Hi,

I'm considering making a class E amplifier for single-sideband output, but
from what I
understand, modulating the plate voltage produces double-sideband. I'd like
to
avoid that if I can.

I'm thinking of taking regular audio, digitally processing it into
single-sideband using a
soundcard, and then digitizing that to 8-bits.

I'd then create 8 class E amplifiers that put out 1/8w, 1/4w, 1/2w, 1w, 2w,
4w, 8w,
and 16 watt outputs for the same signal; modulating each one as CW in
accordance
with the audio bit.

I'd combine the outputs through a power combiner.

Questions:

1) Would this scheme be possible for creating SSB by parallel switching of
CW
amplifiers in accordance with a digitized audio signal?

2) Would I need to create assymetrical power combiners to combine the
amplitudes
of the different signals, or could I get away with simply making a
symmetrical
8-way combiner.

Thanks in advance,

The Eternal Squire


---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.543 / Virus Database: 337 - Release Date: 11/21/03

In concept, it should be possible to generate SSB digitally at any frequency
and any power level. Clearly, it is possible to generate low-level SSB with
a PC sound card, but the output frequency is typically limited by the speed
at which the card can sample, digitize and process the audio. This speed
limitation caps the output at around 15-20KHz with typical sound cards.

The output power delivered to the load is purely a function of how much
voltage the D-A converter can deliver to the load. If, as you suggested, one
took the 8-bit digital signal (pre-D/A converter) amplified each bit in 8
binary-weighted 'lossless' voltage (not power) drivers, and summed the
output in a purely resistive load, then one would have a high-power, high
efficiency SSB exciter.

Here are the problems I see:
1) Speed. The A/D, DSP, and D/A have to be able to process audio samples at
twice the highest RF output frequency (2.5-3.0 times in practicality)
2) Precision: The high power weighted binary output drivers need to produce
very accurate voltages
3) Spectral purity: The output spectrum will have aliases and 'birdies'. The
number and amplitude of these is a function of the number of bits in the
digitizing process, and the accuracy of the high-power D-A at the output.
4) Complexity: Enough said about this.

A more practical way to get a similar result might be to generate low-level
SSB in a high-speed DSP, and amplify the analog result it in a class-B
amplifier whose bias is dynamically adjusted by the DSP processor in
anticipation of the instantaneous RF level. This should produce a
highly-quality signal with good overall power efficiency.

Joe
W3JDR

Check out my project page: http://mysite.verizon.net/jdrocci/.



"The Eternal Squire" wrote in message
...
Hi,

I'm considering making a class E amplifier for single-sideband output, but
from what I
understand, modulating the plate voltage produces double-sideband. I'd
like
to
avoid that if I can.

I'm thinking of taking regular audio, digitally processing it into
single-sideband using a
soundcard, and then digitizing that to 8-bits.

I'd then create 8 class E amplifiers that put out 1/8w, 1/4w, 1/2w, 1w,
2w,
4w, 8w,
and 16 watt outputs for the same signal; modulating each one as CW in
accordance
with the audio bit.

I'd combine the outputs through a power combiner.

Questions:

1) Would this scheme be possible for creating SSB by parallel switching
of
CW
amplifiers in accordance with a digitized audio signal?

2) Would I need to create assymetrical power combiners to combine the
amplitudes
of the different signals, or could I get away with simply making a
symmetrical
8-way combiner.

Thanks in advance,

The Eternal Squire


---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.543 / Virus Database: 337 - Release Date: 11/21/03

In concept, it should be possible to generate SSB digitally at any frequency
and any power level. Clearly, it is possible to generate low-level SSB with
a PC sound card, but the output frequency is typically limited by the speed
at which the card can sample, digitize and process the audio. This speed
limitation caps the output at around 15-20KHz with typical sound cards.

The output power delivered to the load is purely a function of how much
voltage the D-A converter can deliver to the load. If, as you suggested, one
took the 8-bit digital signal (pre-D/A converter) amplified each bit in 8
binary-weighted 'lossless' voltage (not power) drivers, and summed the
output in a purely resistive load, then one would have a high-power, high
efficiency SSB exciter.

Here are the problems I see:
1) Speed. The A/D, DSP, and D/A have to be able to process audio samples at
twice the highest RF output frequency (2.5-3.0 times in practicality)
2) Precision: The high power weighted binary output drivers need to produce
very accurate voltages
3) Spectral purity: The output spectrum will have aliases and 'birdies'. The
number and amplitude of these is a function of the number of bits in the
digitizing process, and the accuracy of the high-power D-A at the output.
4) Complexity: Enough said about this.

A more practical way to get a similar result might be to generate low-level
SSB in a high-speed DSP, and amplify the analog result it in a class-B
amplifier whose bias is dynamically adjusted by the DSP processor in
anticipation of the instantaneous RF level. This should produce a
highly-quality signal with good overall power efficiency.

