GRF5604 test boards finally arrived. I tested one of them using an HP E4436B RF signal generator. The datasheet mentions gain value of 37.5dB at 460MHz and 5V supply, so I used 0dBm signal to find out if we can get anywhere close to the reference value.
Right after measuring about 34dBm of CW power, I decided to take a look at the 10W 40dB attenuator used in the test setup. It turned out that its attenuation changes dramatically (5dB or so) upon applying mechanical stress (bending) to its SMA connectors. Since the attenuator is malfunctioning, I will have to perform the test again after getting a reliable, new part.
For now, let’s focus on the amplified signals. Here’s a set of spectrum plots and eye diagrams for both M17 (4FSK) and TETRA (pi/4-DQPSK):
The “step” effect at the sides of the TETRA signal will hopefully be mitigated by applying digital predistortion at GNU Radio level. Here’s a sample 2nd order polynomial DPD block for GNU Radio. There’s a good chance that the device will be able to apply predistortion in real-time, based on the measured output signal (feedback loop). A similar approach could be possibly used for IQ imbalance and DC offset correction.
Leave a Reply