Hello!
I have MAX2380 in my ISM-receiver project. It has 4-layer PCB routed as in the EVKIT reference design. It works just fine except of one significant issue.
MAX2380 is configured to Rx mode ( nSHDN = 1, RXTX pin = 0), all registers loaded with recommended default values, all controls work well (frequency tuning, filter bandwidth change, LNA, VGA, etc.).
On default settings the common-mode voltage on all Rx IQ Outputs stays about 0.9 V the same on all outputs (pins 35 to 38). It is correctly adjusted by programming bits D11:D10 in Register 15.
Here is the issue. On VGA settings (bits D4:D0 in Register 11) greater then "0" the common-mode voltage of Rx IQ Outputs change. Voltage on RXBBI- (pin 37) and RXBBQ+ (pin 36) rises up. Voltage on RXBBI+ (pin 38) and RXBBQ- (pin 35) falls down. The voltage is different on "+" and "-" branches of diff-pair output. This strange behavior depicted bellow.
On VGA values about 9-11 voltage on RXBBI- and RXBBQ+ grows to about 1.9 V while RXBBI+ and RXBBQ- falls down to 0.5 V.
On greater VGA values voltages grow to about 2.5 V and fall down to 0.4 V tending to VCC and 0V respectively. Actual unaffected VGA control range appears to be 0..8 instead of 0..31 as mentioned in the datasheet.
This behavior causes a IQ gain imbalance and huge waveform saturation on VGA values greater then 10-12. Even internal noise is IQ-disbalanced and saturated with no input connected. Actual spurred
This VGA behavior is observed on different PCB prototypes specially built for this. It does not depend on PCB routing. It appears the same with SPI as well as parallel gain control methods. This does not depend on the Rx outputs termination (open, differential load, pull-up or pull down). It does not depend on other settings and registers. With minor differences it appears on 10+ chips ordered in official distributors.
I found related issue with similar chip here:
MAX2831 IQ output DC offset drifting
but it is not so helpful.
Looking forward for some good advice.
