Dear all,
I am working with transceiver AD9375, and I want to ask for help with proper GPIO switching in observer path. We are using transceiver for LTE Base station, so I will show some problems with signal quality during OBS switching.
My scenario: I want to split LTE frame (10ms) to 5ms for TRX InternalCals and 5ms for ORX to do some DSP/SDR signal measurements for my company’s own purpose. I do it via 100MHz few own sequencers which set GPIO values.
Below images: state analyzer from device which controls AD transceiver, and Keysight MXA Spectrum Analyzer.
Example: switching from INTERNAL_CALS (0x3) to ORX1 (0x1),
ORX_TRIGGER_RISE_TO_RISE – 5ms
ORX_TRIGGER_HOLD – 10us
tMODE_SETUP – 100us
tMODE_HOLD – 5ms
Problem: very short signal quality disruption during triggering Observer switch (I/Q mistake in one symbol). But not for every switch. Please see Table below.
Please look at constellation, and Error Vector Time.
I tried turn on/off trackingCalibrations and I see nothing better or worst in signal quality.
Switch from/switch to -> |
ORX_OFF |
ORX1 |
ORX2 |
CALS |
Sniffer channel |
ORX1 with sniffer |
ORX2 with sniffer |
ORX_OFF |
good |
bad |
bad |
Pretty good |
bad |
bad |
bad |
ORX1 |
bad |
good |
good |
bad |
good |
good |
good |
ORX2 |
bad |
good |
good |
bad |
good |
good |
good |
CALS |
Pretty good |
bad |
bad |
good |
bad |
bad |
bad |
Problem exists in every DC frequency, but for 3.5GHz has high EVM disruption (50% EVM) and for 500MHz, disruption is very small (<10%EVM). 500MHz below.
I wish you can help me.
Best Regards,
Jedrzej