MRF300AN and BN

[joel_tissot]

Hi

I worked on a rf amplifier for HAM used from 1.800 to 50200 Mhz using both MRF3000AN and BN working in push pull.

I found severals drawings on the net and follow the informations but for the 4th time MRF300 is destroy.

Input rf power is 5w and expect 400w output trought appropriate LPF to work on different frequencys.

SWR is controled like current, voltage and temperature, no alarm turn on, all parameters are ok but transistor suddenly goes up.

Here in attached file a picture of a transistor after damage.

My knowledge in RF is not good enought to solved this problem.

Bias current set to 5 v 200mA on each MRF, main power voltage is about 50~52 volts.

What i doing wrong ? why always only BN is damage ?

[mikePJ1967]

It may be too late now, but the back of that tab looks burnt - are you relying on the bolt through to the heatsink to carry the DC as well as the heat ? Is the heat sink not perfectly flat ?  A very small air gap can be fatal, and creams and pastes are  tens of times worse as  thermal conductors than metal, but equally air is several tens of times worse again. The thinnest most intimate contact layer is essential and it needs not to work loose as it heats.. You mention temperature monitoring - are you measuring the device at the exposed tap windows on the top, or are you really measuring the heatsink temperature at some remote point - these can be several degrees apart - even ten of degrees during a rapid heating event. What was the supply current and output power during and after the failure ? More info is needed to comment intelligently on problems like this. M

[ct1eex]

I have built a 600W amplifier using 2 x MRF300 AN/BN     i only blew up one of them , by excess input power. I adjust my bis for 180ma on each Transistor , it is around 2,5volts in the gate of each.

I have a 10db attenuator  on the board input , normally i drive  the amp with 20/35 w. 

I get less power on the extremes bands , it is normal about 550W   160mts  and 450W   6mts  all the other bands i get  500/600 W .

My linear some days workin 4 or 5 hrs in FT8 non stop  , works very good with 51volts.

My palette is from Grece , nice working.

A.Pires

ct1eex

[LPP]

1.
General reason of the fail is that one or more of the operating parameters violates maximum ratings specified in the datasheet.

VDSS [133V max] is the maximum voltage potential that can be applied across drain to source of the LDMOS device. Instantaneous voltages can exceed these value for short periods of time, but static DC voltages above the maximum rating will lead to drain-source breakdown (off-state breakdown). In a reliable RF amplifier design, drain voltage never exceeds breakdown value during normal operation.

Common reason of the VDS overvoltage is that an amplifier operates at high VSWR. It causes electric overstress resulting in excess power dissipation and subsequent melting of the drain runners. The end result is that at least two if not all three of the terminals (drain, gate, source/body) are electrically shorted together.

There might be different reasons of High VSWR:
- the load is mismatched at the target frequency because of poor design
- the load impedance is not nominal or uncontrollable (industrial application)
- the load is mismatched because of operating at a frequency out of the design target band
- amplifier is not stable and oscillates at an arbitrary frequency

Also, see the attached document.

2.
Take precautions to avoid fails during initial board bring up. I recommend to read the section "Precautions for RF Power Transistors" in this document from Polyfet:

http://www.polyfet.com/rftopics.pdf

Use pulse test RF signal to reduce average power dissipated in the device that helps to split electrical and thermal issues.

3.
I don't think there is any reason for BN to damage more often than AN. Probably, all the failed parts were from the same lot.

Note. Production parameters of the transistors (gate threshold..) vary in wide range.
Bias control of an LDMOS FET is done with DC control of the gate voltage. The gate voltage controls quiescent drain current, or IDQ, which is set with no RF present. Each LDMOS transistor will require a unique gate voltage setting for the optimum drain current in a particular power amplifier design.

If you use common gate bias for AN and BN, the quiescent currents of the sides will be different and it may be a reason for their different sensitivity to overstress.

Best Regards,

Pavel

NXP CAS