315-446MHz 27dbm (0.5w at 446MHz) RF amp

[kilohercas]

Hello,
I am making universal garage opener/RF beacon for my motorcycle, and i hit one problem. I simply can't get 0.5W at 446MHz at high efficiency.

My setup:
I have SPIRIT1 capable of 11.6dbm of power at 315-446MHz range. Testes working

Made very high efficiency band-pass harmonic filter, with -55dbc for 315MHz and -70dbc for 446MHz that can handle power levels i need. (7 order Chebyshev filter with 1-2db attenuation)

Modulation type is OOK

After filtering i have around 10dbm of power at 446MHz. I need 15mW at 315-433MHz that is very easy to any RF amp, but 446Mhz gives me problems. Most of RF amps work great in CW mode, my favorite would be CMX901 and CMX902, but they are extremely bad at OOK modulation (but work great in FM or any other constant power modulation type)

MMZ09332BT1 works again in CW mode, i can get in CW around 15-20% efficiency, and it uses 200mA no mater is there is signal or not. And also very unstable, keeps blowing up in OOK mode, while it was good in CW mode.

MAAM-009286-TR3000 Also works, but it also does the same thing of using a lot of power even at idle, again 130mA and it is about 11% efficient.  Also can't reach 27dbm and does produce tones of distortion.

AFM907NT1 can produce power but again, extremely inefficient at low power levels. Somehow 27dbm is a spot where not much RF''amps exist.

Will try eGaN EPC8002 device, this is my last hope.

Device is battery powered and use 4.2V, while i can boost voltage with voltage converter.

So, any ideas how to make efficient ( 50% would be nice for battery powered device) that can give at least 15mW at 315-433Mhz range and can provide high power for 446MHz RF beacon ??

[Da_Stier]

Hi,

first of all I wanted to say that this post propably isn't a real solution to your problem, so don't be dissapointed, however I have a few ideas and points to add.

Considering your low efficiency, I don't think that you can get much better than maybe 25%.
RF transistors need a (relatively high) bias current to operate in the linear region needed for amplification. This current stays mostly constant, as long as you don't push to the max. limit (as you already mentioned). Since OOK has a lot of off time, the efficiency depends directly on the Dutycycle and therefore on the data being sent.
Most amplifers have the best efficiency if they are driven quite hard but even then the you hardly reach more than 50%.
On the bright side, you can propably get 1W output power before the current will go up after all 

The only thing you can do to improve this a little bit is to switch off the amplifier, as long as there is no signal transmitted.

Another possiblity would be to find a very unlinear amplifier that also has a very small bias current. For larger power amplifiers, this would be a Doherty amplifier or even better a class E amplifer, however I don't think there are any in your power range, especially not as a single IC.

The distortion problem as well as the amplifers blowing up are due to the fast swichting of an OOK signal, which produces a lot of harmonics. This will add a lot of signal power outside your wanted band. The easiest thing to do here is propably to have enough bandwith to accept like 5 harmonics of the carrier frequency.

The better solution after all would be to genearte a 27dBm CW tone of the frequency you need and use a fast RF switch to do the OOK Modulation on the full power signal.
This way you don't need an amplifier that can handle the pulsed power.
I have seen this on quite a few applications where ASK was needed which is quite similar to OOK after all.

Hope this helps or atleast gives you some ideas.

 
Greetings,
Michael

[johnf]

Firstly your filter with a 2dB loss that is a 40% loss of your output power so you are going to need more than +30dBm.
The best solution would be to design a single transistor stage to give the required amplification. Your choice of 4.5 volts is quite limiting as it limits the available choices and gain /efficiency available
there is a neat little amp family that is ideal
AFIC901N.pdf

[SteveN87]

I see that the SPIRIT1 offers an 8-level ramp-up/down in ASK but (oddly) not in OOK mode. Could you achieve the same result (modulation-wise) using ASK mode? The ramp may help reduce the switching harmonics and save the PA from blowing up.