from datasheet maximum input level is +7dBm or +16dBm with 0.5V bias.
But +7dBm means a negative peak of -0.7V, while +16dBm + 0.5V DC means a peak of -1.5V.
I would expect a +12dBm maximum level to reach the same -0.7V.
Yes, at 350MHz, 15dBm input and a 0.5V DC bias the parasitic diode will not get chance to turn-on fully so no significant attenuation of negative portion of signal.
When the part is handling an RF signal then the ADG9xx can handle a 7dBm (0.7Vp) signal and with a 0.5V DC bias it can pass a 16dBm (2Vp) signal.
This is determined by the lower limit in datasheet of -0.5V DC; so to simplify things you can relate the AC signals to the Vrms level. 7dBm = 0.5Vrms and we recommend this is max signal you should use with 0V bias. With 16dBm signal we recommend a 0.5V DC bias to offset your input and this ensures you are not clipping your signal.
The power handling of the ADG9xx decreases with frequency because parasitic diodes will begin to turn on if input signal goes below ~-0.6V. The attached app note (Power Handling section) explains in more detail why this is so.
i have already read the AN952, the problem are the levels.
With 16dBm the peak voltage is 2V so also with a 0.5V bias the negative peak is -1.5V
The internal parasitic diodes will turn on if your signal goes ~-0.6V below GND. Hence the -0.5V DC limit.
However, these parts are designed for RF switching.
So the 16dBm, 2Vp is an AC signal and the -0.5V DC limit does not directly apply.
What is the frequency and power levels of the signals you are switching?
With 16dBm input and 0.5V DC bias, you should not see any distortion.
You can increase your DC bias but there is trade-off with increased On resistance and hence higher Loss.
so you mean that the parasitic diode is slow and if the RF frequency is enough it simply don't switch on?
My working frequency is 350MHz with 15dBm.
Retrieving data ...