How to improve the performance of a power amplifier-TGA2508

[LTCC]

Hei, guys. I recently has tested the triquint PA MMIC-TGA2508 integrated with Roger5800 , and can't get the output power at the recomanded bias condition Vd=7V, Vg=-0.36V. Actually, all the performance is worse than the spec. in the datasheet. The P1dB output power i get is 23dBm with a gain of 25dB.

The best P1dB output power i have is 26dBm with a gain of 28dB at 16GHz when i adjust the gate and drain voltage to Vd=4V, Vg=-0.44V. But still not as good as the performance in the datasheet.

Does anyone have similiar problem? Not limited to this mmic. When you use the MMIC PA, do you follow recomanded bias condition and get the good results, or adjust it a little and also good results? or just like me.

The problem is that I don't know where the problem is? Or the problem is the mmic itself?

Here is the datasheet of TGA2508. Any advices will be appreciated.
The spec. of TGA2508 is 30dBm P1dB, 30dB gain in the freq band 12~19GHz.


Sherry

 

[LTCC]

Any one has similiar experience?

 

[BigBoss]

Have you ever taken into account the bondwires ?? because bonding should be in according with manufacturer's specs.Otherwise, you loose the specified performance because the operating frequency is very high.

 

[LTCC]

Re: How to improve the performance of a power amplifier-TGA2

Have you ever taken into account the bondwires ?? because bonding should be in according with manufacturer's specs.Otherwise, you loose the specified performance because the operating frequency is very high.

Thanks for opinion. Yes, the specs in datasheet is tested on wafer and the probe, and i test it using bondwires and the RF connectors, but i don't think the bondwires would cause so much loss since it's less than 8mil. The operating frequency is high, maybe that is the reason.

 

[BigBoss]

Re: How to improve the performance of a power amplifier-TGA2

Have you ever taken into account the bondwires ?? because bonding should be in according with manufacturer's specs.Otherwise, you loose the specified performance because the operating frequency is very high.

Thanks for opinion. Yes, the specs in datasheet is tested on wafer and the probe, and i test it using bondwires and the RF connectors, but i don't think the bondwires would cause so much loss since it's less than 8mil. The operating frequency is high, maybe that is the reason.

8mil is approx.0.2mm.( @15GHz it's equivalent to 18Ohm !!)
This length has around 0.2nH each and in additional to this, measurement pads and their parasitic capacitances will be added..
So, finally measuring the circuit on its die is quite different with bonding wires and associated pads..

 

[LTCC]

Re: How to improve the performance of a power amplifier-TGA2

Have you ever taken into account the bondwires ?? because bonding should be in according with manufacturer's specs.Otherwise, you loose the specified performance because the operating frequency is very high.

Thanks for opinion. Yes, the specs in datasheet is tested on wafer and the probe, and i test it using bondwires and the RF connectors, but i don't think the bondwires would cause so much loss since it's less than 8mil. The operating frequency is high, maybe that is the reason.

8mil is approx.0.2mm.( @15GHz it's equivalent to 18Ohm !!)
This length has around 0.2nH each and in additional to this, measurement pads and their parasitic capacitances will be added..
So, finally measuring the circuit on its die is quite different with bonding wires and associated pads..

How do you come to 18ohm@15GHz, i don't get it ? Can you explain? Thank you.

 

[BigBoss]

Re: How to improve the performance of a power amplifier-TGA2

How do you come to 18ohm(at)15GHz, i don't get it ? Can you explain? Thank you.

As a rule of thumb, I assumed 1mm=0.8nH
Your bondwire has 0.2nH (maybe more or less because of mutual couplings ) inductance value that is equal to 18Ohm inductive reactance @15GHz
And don't forget the pad parasitic capacitances which are tied to measurement pads regarding to ground.

 

[LTCC]

Thank you so much, Bigboss