LNK304D High-Side Buck - Direct Feedback

600V DC input is too high for LNK-TN direct buck application. the combined high current and voltage stress at start up and also during times when catch diode is in reverse recovery will be beyond absolute max rating of device.

I understand.  

I have a high voltage NPN pass transistor with zener diodes to limit the input voltage to ~500VDC.

My problem is that LNK304D quits working well with Drain voltage of >=375VDC.

Should this device work with 400VDC Drain voltage?

Yes, If there is a problem at 400V or below then look at drain current and see if there is a large intitial current spike due to recovery of catch diode. If so, then cycle is probably getting terminate on current spike. You may need to find a faster recovery diode if this is the problem

I am using the suggested US1A-F catch diode.

Should this be okay?

I am testing the device with a PowerStat Variable Autotransformer with a step-up transformer to get to the 280VAC input voltage, and ramp up the input voltage slowly from 0VAC to 280VAC, where the LNK304D stops operating, so I don't think this is initial current spike.

Any other hints? 

Please send waveforms of drain current at input VDC level just before failure.

Here are scope shots of the drain voltage just before failure and just after failure.

Waveforms are not useful -- you need to send drain switching current showing transistions -- perhaps 1uA per div for resonable resolution

Here are the drain current waveform pictures.

This is actually V1--V2 voltage waveform across a 1-Ohm resistor  in series with the drain. 

My AC Ammeter reads 14mA during this time. 

EPR48, page 12 shows source voltage and drain current for normal operation. The waveforms you have here do not look at all like drian current.

You are correct.

Do you have any suggestions on how to improve my design so it will work as intended? 

Makes no sense -- suggest you use a rwal current probe. Please look at EP48 for typical drain current waveforms