开云体育

We've got an interesting discussion going over in the qrplabs group regarding
a possible way to transmit SSB signals much more efficiently (and cheaply) than
through the use of a linear amp:? ??/g/QRPLabs/topic/29572792

EER is an acronym for Envelope Elimination and Restoration.
Traditionally, this apparently meant to build an SSB exciter, then amplify and limit the?
output such that all that remains is a square wave with the frequency information.
Amplify that square wave efficiently with a class C or D or E amp, and restore the
envelope information by modulating the power supply to that amp.
The envelope information could be obtained with a diode detector that watches the output
of the SSB exciter.

A more interesting approach might be to do away with the SSB exciter entirely.
Audio comes into a uC through an ADC from the mike, calculations are performed
on that audio to compute the desired frequency and envelope information,
then the uC steers the frequency of a PLL (such as the si5351) and also modulates
that final amp with envelope information.

Seemed only fair to mention it here.
One possible piece of the puzzle might be to modulate an si5351 clock by
jerking around the reference oscillator as Farhan first described in this forum:
? ? /g/BITX20/message/34020

Jerry, KE7ER?

Jerry

Be careful what you spread. Of course feeding accurate info into a D to A converter can produce the voice, and with some predistortion correct for non linearity in the power stages. But modulating a PLL, that sounds like dreadful analog complexity. I suggest emailing an expert on PA designs, like Paul NA5N versus a debate. Anything that seems real can be tried, but this seems a bit unwieldy to me, but I don't have expertise in modulation.

Curt

I think that would be a cool way to build a SSB transmitter. I've been doing some reading as I'm investigating the possible use of a FPGA for creating a SSB signal to drive my uBITX (in this case the signal will be feeding a linear TX chain). In the analog days it seems the results were somewhat marginal. One technique described splitting things up so the carrier was "FM" modulated (I imagine that is an over simplification) and the envelope was used to modulate the carrier with another class D audio range amplifier modulating the voltage to the class E final.

It would seem that in this age of DSP processors and such this should be a realizable design. In my limited searching I haven't found much on realized designs. I'm really interested if others with more experience in this, know of working designs.?

I'd been thinking about building a tiny, portable, low power battery operated digital mode HF transceiver. By limiting it to just digital modes with constant power modulation types, it could use a class E output and be easy on batteries.?

Tom, wb6b

Guido has a first pass at something that sort of works, running on a QCX.
In his case, he avoids analog entirely, PLL updates are carried out over the i2c bus into the Si5351.
If that proves not precise enough, perhaps use the $2 Si5351B, which has a "VCXO" pin, could be driven with a DAC.
The challenging (dreadful?) part will likely be in developing the DSP code that steer this thing,?
trying to create an acceptably clean signal.

No need for a debate here, this is just an invite to join the conversation if interested in technical puzzles.
And no need to be careful, these are all $1 and $2 parts.?

If you want an expert designing your gear, go buy an Elecraft.
But why should they have all the fun?

Jerry

That's really interesting. I googled Guido and QCX and found the thread on the QRP Labs group and another interesting hit.

I'll include the links here for anyone interested.

/g/QRPLabs/message/30792

This is a fun topic with some interesting experimentation going on.?I think FT8 could be generated just by updating the Si5351 over i2c?à la what CEC did for WSPR in his firmware. But that is a side issue not directly related to the SSB topic.

Yes, the DSP code for a digital approach would be interesting.

Tom, wb6b

On Fri, Feb 1, 2019 at 09:35 PM, Tom, wb6b wrote:
I think that would be a cool way to build a SSB transmitter. I've been doing some reading as I'm investigating the possible use of a FPGA for creating a SSB signal to drive my uBITX (in this case the signal will be feeding a linear TX chain). In the analog days it seems the results were somewhat marginal. One technique described splitting things up so the carrier was "FM" modulated (I imagine that is an over simplification) and the envelope was used to modulate the carrier with another class D audio range amplifier modulating the voltage to the class E final.

The central feature is doing half the work DSP at 16mhz (arduino uno or nano speeds) and a little analog.

Yes the carrier is sorta FM... what it is and needs to be is the mic audio signal stripped to the frequency content
and added (or subtracted) to the base frequency.? FM tends to revolve around a central frequency and has
more sidebands.? In the analog world we would hard limit the audio and mix it with a carrier the result would
have sidebands of the base plus sum (or difference).? That limiting thing in the digital realm takes processor
speed. The other half is then modulating an amplifier with a signal that represents the envelope of the audio
at audio rate and is used to re-create the power out for the driving level.? ?The simplest form of that is not
more than 100years old and known as AM save for that has carrier(largely useless save for the detector at
the other end needed it then) and both sidebands (both sum and difference of the source signal applied
to carrier).? SSB removes the carrier? and the receiver uses a BFO as replacement.? the excess
sideband can be filtered away or using Analog signal processing, we can inhibit the unwanted
sideband.? Phasing side band is more than 80 years old.? Filter mode of sideband has about the same
origin.

All of the digital techniques do that and its all math but at some point the result is RX audio or TX RF.
The fastest designs grind the numbers and drive an D/A and feed the rest directly to the antenna
without the need for frequency change and on RX grab the spectrum using a very fast A/D and
process the data down to the desired signal.? Just about all the radios with IF DSP do this but
at 12 or 30khz with conversion to user frequency as that was were the tech was 20 years ago.

>>>It would seem that in this age of DSP processors and such this should be a realizable design. In my limited searching I haven't found much on realized designs. I'm really interested if others with more experience in this, know of working designs. <<<<

If you have a MPU fast enough to do dsp (STM32F4 at 180mhz is fast enough) then the mainline techniques
can get you a very good SSB signal and also do the RX as well.

As to realized designs I give you softrock (RX), SDR2GO, Flex radio, ANAN 100, KX2, KX3, K3S, IC7300...
And just about any IF DSP radio in the last 20+ years.? However there are several ways to do it and the
EER form we are talking about has never been very successful. One amp I know of was SGC 500W,
very short market life.

Allison

I have about 5 papers on this some really old.

Bottom line, it sorta works.

Allison

On Sat, Feb 2, 2019 at 09:05 PM, ajparent1/KB1GMX wrote:

The homebrew G2DAF class C?linear amplifier of many years ago was a successful example.

73 Bob GM4CID

?

Define successful.? Its also a bit of a jump from EER amp to generating SSB?
using the Khan technique.

Right now most of the AM broadcast market is class D or E but that is far easier and?
extremely successful.? Lot of the local to me hams run that.? They have a really
nice signal and clean spectrum and run the full 1.5Kw 100% modulated signal
(about 400W carrier).

Allison