Active multiplier design

Hi,

I am designing an active doubler,using reflector networks at both sides, reflector network consist of a series line as well as an idler circuit ( quarter wave open stub), idea is to reflect the undesired harmonics back to device in phase,.
since the quarter wave stub is resonant at one particular frequency,i want to know what is usually the percentage bandwidth of the stub and how to calculate it.

My input frequency is 5.5GHz-6.0GHz, i am using HEMT device for multiplication.

How much bandwidth the stub can sustain ?How to increase the bandwidth?

Avijit

hmmmm. the method you are describing works well with a passive device, such as a varactor diode. In fact there is a paper by rafuse, and a book by Penfield, telling exactly what impedances you present to the diode at all the idler frequencies, to achieve maximum efficiency.

However...that is not quite how an ACTIVE multiplier works. You add structures at the input, output, and control the DC bias in such a way as that the large signal WAVEFORM has either rich even OR odd harmonics. You are not so much "bouncing back" idler energy, but instead more "waveshaping" the output current and voltage waveforms.

It is a good idea to use a series transmission line at the output of the active doubler, to adjust the phase of the fundamental frequency.
You need to place a BPF after the transmission line, tuned to 2*fo.
I think the 10% bandwidth which you are looking for, its easy to get.

Hi,

For active multiplier design, which PHEMT is better ? E-PHEMT or D-PHEMT

E- Enhancement mode.
D-Depletion mode.

Did you considered radial stubs?
I've used NE3210S01 with negative biasing to double 5GHz to 10GHz, and only shorted F0 at drain using radial stub, nothing about 2*F0 at gate.
For really good multiplier I suggest prototyping array:
5 boards, N-th board drain to stub distance: Loffset=180deg/N
Test N-th board by cutting stub to different lengths Lstub, each time do measurements.
Result is 2d-array of data: 2*F0 amplitude, F0 suppression for each Lstub vs Loffset