CPW SM attenuator chip 1

[logbook]

I am using a nice little 3dB surface mount attenuator, roughly 0805 size, which works up to 20GHz. Unfortunately it only has three terminals; in, out, and ground. What I really want is the same thing but with two ground terminals so it fits more symmetrically on coplanar waveguide.

Has anyone seen such a device? Manufacturer’s name?

 

[MKimagin]

not familiar with the 0805 size attenuator , however I kown atttenuator that have 4 terminals (in out and 2 seperate grounds) that is small. One of the PAT series from mini-circuits, unfortunately they work only up to 7GHz.

http://www.minicircuits.com/cgi-bin/inquery.cgi?querystring=af320&searchtype=case&x=24&y=16

 

[logbook]

Thanks for the link.

It is interesting that all three of the mini-circuits offerings have two ground connections. It is hard to imagine being able to design such a component to have only one ground in fact!

Pity they only work to just above DC

 

[Higgler]

Why not make your own attenuator using R Card (Resistive Card).
It's perfect for CPW, it's a few mils thick and just lay it across your contacts similar to a chip resistor used in an isolated power divider. Trim the length to get the 3 dB loss. RCard comes in impedance 10-400 ohm's per square. This would be similar to a rotary vain attenuator in waveguide too, just some lossy stuff in your field.

Now the bad part, the perfect 3 dB loss won't be seen over a wide bandwidth. It's like the loss of a cable, more loss at fhigh. So depending on your bandwidth, it may be usable.

kch

 

[logbook]

It's a broadband application, say 1GHz to 20GHz, and non flat response is unacceptable. However I have not heard of this stuff before so the idea is interesting anyway. Thanks.

 

[Higgler]

I believe that a normal attenuator has a T or Pi shaped arrangement of resistors. If you have chip resistors available with the proper impedances, can't you add them into your CPW lines and between the lines to get what you want. Probably won't work at 20 ghz unless the parts are very tiny and you can get them very close together.

Good luck.

kch

 

[logbook]

Well that is of course the other option I was looking at. It is difficult (= impossible) to quantify the VSWR of such a "homemade" attenuator, but if I make them into an "H" (5 resistors) or a "+" (4 resistors) it will certainly have the desired physical symmetry.

 

[Higgler]

You can buy dielectrics with Rcard underneath the copper cladding. We are doing that now for antenna development. I think the company is Omegafoil I believe, a coworker is designing it. With that you may be able to use etching to make your circuit. I saw that

http://www.E-fab.com

uses it
Previously, I worked on stripline Microwave Lenses using this material where we etched this Omegafoil over 4-18 ghz bandwidths to make a 2:1 power divider resistive load for isolation purposes. The layers are dielectric/Rcard/copper cladding. Removing the copper cladding leaves Rcard which is in contact with the copper foil.
If you etched your 4 or 5 resistors, that might work.


kch.

 

[divebuddydave]

You might want to look at Micrometrics.

I have designed both T and Pi attenuator chips that they have manufactured for me for use to 18 Ghz, but they are larger than 0805 in order to handle higher power levels.

These chips

http://www.micrometrics.com/pdfs/AP_Datasheet.pdf

are very small (.03" x .03") and have 4 leads (Pi configuration) and are rated to 40 GHz.

David

 

[logbook]

Nice one. Thanks.