Rogers 4003c substrate losses 50-70 Ghz

[Salving]

Hello everybody, could anyone give me an information about losses in rogers 4003c at frequency range 50-70 Ghz?

 

[johnjoe]

Uhh, honestly i think rogers 4003c is completely unsuitable for such frequency range. Losses of what? Microstrip ( ;-) ) or cpw? Better think about Rogers3000 etc.

 

[FvM]

Two ideas:
- ask Rogers
- loss curves are linear in dB vs frequency up to 20 GHz according to datasheets. For a first estimation, I would extrapolate it.

 

[Salving]

Thanks to all of you, i was interested of finding out loss tangent, but now i gotta measure one

 

[johnjoe]

Can you post your results?

 

[Bob60]

I`ve made some device with Rogers 4003c at 40GHz. Hard to say but it worked quite well gain OK, Fn~ 4-5dB. It was frontend with Mimix-Brodband naked chip.

 

[johnjoe]

I`ve made some device with Rogers 4003c at 40GHz. Hard to say but it worked quite well gain OK, Fn~ 4-5dB. It was frontend with Mimix-Brodband naked chip.

Can you tell a little bit more about your pcb design. What kind of connector did you used? Transmission lines, CPW? I did not expected results at such frequency with rogers 4003.

 

[Bob60]

8mill and main idea maximally fast covert 40GHz to a IF frequency, micro stripes was short, below may be 5mm. And waveguide adapter. I`ll find some photo, that is old story 10 years should be. And completely crazy story how I bonding Mimix chip with iron. I cant repeat any more that trick.

 

[johnjoe]

If you can post a pic, it would be great. And you used iron for bonding? Kind of wedge or ball bonding?

 

[Bob60]

No it was tricky gold to gold soldering. Bonding machine for manufacturing appear later, when device was already checked.

 

[ge]

Southwest Microwave did a nice write-up (advertising their connectors), 30mil Rogers 4350 50GHz
Search using: Optimizing Test Boards for 50GHz End Launch Connectors

 

[Bob60]

Here we are. Mimix Broadband on the Roger4003. 40GHz.

 

[johnjoe]

A little bit hard to see something, but nevertheless quite impressive. Thanks!

 

[Bob60]

If you will knowing in detail, you would impress much better. I`m still impressed as well ( after 10years ), how I made these design with nothing. Some bosses slept at these historical moment, but not for a long time. Then they wake up ( when It was dan ) and as usually no bonuses no champagne, nothing. Capitalismus brother.
If you seriously I can assist in detail. Strongly saying it is not right design, but it works. At first, chip directly connect with glue to the PCB without thermo compensated die ( titanium?) , but in these case was OK. PCB bondable gold with palladium.
Finally it was budget variant. Chip cost ~ 50$, PCB ~ 5$, amount less than 100$. But if we go to a microelectronics right configuration, instead PCB should be polikor, titan dies, eutectic and so on. The price will be on by 10.

 

[tony_lth]

Yes, more than 5 years ago, i did 40GHz point to point microwave telecomm, thanks to the simulated software, it looks my design much neat than yours.

 

[Bob60]

design much neat than yours.

Thank you. Is it possible to check how it looks.

 

[tony_lth]

Hi, Bob, sorry, you did a very good job without SMT, that is impossible for me in your situation.
For the probe, I had done a lot of simulation, so I did not do any error-and-try in the waveguide port.
And some de-coupling capacitors are differet from you.
In my hand, I don't have picture. Sorry again.

 

[Bob60]

Here is only one significant cap from ATC, others newer mind.