AnsweredAssumed Answered

E8267D and I/Q modulation

Question asked by florentlecocq on Apr 4, 2013
Latest reply on Apr 22, 2013 by tabbott
I'm writting this thread to discuss aboutthe I/Q mixers on the vector generator E8267D.

I mainly use this instrument to generate 10GHz Gaussian pulses.
"Mod" is on, "I/Q" is on and "ALC" is off.
Using an arbitrary waveform generator I send a Gaussian pulse to the I/Q port of the E8267D (50 Ohm, Front Panel).
Typically the Gaussian pulse is 0V to 1V amplitude, rise time of about 300ns, with a repetition rate of 5 kHz.
At the output of E8267D I get  the product of that envelope and the main carrier at 10GHz.
To cross check I send this tone to a signal analyzer (N9030A).

Here is what I see :

- If all the "I/Q" and "Mux" settings are on factory default, I get a max power at the peak of the Gaussian of about +6dBm compare to the generator output power (measured used the "Burst Power" mode of the N9030A).

- If I change the amplitude of the Gaussian pulse amplitude by -6dB (factor of 2), I get a change in burst power of -6dB, unless I get close to the max amplitude, where the mixer saturate (from 0.5V to 1V, I only get a power change of 4dB). When the mixer seems to saturate, my pulse do not look Gaussian anymore in the frequency domain (FFT with the N9030A), and it acquires higher frequency component ripples.

- If I know start to change the "Mux" parameter called "Modulator Atten" (wich I assume is an attenuator between the front panel I/Q ports and the mixer ?) I can now change the maximum output power of the burst, and also the saturation level of the mixer (as measured by the degradation of my Gaussian envelope). But for a fixed "Modulator Atten" of 8.84dB (factory default), if I switch between "auto" and "man" mode, WITHOUT CHANGING the attenuation, my pulse shape is clearly affected ! why ? is there any "dynamical" change of the attenuation when in Auto mode ?

- In the sub menu "settings for Auto mode", what is the "optimize for X Msps bandwidth" means ? What does it change ? And as I'm far from being able to give a meaning full definition of RMS input level (I send a micro second pulse of amplitude 1V every 200 micro second), what "Input level" should I choose ?

Long story short, my pulse requirements are :
No distortion from the I/Q mixer (a Gaussian pulse in the time domain look Gaussian in the frequency domain)
An output power behavior as linear as possible (or otherwise with a minimal amount of calibration Vs Carrier Frequency, Carrier Power, I/Q gaussian pulse amplitude and shape)
And last I can't really afford any big loss of power during the mixing step.

All right, that was a lot of information, I hope I've been clear enough in my description !

Thanks a lot,