Designer’s Guide Community

gorkhim

New Member

Offline

Posts: 6
San Jose

High Freq. Class E PA
Sep 22^nd, 2008, 4:52pm

I was trying to simulate Class E PA with over 50GHz frequency using nanowire transistors in spectre (cadence). But, the input impedance (real) is negative. Is there any resource or solution to solve this problem? I was not able to find enough resources to solve the negative input impedance due to high frequency. I tried to use input matching network but the fmax went down to less than operating frequency.
Thank you

Back to top

IP Logged

nano_RF

Community Member

Offline

Posts: 50
madison

Re: High Freq. Class E PA
Reply #1 - Sep 27^th, 2008, 9:15am

Is this a single ended or differential design? In differential you can have some positive feedback given that you are working in nano regime.

Would you be willing to share your schematic?

The only way i can think of a negative real impedance in SE config is that your device is degenerated by a capacitor. Does this real part drastically changes with frequency?

Back to top

IP Logged

gorkhim

New Member

Offline

Posts: 6
San Jose

Re: High Freq. Class E PA
Reply #2 - Sep 28^th, 2008, 3:19pm

Thank you so much for your reply. Yes, it is single ended design. I have also attached my schematic ( used for impedance simulation). I am getting -ve real impedance around the operating frequency. For example, for 50GHz of design, I am getting -ve impedance for arround 45 to 70GHz. For other frequency it is +ve.
Do you think, I should not use single ended design? Please let me know, really appreciated.
Thank you....

Back to top

single_ended_schematic.jpg

IP Logged

nano_RF

Community Member

Offline

Posts: 50
madison

Re: High Freq. Class E PA
Reply #3 - Oct 1^st, 2008, 6:37pm

Hi,

Your schematic seems pretty straightforward, do not see any problem with the set-up to cause negative resistance. No, i did not mean to suggest that a differential design would take care of this issue. However differential design do offer lot of advantages.

Now coming back to the negative real impedance. Can you simulate the impedance of your final device without applying the class-E set-up. In other words what do you really get out of your device model? By the way what is input signal level in your PA setup?

Thanks

Back to top

IP Logged