VHF RF filters - how many and where?

[jgalak]

I'm designing an APRS tracker using a clock chip as the signal source.  This works well, except the carrier signal (144.39MHz) is a square wave, not a sine.  An LC low pass filter cleans that up nicely.  The signal, however, needs to be amplified, and I'm not sure of the exact way to "process" the output.

I'd like to use some sort of MMIC amplifer to take it from the current output level of about 25mW to somewhere in the 100-500mW range.

So the question is where the filters should go.  I can see at least three options: 
1) source -> amp -> filter -> antenna
2) source -> filter -> amp -> filter -> antenna
3) source -> filter -> amp -> antenna

I have always learned that a low pass filter is required on the output of any amplifer, to block any spurs generated by the amp.  This would tend to rule out #3.

But I'm wondering if I should have any sort of filter between the signal source and the amplifier itself.  Is that better than amplifying all the odd harmonics only to need to filter them back out?  I'd prefer to avoid the extra components involved with a second filter if I can, but not sure what's best practice here. 

Thanks.

[eb4fbz]

2. Otherwise the amplifier will have to deal with the source harmonics, leading to lower efficiency, higher distortion and amplifying them further in some degree depending on the amplifier bandwidth. With just one filter, you would need a much higher selectivity filter to get the same cleaniness at the output, so better to use two simpler filters.

[hagster]

Option 1 is the most power efficient way.

Your amplifier(buffer) is just a on off transistor. Very little power disapater in fully on of off states.

As soon as you switch to analogue signal levels(via the filter) you need to have a analogue amplifier(class A for example).

The next problem you have is impedence matching.

[xaxaxa]

Option 1 is the most power efficient way.

Your amplifier(buffer) is just a on off transistor. Very little power disapater in fully on of off states.

As soon as you switch to analogue signal levels(via the filter) you need to have a analogue amplifier(class A for example).

The next problem you have is impedence matching.

Sidenote: for that to work well ensure the impedance of the output matching network (looking in from the transistor side) is high at the 3rd harmonic frequency; then the amplifier can actually have a square looking waveform at the drain. I think this is called class E or something.

[jgalak]

The signal source is a clock chip with a 50Oh, impedance.  So I was expecting to maintainnthat for the entire rf signal path.  One of the nice things about mmic amplifiers is that they already have 50Ohm inputs and outputs.

[jgalak]

Option 1 is the most power efficient way.

Your amplifier(buffer) is just a on off transistor. Very little power disapater in fully on of off states.

As soon as you switch to analogue signal levels(via the filter) you need to have a analogue amplifier(class A for example).

The next problem you have is impedence matching.

Sidenote: for that to work well ensure the impedance of the output matching network (looking in from the transistor side) is high at the 3rd harmonic frequency; then the amplifier can actually have a square looking waveform at the drain. I think this is called class E or something.

Having now looked at class E amps, that does sound interesting.  It lacks the simplicty of a single chip MMIC, but the potentially lower power is very attractive in my application.  Will need to investigate more.