10 W power transirtor is burning out

i have a problem with the SLD-2083CZ , when i apply the VGS , to check the DC char , the transistor burn out ,

i have followed the datasheet , but my transistor is not loaded

any suggestion ,

i am using it to design 10 W PA

thanks

Khouly

In the datasheet Sirenza specifies that this transistor passes ESD at 500V, but in this Reliability Qualification Report is specified that fails below 250V.

http://www.sirenza.com/documents/pro...ion_Report.pdf

yeah , i have seen this

but i have take the precations .

khouly

Do you have stable application? Have you terminated the input and output ports correctly? It is possible that after biasing the transistor, it starts to oscillate and burns right away. Also Drain bias has to be applied first and the gate bias after that (in that particular LDMOS transistor) In case you would have GaAs FET, you should apply first the negative gate voltage and after that the drain bias.

i have matched the input and the output , but didn't connect the load and source , is this the problem ?

khouly

Hi Khouly,

Most likely the unconnected load and source are the reason why the transistor burns

Karote

i will try and feedback u , thanks all

khouly

Added after 4 minutes:

is it also useful to use an isolator after the PA , or , just connect the load ?

khouly

Added after 2 hours 20 minutes:

the DC point is now , OK and the transistor didn't burn

thanks all

now i have applied the RF but the output is not as i expected ,

i think i need to modify the matching circuits

any suggestions ?

khouly

Yes, i think you need to redesign your matching networks.
Insulator is good idea because it protects your PA (usually from the Standing wave) if something happend with conection with antenna.

If the PA dies without source and load connected means that is oscillating, which is not very unusual in these conditions.

You have to tune first the input match for best input return loss (S11) using low input power.

Check for out-of-band high gain that could degenerate in oscillations at high output power, and/or poor load.

After that increase the input power and tune the output match for maximum power.

Isolator has not an infinite BW. If your PA is oscillating at low frequencies the isolator is useless.

yet put current sensing to supply side to stop paying for burned parts.

mm , Vfone , i have used the load pull data to design the input and output impedance , and followed the recommendations , but in the datasheet they mentioned not to apply the bias till the PA is terminated

khouly

Yes, that's right..
The PA's can "potentially" oscillate while they are not connected or terminated with a proper load even the matching circuits are connected. Because matching circuit input impedance which is seen by PA will be different than when the its output is terminated with load.If the reflection coefficient that is seen by PA transistor is high enough, the reflected voltage will increase till the power transistor is dead.
For instance if the reflection coefficient is around 1, the VCE voltage acrossing C-E junction will be doubled, if PA works in Class A, this may vary 0-4Vcc ( theoritically,depending on reflection phase etc.).And also, it should be considered the reflection current from non-proper loads.
Output circuit must be double terminated for stabiliy..

Also, input termination will influence the overall stability ( less than output). And finally , both input and output should be terminated by proper impedances for an exact stability.
Otherwise, the PA will "probably" oscillate and destroy the power transistor.

yeah i have noticed that , and when i terminated , the Input and output , the transistor didn't die , but i didn't get the output power ,

i will make a new board with the new matching network and see how things will go

khouly