Joe
W3JDR

Check out my project page: http://mysite.verizon.net/jdrocci/.



"The Eternal Squire" wrote in message
...
Hi,

I'm considering making a class E amplifier for single-sideband output, but
from what I
understand, modulating the plate voltage produces double-sideband. I'd
like
to
avoid that if I can.

I'm thinking of taking regular audio, digitally processing it into
single-sideband using a
soundcard, and then digitizing that to 8-bits.

I'd then create 8 class E amplifiers that put out 1/8w, 1/4w, 1/2w, 1w,
2w,
4w, 8w,
and 16 watt outputs for the same signal; modulating each one as CW in
accordance
with the audio bit.

I'd combine the outputs through a power combiner.

Questions:

1) Would this scheme be possible for creating SSB by parallel switching
of
CW
amplifiers in accordance with a digitized audio signal?

2) Would I need to create assymetrical power combiners to combine the
amplitudes
of the different signals, or could I get away with simply making a
symmetrical
8-way combiner.

Thanks in advance,

The Eternal Squire


---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.543 / Virus Database: 337 - Release Date: 11/21/03

"The Eternal Squire" wrote in message ...

I'd then create 8 class E amplifiers that put out 1/8w, 1/4w, 1/2w, 1w, 2w,
4w, 8w,
and 16 watt outputs for the same signal; modulating each one as CW in
accordance
with the audio bit.

I'd combine the outputs through a power combiner.

Questions:

1) Would this scheme be possible for creating SSB by parallel switching of
CW
amplifiers in accordance with a digitized audio signal?

nope, you will instead need to apply simultaneous frequency modulation
to the carrier as well as amplitude modulation.

2) Would I need to create assymetrical power combiners to combine the
amplitudes
of the different signals, or could I get away with simply making a
symmetrical
8-way combiner.


you will need an infinite way combiner, the easier way is to frequency
modulate the carrier so that it shifts to wherever you want the
carrier to be and control the envelope (this is a white lie, the
acutal explanation is a little deeper).

the dsp software required is not going to be easy. i have written a
dsp shell for the sound card, you can download from
http://farhan.net.co.nr and play with it. once i am finished with my
VHF transceiver i plan implement a hilbert transform using this shell
and make a simple SSB exciter with it.

- farhan

"The Eternal Squire" wrote in message ...

I'd then create 8 class E amplifiers that put out 1/8w, 1/4w, 1/2w, 1w, 2w,
4w, 8w,
and 16 watt outputs for the same signal; modulating each one as CW in
accordance
with the audio bit.

I'd combine the outputs through a power combiner.

Questions:

1) Would this scheme be possible for creating SSB by parallel switching of
CW
amplifiers in accordance with a digitized audio signal?

nope, you will instead need to apply simultaneous frequency modulation
to the carrier as well as amplitude modulation.

2) Would I need to create assymetrical power combiners to combine the
amplitudes
of the different signals, or could I get away with simply making a
symmetrical
8-way combiner.


you will need an infinite way combiner, the easier way is to frequency
modulate the carrier so that it shifts to wherever you want the
carrier to be and control the envelope (this is a white lie, the
acutal explanation is a little deeper).

the dsp software required is not going to be easy. i have written a
dsp shell for the sound card, you can download from
http://farhan.net.co.nr and play with it. once i am finished with my
VHF transceiver i plan implement a hilbert transform using this shell
and make a simple SSB exciter with it.

- farhan

In article ,
says...

Hi,

I'm considering making a class E amplifier for single-sideband output, but
from what I
understand, modulating the plate voltage produces double-sideband. I'd like
to
avoid that if I can.

I'm thinking of taking regular audio, digitally processing it into
single-sideband using a
soundcard, and then digitizing that to 8-bits.


Base audio has no sidebands, it is monoband so to speak. It does not become
sideband until you modulate a carrier, remove the carrier, and possibly also
remove one sideband. Even if you simulate this process in a sound card, which
can be done, and you produce the 'garbly' sound, you still have 'audio' which
when impressed upon your transmit carrier will still create normal double
sideband modulation. The only way to get the result you are looking for would
either to be taking your digitally generated single sideband and run it through
a doubly balanced mixer with your intended carrier frequency, or perhaps
digitally mixing your digital sound source with a digital representation of a
sinewave who's repetition rate is the same as your intended carrier frequency.
Your digital power output idea is a good one however.

In article ,
says...

Hi,

I'm considering making a class E amplifier for single-sideband output, but
from what I
understand, modulating the plate voltage produces double-sideband. I'd like
to
avoid that if I can.

I'm thinking of taking regular audio, digitally processing it into
single-sideband using a
soundcard, and then digitizing that to 8-bits.


