Anyone know what it is?

[boy]

It should be a 90 degree hybridge. But I have never seen it before. This is the input of a balance amplifier.

 

[biff44]

It is obviously an in-phase power splitter. You can tell that from the equal distances from the feed point to the two take-off points. The rest is some sort of network that either
1) isolates reflected energy from one device from getting back into the other device. It kid of looks like that reflected energy would go λ/2 by one path, and 2λ/2 by the other path--cancelling at the other take-off point
2) could be some sort of matching network to keep the voltage low at the white chips, which I assume are resistors to ground. Sometimes you have voltage arcing problems at such devices, and maybe this structure minimizes the reflection standing wave voltage at the resistors. Note that those white chips are mounted on a metal standoff for heat dissipation, so there is probably a lot of RF power in this circuit.

 

[boy]

It is a 200 W pulse amplifier. I know only wilkinson and T-jutction power divider. In wilkinson, we did not connect any resistor to ground. that why i am dont know why they did that here. I doubt that this is a balance amplifier but i could not figure out what that circuit is. They use balance amplifier structure for pulse amplifier for protecting the circuit from rapidly changing impedance due to input pulse.

Added after 1 hours 46 minutes:

dear biff4
I have simulated the circuit. It is indeed a power divider with balance configuration for a balance amplifier. All reflected power will be discipated at those two resistors.

 

[biff44]

Hmmmm, very interesting. If you did that with a normal wilkensen splitter, all the reflected energy would show up across a single 100 ohm chip resistor, and it would probably arc over, and would also be much harder to heat sink since the resistor would be in series between two points on top of the substrate. I guess this rat-race splitter is a nice improvement for high power applications.