Tradeoffs of filtering exciter vs after PA
 

I'm still working on my HF APRS beacon idea (25MHz-14.85MHz = 10.151)
and I've done some experiments using a 74HC86 quad XOR with one gate for
each oscillator and one as a mixer. In spice simulations, the '86 as a
mixer produces lots of products resulting from the odd harmonics of the
inputs, some of which are close to or below the 10.151MHz I want. If I
run that output through a 2nd order Butterworth 30m bandpass filter I
get a clean result.

My question is about the tradeoffs between filtering the exciter output
to the PA vs filtering after the PA. Obviously a post-amp filter has
to handle more power, but has the opportunity to eliminate amplifier
distortion. But eliminating unwanted inputs to the amp also reduces
unwanted outputs.

Are there rules of thumb for suppression of unwanted signals at each
stage of an amplification chain?

--
Ben Jackson

http://www.ben.com/

Tradeoffs of filtering exciter vs after PA
 

Ben Jackson wrote:
I'm still working on my HF APRS beacon idea (25MHz-14.85MHz = 10.151)

I realize this doesn't really answer your question, but isn't 10.15 the
band edge - and out of the band once you apply modulation?
--
Doug Smith W9WI
Pleasant View (Nashville), TN EM66
http://www.w9wi.com

Tradeoffs of filtering exciter vs after PA
 

On Tue, 20 Jun 2006 13:58:31 -0500, Ben Jackson wrote:

I'm still working on my HF APRS beacon idea (25MHz-14.85MHz = 10.151)
and I've done some experiments using a 74HC86 quad XOR with one gate for
each oscillator and one as a mixer. In spice simulations, the '86 as a
mixer produces lots of products resulting from the odd harmonics of the
inputs, some of which are close to or below the 10.151MHz I want. If I
run that output through a 2nd order Butterworth 30m bandpass filter I
get a clean result.

Since the filter loss depends on the ratio between the loaded-Q and
the unloaded-Q and in order to get a steep bandpass filter, the losses
are going to be large with ordinary LC components.

If you put the steep bandpass filter after the power amplifier, a
large amount of expensive RF fundamental power is lost in the filter.
However, if you put the steep filter before the power amplifier, the
loss of signal level is easily compensated for by an extra low level
amplification stage, with minimal extra cost.

Depending of the linearity of the power amplifier, some low pass
filtering may be needed after it, but now you have to worry only on
the harmonics of the 10 MHz signal, not the various mixing products.
If you use a push-pull power amplifier, the first significant harmonic
would be the 3rd, which is above 30 MHz and since this is a single
band design, it might even make sense to use an elliptic low pass
filter, with a resonant notch at the 3rd harmonic.

Paul OH3LWR

Tradeoffs of filtering exciter vs after PA
 

On 2006-06-20, Ben Jackson wrote:
I'm still working on my HF APRS beacon idea (25MHz-14.85MHz = 10.151)

The replies to this aren't showing up on my news server, but from google
groups I see that

Doug Smith asked:
I realize this doesn't really answer your question, but isn't 10.15 the
band edge - and out of the band once you apply modulation?

With LSB and ~2kHz tones it's actually back *in* the band with modulation.
HF packet is FSK, and the mark/space for APRS are at 10.149.200 and .400.
When implemented as AFSK + LSB + modem you have to take into account the
modem tones, which vary considerably. In fact, you can use USB and tune
below the band as well. If you google "hf aprs" you'll find a bunch of
tables for various radios and TNCs.

--
Ben Jackson

http://www.ben.com/

Tradeoffs of filtering exciter vs after PA
 

On Tue, 20 Jun 2006 13:58:31 -0500, Ben Jackson wrote:

I'm still working on my HF APRS beacon idea (25MHz-14.85MHz = 10.151)
and I've done some experiments using a 74HC86 quad XOR with one gate for
each oscillator and one as a mixer. In spice simulations, the '86 as a
mixer produces lots of products resulting from the odd harmonics of the
inputs, some of which are close to or below the 10.151MHz I want. If I
run that output through a 2nd order Butterworth 30m bandpass filter I
get a clean result.

You ant to do that.


My question is about the tradeoffs between filtering the exciter output
to the PA vs filtering after the PA. Obviously a post-amp filter has
to handle more power, but has the opportunity to eliminate amplifier
distortion. But eliminating unwanted inputs to the amp also reduces
unwanted outputs.

if you do it agter the PA you get those products plus IMD to create
more products and it will be very hard to secure a clean signal
without high losses.

Are there rules of thumb for suppression of unwanted signals at each
stage of an amplification chain?

Start clean, then clean up anything that results from the stage(s).
Usually if you have a clean source and run that through multiple
linear stages you should get the same only bigger with only a
small number of artifacts that are easy to clean up.

Allison

Tradeoffs of filtering exciter vs after PA
 

Ben Jackson wrote:
With LSB and ~2kHz tones it's actually back *in* the band with modulation.
HF packet is FSK, and the mark/space for APRS are at 10.149.200 and .400.
When implemented as AFSK + LSB + modem you have to take into account the
modem tones, which vary considerably. In fact, you can use USB and tune
below the band as well. If you google "hf aprs" you'll find a bunch of
tables for various radios and TNCs.

Ah. Makes sense.

I do direct FSK so have been taking carrier frequencies literally. (and
haven't done datamodes, except CW, on 30 meters so I'm not familiar with
APRS practice on that band)
--
Doug Smith W9WI
Pleasant View (Nashville), TN EM66
http://www.w9wi.com