Base audio has no sidebands, it is monoband so to speak. It does not become
sideband until you modulate a carrier, remove the carrier, and possibly also
remove one sideband. Even if you simulate this process in a sound card, which
can be done, and you produce the 'garbly' sound, you still have 'audio' which
when impressed upon your transmit carrier will still create normal double
sideband modulation. The only way to get the result you are looking for would
either to be taking your digitally generated single sideband and run it through
a doubly balanced mixer with your intended carrier frequency, or perhaps
digitally mixing your digital sound source with a digital representation of a
sinewave who's repetition rate is the same as your intended carrier frequency.
Your digital power output idea is a good one however.

(gudmundur) wrote in message ...

Base audio has no sidebands, it is monoband so to speak. It does not become
sideband until you modulate a carrier, remove the carrier, and possibly also
remove one sideband.

not true. consider that you are whistling a pure tone of 1KHz into an
SSB transmitter's input. If the SSB transmitter was tuned to 7.000MHz,
you would get just a carrier at 7.001MHz. Thus, a single tone SSB
signal is virtually indistinguishable from just a carrier.

now consider that the human voice is expressed as a number of
simultaneous pure tones superimposed over each other (easily obtained
by converting the time domain signal into frequeny domain).
Now imagine that for each of the tones, you have a separate
transmittter offsetted from the carrier by a frequency obtained by
frequency analysis and also with equivalent amplitude. this is in
effect, the SSB signal.

to rephrase, we need to modulate a carrier and then remove it because
that is seemingly the easiest way to generate SSB (but not a necessary
way).

Even if you simulate this process in a sound card, which
can be done, and you produce the 'garbly' sound, you still have 'audio' which
when impressed upon your transmit carrier will still create normal double
sideband modulation. The only way to get the result you are looking for would

there is an entirely different approach, we are advocating that we
split the signal into a number of component audio frequencies and
essentially run a carrier that is the equivalent of each of those
frequencies.

From the above example of a single tone, lets expand this to a two
tone ssb exciter. Lets imagine that such a signal can be represented
by two carriers that have relative amplitudes of the two tones and are
shifted from an imaginary carrier by amounts equivalent to their tone
frequencies.

a frequency modulated signal generates sidebands according to a bessel
function. Thus a correctly modulated FM signal will also result in two
tones being generated at two frequencies. In order to control multiple
sidebands, an envelop shaping will have to be applied to the FM
signal. The frequency modulation and the amplitude modulation can be
easily computed digitally.

What is really difficult is to accurately frequency modulate the
carrier. One way is to directly generate the carrier digitally using
DDS. I am not aware of any amateur effort in trying this method out. I
seem to have vaguely read about it in either EMRFD or the new ARRL
Handbook.

- farhan

(gudmundur) wrote in message ...

Base audio has no sidebands, it is monoband so to speak. It does not become
sideband until you modulate a carrier, remove the carrier, and possibly also
remove one sideband.

not true. consider that you are whistling a pure tone of 1KHz into an
SSB transmitter's input. If the SSB transmitter was tuned to 7.000MHz,
you would get just a carrier at 7.001MHz. Thus, a single tone SSB
signal is virtually indistinguishable from just a carrier.

now consider that the human voice is expressed as a number of
simultaneous pure tones superimposed over each other (easily obtained
by converting the time domain signal into frequeny domain).
Now imagine that for each of the tones, you have a separate
transmittter offsetted from the carrier by a frequency obtained by
frequency analysis and also with equivalent amplitude. this is in
effect, the SSB signal.

to rephrase, we need to modulate a carrier and then remove it because
that is seemingly the easiest way to generate SSB (but not a necessary
way).

Even if you simulate this process in a sound card, which
can be done, and you produce the 'garbly' sound, you still have 'audio' which
when impressed upon your transmit carrier will still create normal double
sideband modulation. The only way to get the result you are looking for would

there is an entirely different approach, we are advocating that we
split the signal into a number of component audio frequencies and
essentially run a carrier that is the equivalent of each of those
frequencies.

From the above example of a single tone, lets expand this to a two
tone ssb exciter. Lets imagine that such a signal can be represented
by two carriers that have relative amplitudes of the two tones and are
shifted from an imaginary carrier by amounts equivalent to their tone
frequencies.

a frequency modulated signal generates sidebands according to a bessel
function. Thus a correctly modulated FM signal will also result in two
tones being generated at two frequencies. In order to control multiple
sidebands, an envelop shaping will have to be applied to the FM
signal. The frequency modulation and the amplitude modulation can be
easily computed digitally.

What is really difficult is to accurately frequency modulate the
carrier. One way is to directly generate the carrier digitally using
DDS. I am not aware of any amateur effort in trying this method out. I
seem to have vaguely read about it in either EMRFD or the new ARRL
Handbook.

- farhan