RF channel characteristics

[jmatyas]

I did some measurement work on Silpa and its RF signal path and here are the results.

AM/AM measurements

Measurements were performed using a signal generator and a spectrum analyzer. A continuous wave single tone signal at 200 MHz from the signal generator was fed to Silpa channel's input. Channel's output was connected to the spectrum analyzer's input. For each input power value, a number of measurements was taken and their average value was used for processing. The procedure was repeated 4 times - for each of the Qorvo amplifier bias settings. Presented below are results for the most linear bias setting.

This allowed to obtain the following RF channel parameters:

• fixed Gain of around 52 dB
• P1dB of around -19 dBm (@gkasprow is this OK? considering it's supposed to work together with Urukul, which nominal max output power is 10 dBm)
• Output power at P1dB of around 32 dBm

Frequency analysis

Frequency analysis was performed using spectrum analyzer with its built in tracking generator. Tracking generator's output was set to -20 dBm and connected to Silpa channel's input via 20 dB attenuator so the signal was in Silpa's linear region. Analysed frequency range was between 5 MHz and 2.005 GHz. Presented results are for the most linear bias setting. Cutoff frequencies were defined as points were the power gain drops by 3 dB.

RF channel frequency parameters:

• maximum gain of 53.3 dB at around 40 MHz
• lower cutoff frequency at 12 MHz
• upper cutoff frequency at 385 MHz (@gkasprow is this OK? considering it's supposed to work together with Urukul, which output frequency is > 400 MHz)

[gkasprow]

The gain is too high.
I'd opt for 40 dB.
The way you define cutoff frequency depends on reference. If you define +/- 3dB, it would be much higher.
The SilPA HL is intended to work with Phaser.
We will create an HE version of it that is suitable for Urukul.
We don't want to operate Phaser/Urukul with max power due to distortions.
I expected 2W of output power with low distortions, at least such were my measurements of the power stage.
Please measure with RF probe (1k resistor in series with coax cable) where the compression occurs. I'm not sure if it is the power stage. Silpa HL is intended for low harmonic operations, so we are more interested in second/third harmonic distortion than bandwidth or P1dB.
We probably need to linearise the previous stages and lower the R3 value, which will reduce gain and improve linearity.
I'd add feedback to ADL5536 or replace it with lower-gain chip

[gkasprow]

the gain is too high. And we get some oscillations. Just add the attenation pads that will add 10dB of attenuation