To whom may concern,
I use AFT09S282N Transistor from NXP (formerly from Freescale) to achieve 53.5dBm (equals to 225Watts) RF peak power in Pulse mode in my transmitter module with the following pulse specifications:
Pulse Width: 250 μsec
Pulse Repetition Interval: 1250 μsec
Using the above parameters the average RF Output power is 45Watts which has a great back off from the transistor maximum average power which is 80Watts as reported at the transistor datasheet Rev. 0, 10/2012.
I have mounted the transistor on a brass carrier (or copper carrier) as a flange and the module is well cooled so the final flange temperature of the transistor is about 70° Celsius at 225W RF peak output power.
The operating working frequency is about 850MHz and the matching networks are realized on a 0.02” Rogers RO4003 PCB.
Gain, Output power and efficiency seems to be normal due to ADS simulations (using the large signal model published by NXP) and my measurements on the module.
The problem is that after continuous operation (several days for example) of my module, the transistors begin to fail on several modules. The transistor gate becomes short circuited.
My questions:
1. Is this transistor suitable for my pulse condition (250μsec/1250μsec) with 225W RF peak power and 45W RF average power?
2. Is 70° Celsius suitable for long term operation of this transistor?
Any additional guidance and recommendations will be appreciated.
Sincerely yours.
1. This transistor can be used in your pulse condition.
2. 70° Celsius case temperature is suitable for long therm operation.
I think, you problem is not thermal management related but caused by electric overstress.
EOS occurs when the maximum voltage potentials have been exceeded across any two terminals of the device inducing device breakdown resulting in excessive currents.
AFT09S282N maximum drain-source voltage is 70V. Most probably this parameter is violated in your design.
Typical reasons of EOS for pulse mode are:
(a) bias feed inductance determines minimum rise/fall time of the signal.
L*dI/dt < Vdss, dt > L*dI/Vdss.
Suppose, bias L = 100nH, Vdss = 70V, dI=10A (for 280W) -> dt > 14ns
It is reasonable to decrease bias inductance. So, if L=5nH -> dt > 0.7ns. It would not cause a problem.
(b) Very low resonance. Bias feed inductance and output capacitance cause resonance at low frequency. If the ramp of the signal is not properly controlled, the signal will cause ringing in the bias feed circuitry and may cause overstress.
By properly controlling the ramp time of the RF signal, the overshoot can be eliminated. RF rise time should be more than 1/F_low_resonance.
Also, it is possible to shift up low resonance frequency by using two bias feed lines (doubles resonance frequency) and/or using shortened feed lines to decrease feed line inductance at low frequency.
(c) RF Reflections – high VSWR.
Have a great day,
Pavel
NXP TIC
-------------------------------------------------------------------------------
Note:
- If this post answers your question, please click the "Mark Correct" button. Thank you!
- We are following threads for 7 weeks after the last post, later replies are ignored
Please open a new thread and refer to the closed one, if you have a related question at a later point in time.
-------------------------------------------------------------------------------
Hi, thank you for your reply and great guidance.
I have some more questions if possible:
Due to the details provided above, and due to this point that I have not changed the transmission lines dimensions and characteristics compared to the original evaluation board presented in the datasheet, and due to about 40nsec rise time of the RF signal, is there any chance to the EOS problem?
Sincerely yours.
Hi, thank you for your reply and great guidance.
I want to send you my private design with some questions would you plz tell me your emaild?