LPF after or before power amplifier?

[neazoi]

Hello,
Why should many include a LPF AFTER the power amplifier and not BEFORE?
It seems to me more suitable to cut any unwanted harmonics before they are amplified.

Any ideas?

 

[albbg]

The power amplifier itself generates harmonics that have to be reduced.

 

[BradtheRad]

There is a line of thinking that advocates bi-amping. Use two power amps, one for low frequencies and another for highs.

The reason is that a single amp needs to have ten times the power in order to amplify the entire spectrum.
Whereas with bi-amping you can use two smaller, less expensive amps. As well as a smaller, less expensive power supply.
It's hard to argue with that logic.

Article at website:

**broken link removed**

 

[godfreyl]

I strongly suspect Neazoi's talking about RF amplifiers, whereas Brad's response refers to audio amps. Looks like a slight misunderstanding?

 

[BradtheRad]

I strongly suspect Neazoi's talking about RF amplifiers, whereas Brad's response refers to audio amps. Looks like a slight misunderstanding?

I believe you're right. His avatar has the word 'microwave' in it.

 

[neazoi]

The power amplifier itself generates harmonics that have to be reduced.

Yes, a filter before AND after the RF amplifier would be ideal. But filters are expensive to switch in a multiband approach where filter switching needs to be done.
Furthermore, if used at the output of the amplifier, the filter components have to withstand the RF power.

Would the amplifier intermodulation products be too high, if a bandpass filter is used at the input?
Then less expensive components can be used. I am talking for something like 5W output from a broadband HF amplifier, with a BPF at the input.

 

[chuckey]

if you are using a broadband HF amp (say 1-30MHZ), if you follow it with a band pass filter, it must have this sort of band pass, so a 1MHZ signal will whistle through it complete with all its harmonics. Same argument for an input filter. In actual real life you would need a resonant aerial tuning unit which you will have you adjust for every significant change in frequency. So you will have one anyway !
Frank

 

[hamidrezakarami]

I think the amplifiers usually connected to the transmission lines or the antennas, so the inter-modulation terms can not propagated in them. using the sharp filter after the amplifiers can be eliminated these terms.

 

[chuckey]

The most imporetent i/ps are always within the pass band and are the third order products 2F1-F2 and 2F2 - F1. Band pass filters can only get rid of harmonics, 2 F1, 2 F2 etc.
Frank

 

[FvM]

Why are expecting harmonics in the input signal? Which signal source are you imagining?

Classical small band power amplifiers are mostly build with LC impedance matching between amplifier stages, which is usually designed as low-pass. In so far filters on both sides of the PA are quite standard. They are simple circuits and not "expensive".

 

[neazoi]

Why are expecting harmonics in the input signal? Which signal source are you imagining?

Classical small band power amplifiers are mostly build with LC impedance matching between amplifier stages, which is usually designed as low-pass. In so far filters on both sides of the PA are quite standard. They are simple circuits and not "expensive".

I am talking about simple oscillators that have harmonics. Most radio amateurs do not include input filters but just LPF after the amp

 

[FvM]

Simple oscillators don't have much harmonics, PA always have. In so far the standard solution is reasonable.

 

[neazoi]

Simple oscillators don't have much harmonics, PA always have. In so far the standard solution is reasonable.

Hm... ok, so it would be better to follow the standard LPF after the amplifier I think.
My guess was that if including this filter before the power amplifier smd components could be used, since the power reguirements would be low at that stage. But at the output of the power amplifier I do not know if smd can tollerate the power (simething like 5W rf power output)

 

[FvM]

The problem is, that even a pure class A power stage driven by clean carrier signal most likely won't comply with the harmonic suppression requirements in any applicable regulation. Thus you can hardly avoid a low-pass filter. If it can be implemented in SMT for 5W depends on the frequency band, I think.