Industry News

Ask Harlan: December 9, 2003


You may know Harlan Howe from his twelve years as publisher and editor of Microwave Journal ®, or from his 34 years as a Microwave design engineer and engineering manager, or from his service as an IEEE fellow and past president of MTT-S.

Now, although semi-retired, Harlan is available to answer your questions about RF and Microwave engineering. If he doesn't have the answer, he will find an industry expert who does.

Click here to submit your question now.

FROM: John Mohr, Raytheon

I am looking for an explanation of the properties of a Butler Matrix.


Dear John,

A Butler Matrix is a beamforming network made of a cascade of 3 dB, 90 degree hybrids. A signal input at one port produces equal amplitudes at all the output ports with a linear phase progression from port to port. Fixed or variable phase shifters are sometimes employed as well, for beamshaping, steering, sidelobe suppression and adaptive nulling. There are many, many configurations that fall into the general category of Butler Matrix. I have not been able to find a textbook on the subject, however, there are a mutitude of papers in the IEEE Transactions on Microwave Theory and Techniques as well as Antennas and Propagation. A search of their archives will probably turn up something on the specific configuration, which concerns you.

FROM: Arun Sharma, Otto-von Guericke University

Dear Dr. Harlan,

What could be the reasons for getting very high noise/oscillating power beyond the band of operation in a Coupled cavity TWT, which does not change with the input drive level, and almost nill power in the requisite band?

Sincerely yours,

Arun Sharma


Dear Arun,

The most likely cause is a high out-of-band mismatch, which is refelcting noise output back into the tube and causing oscillations. As an experiment, try putting a broadband matched pad on the output and see if that stops it.

FROM: Arion Lleshi, KTDN

Hi, Harlan

We have Western Multiplex-Tsunami 45 Mbps+2E1, which operates in 5,8 GHZ.

My question is how can we change the operatinig frequency from 5.8GHz ( unli) to 5,925 - 6250 Ghz.

What do you propose to us, for change of frequency but not to change the radios.

Arion Lleshi


Dear Arion,

I am not familiar with the specific radio, so I can't tell you whether the frequency can be changed. However, even if the source can be changed, most radios have a relatively narrow output filter, which would have to be replaced. In addition, the radio is probably licensed only at the original frequency.

FROM: Chang-Luen Wu, Transcom, Inc.

Dear Harlan:

It's nice to know someone like you -- an experienced expert who provides Q&A in this field. I have several questions in microwave applications as follows.

(1) What are the definitions and applications for so-called Class D, Class E, Class F, and Class G circuit designs?

(2) In some MMICs, I found some designs exhibit RF current pull-up, while some designs exhibit RF current pull-down even at the same DC bias conditions. Could you figure out the reasons for the RF current behaviors in HBT and PHEMT applications? Does that depend on the solutions on wafer level or design level?

Thank you very much.

Best regards,



Dear Wu,

Class D, E, F and G amplifiers are all switching amplifiers that are used in portable equipment where high efficiency and low battery drain are needed. However, there are subtle differences between them. Class D has a complementary output stage. Class E is single-ended. Class F uses third-order loading networks to improve linearity and Class G is complementary with two or more power supply voltages on each side, which improves efficiency at the cost of a more complex power supply.

Pull-up and pull-down currents are dependent on the PU and PD resistances. The function is used for shutdown and re-set applications in high eficiency TX/RX circuits. The current behavior is both device and circuit dependent.

FROM: Aviv Shachar, Motorola

Harlan hello

My name is Aviv. Ihave a question about RC or RL filter design:

How you decide for the values of the components (because i use in a regular equation Xc=1/(w*c) or XL=w*L) and I am not sure that I can use R (in the filter) if equal to Xc or XL and to find the cut off frequency of the filter.

In addition, about ferrite bead that in series to resistance which an equation can help me to find the cut off frequency of the ferrite bead (Note: ferrite bead behave like inductor and resistance in series, this is the equivalent circuit?)




Dear Aviv,

The design equations for simple RC and RL filter are in: Reference Data for Radio Engineers (Also known as the ITT Handbook). My copy dates back to 1956, however, I'm sure it is still available. The publisher is Stratford Press in New York.

The equivalent circuit for the ferrite bead is a series RL connection. Its characteristics are dependent of the permeability of the ferrite, its dimensions and the frequecy of operation. You need to get the data from the manufacturer of the bead.


Hi Mr.Harlan,

I would like to design a narrowband fiter of the following specifications :

passband freq.range=1920-1980 MHz
insertion loss in passband= less than 1dB

Since the percentage bandwidth is only 3%,and the required rejection is at 10%,the filter need to have a very high Q-factor. Could you pls suggest any topology or methods to design this filter.

Thanks & regards

Research Engineer,


Dear Ginka,

You have a problem! To meet the skirt rejection, a standard .01 dB ripple filter will require 5 sections. While the available unloaded Q of microstrip is a function of the materials and dimensions, I have never seen Qu at 2 GHz greater than 200 in a microstrip construction and it is usually more like 100. As a result, the bandpass loss will be between 5 and 11 dB, assuming that it works at all. You need to build the filter in some form of symmetrical, air-dielectric configuration. I suggest that you refer to: Microwave Filters, Impedance-matching Networks and Coupling Structures , Matthaei, Young and Jones, Artech House, 1980, ISBN# 0-89006-099-1.

FROM: Sarwan Kumar Sharma

Which one is better to use: wave guide or coaxail cable transmission line on normal condition? Please expalin with necessary calculation.


Dear Sarwan,

The choice of waveguide or coaxial transmission lines is dependent on the frequecy and bandwidth, power levels, permissable loss and additional reqirements such a phase stability and the environment. There is no universal answer, although waveguide will have a lower loss at a higher expense. Equations for both types can be found in: Microwave Engineer's Handbook , Vol. 1, Saad, Artech House, 1971.

FROM: Zhang Kai, UESTC

Mr Harlan:

Would you like to tell me of some software about how to compute the impedance of the suspended microstrip, especially about the broadside-coupled suspended microstrip


Dear Zhang,

There is commercial software available, however, we do not recommend specific products or companies in this column. I suggest that you contact some of the suppliers listed in our On-line Buyer's Guide .

FROM: Sachendra Sinha

Dear Harlan,

I am interested in the design of broadband microstrip antennas. Since these require a low dielectric constant thick substrates, I would like to know the names of the companies which can supply them. I have gone through the buyer's guide but could find only ARLON, which supplies a thin Foam substrate (Dk=1.15).

Could you kindly let me know what other companies manufacture these substrates?


S.N. Sinha


Dear Sachendra,

Since you have already checked out the other substrate suppliers, I suggest that you look at material manufacturers, many of whom make foams. They might supply something to you or you may have to machine and clad it yourself.

FROM: Matteo Biggi, Officine Pasquali SRL

Dear Mr Harlan,

About the Microwave Journal paper that I would like to receive the title is:

"Directional Coupler Design Nomograms ," T.N. Andersen, Vol II, May 1959.

If You can send me it , You may send at the address appearing in my subscription user profile.

Thank You again for your help,

Best Regards



Dear Matteo,

A copy of the paper has been sent. In the future, however, you and any other reader, who requests a paper should provide us with a mailing address. Since our circulation list is maintained by an outside company, it puts a burden on our staff to have to go to them to look-up a subscriber's address.

Post a comment to this article