Device failure after application of modulated signal,

Hello,

Recently came across an interesting problem. In a circuit, a Triquint devicce was being used which has a capability of 39 dBm P1dB. The deivce was working fine until a modulated input was applied. The output power then dropped to 30 dBm and also the device failed (died :) ). Without a modualated input, they were able to achieve 39 dBm. Just as a point in discussion, we thought that maybe there was a problem with the biasing circuit which is probably affected the dynamic range. Is the thought process correct or is is atleast along the right path. If not, what are the possible factors affecting this change in performance once a modulated signal is applied?

Thanks.

just a guess, but the two things I would look for would be some low frequency oscillation, perhaps at 20 MHz, due to the bias network. The modulation pushed it over into oscillation. the other thing would be simple thermal runaway--insufficient cooling.

Maybe the ALC problem, what is your power detector?

Perhaps the reason of the failure is the Peak-to-Average of the modulated signal. Most of the single carrier digital modulations have a Peak to Average in a range of about 4dB to 8dB. Multi-carrier modulation types (as OFDM) have even higher peaks.
If your PA works fine at P1dB using a CW signal, when the input reach higher peaks (due to the modulation), the PA is possible to die.

I'm thinking along the line of vfone. I've seen other junior engineers make this mistake with high-power PAs and blow finals before. A +39 dBm CW signal is NOT the same as a +39 dBm modulated signal. In the latter case, +39 dBm is the average power, but, since it's modulated it will have some PAR (Peak-to-Average Ratio). If the signal has a 5 dB PAR, then a +39 dBm signal will have peak power pulses that would drive the PA to 39+5 = 44 dBm! Unless the PA will compress without detrimental effects, can handle the extra dissipated power, and can survive the higher peak input voltages, it may die.

If you run a modulated signal through a PA, you generally run it "backed-off". That is, if your signal has a 5 dB PAR, and your peak signal excursions need to be at P1dB (+39 dBm), then your average output power should be 39-5 = +34 dBm when running the modulated signal. That way, during the moments when it's outputting peak power, the finals won't exceed your desired maximum operating power.

Setting Pmax to P1dB also keeps the signal amplification more linear, and reduces amplitude distortion due to the peaks going into compression. This is a form of class AB operation... good linearity and slightly better efficiency than a simple class A amp.

Thanks for the replies.

I checked and :

1. The input modulated signal is in the range of 0 dB.
2. The detector is a PIN diode detector IC.
3. When the device is said to have failed, there is a increase in the current consumption.

In the meanwhile, am checking w.r.t. biff44's suggestions regarding the low freq. osc. issues. WIll post the observations here.

Thanks.

So that's maybe your detector problem, your detector is a peak-voltage detector, not a RMS detector. RMS detector is a power detector, you should use that.
This is almost the same as vfone suggestion.