RF Power Amplifier 20-2500MHz 20W design question

[ddfromuk]

I need to design a linear RF Power amplifier with flat gain of at least 45dB over 20-2500MHz.
flatness +-2dB, P1dB min should be 37dBm, VSWR(in) 2:1

I suppose the technology of choice should be GaN, but could you please recommend the active device and the architecture
that can achieve this.

thanks a lot for your help

 

[Hi-Tone]

Gain is not the issue as you can just put several MMIC (GaN/GaAs) in series. Try rather to look for P1dB for 37 dBm and your return loss. Do you have other specifications such as IP3 or NF?

 

[ddfromuk]

Gain is not the issue as you can just put several MMIC (GaN/GaAs) in series. Try rather to look for P1dB for 37 dBm and your return loss. Do you have other specifications such as IP3 or NF?

Output Power, 20 Watts
P1dB Compressed 12 Watts (which is actually 41dBm, sorry)
input return loss of 10dB
NF is not very important (ie 10dB)

Looking forward to your design recommendations. Thank you!

 

[Hi-Tone]

So there are several options here:
1. Find a decent MMIC within your frequency area and with HIGH P1dB. Then choose two of them to collect the gain you want (your 43 dBm/20W) and design them to be in series.
2. You could find a MMIC with HIGH gain (minimum 25 dBm) and design them in parallel with two power splitters (like a Wilkinson, so you only have 6 dBm loss (2x3 dBm)). It's a bit more complicated since the design has to be exactly the same because of the phase when you combine it again. But with this approach you get a higher total P1dB because the power is split in two.

Alternatively, if you loosen your requirements a little you could use the following MMIC. It has bad return loss, but with some matching you could make it fine. There's an example at the bottom of the datasheet:
**broken link removed**

Best of luck,
Steve

 

[BigBoss]

Power Combining may be trouble because the bandwidth is very wide...Passive power combiners have very high losses..
Distributed Architecture may be an altenative but the design will be pretty difficult because of very high power...
The other ways don't seem to be feasible with this bandwidth and power level...

 

[tony_lth]

The Gain flatness is also difficult.
I think you need some switch to divide 20-2500M into several subbands.

 

[young.microwave.eng]

hi
I needed to design some divider for 900-2500 MHz and I simulated both distributed and lumped element power dividers.the result for distributed power divider was better.But when I use distributed I meant microstrip wilkinson power divider.I don't understand what bigboss says.
and an other point is that if you use balanced amplifier using dividers for example branch line coupler then it is not necessary to consider VSWR.because one of the advantages of such dividers is to make the input and output return losses better
any way I think the design of such a wide band and low frequency divider is complicated

 

[Hi-Tone]

BigBoss is absolutely right. It's hard to find a power divider with that range of that frequency. You most likely have to make it yourself with a lot of stages, but it won't be pretty. As tony says, it might be possible to use switches, but then again, I'm not sure if that is allowed in his project.