I'm using 2 HMC253QS24 for a switched filter bank, and I have encountered a weird problem.
When DC supply is switched ON, the system works fine. However, when switched from one channel to another, the overall insertion loss of the system would need to have a 1-1.5 second period to reach it's final value. To be specific, at the moment when the SFB is switched from channel 1 to channel 2, the insertion loss read on the vector network analyzer is 4.54 dB, and then 1~1.5 second later, the insertion loss gradually reaches its stable value, which is 4.49~4.50 dB something. see the attached image.
The problem only exist when switching between channels, and all channels have similar situation.
The averaging of the network analyzer is turned off, and the IFBW is set to be 1.5MHz. the same situation appears with different IFBW and number of point setting.
The two HMC253 are controlled by a Hex inverter. When the Hex inverter is bypassed and all capacitors are removed from the pins of the HMC253, and all control pins are connected to a DC supply directly, same problem occurs.
Using the same PCB, and change the HMC253 to HMC183QS24 (with some rewiring done) and all other variables kept same as the HMC253, The HMC183 preforms well during channel switching, No "insertion loss gradually change" is observed.
T'm worried that when the switch filter bank reaches it's max switching speed, would the insertion loss difference be 0.04 dB or much more than the designed insertion loss?? That is to say, when switching at a speed of 150 ns between channels, would the actually insertion loss of the switched filter bank be 0.04 dB bigger or much bigger than the designed one?
So anyone has similar problems, and can you help?
Thanks in advance