > 4) Why is optical computing getting no where? Is this because lasers are
> > expensive? Or because of the lack of fast optical switch ( any non-linear
> > optical device)?
>
> ANSWER: I think it is more of the minimization and integration of optical
> switch.
>
> >
> > Thank you very much.
> >
> > Subas
> >
> > **** 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 ****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Tue Feb 29 2000 - 11:39:31 PST