Re: LF: MSK with a Class D transmitter...?

[Andy Talbot <andy.g4jnt@gmail.com>]

If you can generate the message by replaying from a .WAV file, this ought to be straightforward.   Change the sampling rate value in the .WAV file header and it will replay at that rate.   But... of course ...  you need to be able to generate the file in the first place.

 

The real solution is to do away with the rather pointless divide by two on the input circuitry of these PAs.  

It is not at all necessary and just makes things complicated.

 

If you can ensure the drive waveform is close to a 50% duty cycle, it can be used directly to drive push-pull FETs after squaring up and generating 180 degress out of phase versions - ie  a single inverter..  

 

One way to get a good square waveform out is first to pass your input waveform through a low pass filter, cutting  off sharply just above the highest baseband frequency.    The reason for the LPF is to eliminate even order (second mainly) harmonics that would skew the subsequent squaring.   Even order harmonics aren't important as they will be symmetrical

 

Then pass the waveform into a mid-point biassed comparator with symmetrical hysteresis.   Line receivers such as the 26xx32 or 33 will do the job admirably, and work at frequencies up to several MHz.  Finally generate antiphase drive signals via logic inverters./ buffers

 

So - LPF +   line receiver. +  Q / NotQ drive .    Simples !

 

Andy

www.g4jnt.com

 

 

 

On 21 April 2011 21:58, Rik Strobbe <Rik.Strobbe@fys.kuleuven.be> wrote:

Hello Andy, Scott, Wolf,

 

indeed WSPR is MFSK, I should be reading mails more carefull.

Some time ago I was thinking about running WSPR with a "divide by 2" class D PA and wrote a small piece of software that created WSPR audio at double or quadruple frequency shift. Modulating a SSB TX at 2MHz with quadruple shift WSPR audio signal produced a nice WSPR signal at 500kHz after divide by 4.

Due to lack of time I haven't done any on air tests, but if anyone wants to try I can provide the software.

 

Another thought :

If one could "fool" the software about the soundcard sample rate it would be possible to get frequency doubled audio from any existing sofware.

Ie: Let the software believe that the soundcard is running at 11025Hz while it is running at 22100Hz.

 

73, Rik  ON7YD - OR7T

 

---

Van: owner-rsgb_lf_group@blacksheep.org [owner-rsgb_lf_group@blacksheep.org] namens wolf_dl4yhf [dl4yhf@freenet.de]
Verzonden: donderdag 21 april 2011 20:06
Aan: rsgb_lf_group@blacksheep.org
Onderwerp: Re: LF: MSK with a Class D transmitter...?

Hi Rik,

AFAIK WSPR is not MSK. It's 4-FSK. One of the very exotic modes in WSJT (not WSPR) was indeed MSK but I don't remember which one.

About the FET push-pull drive: I wouldn't rely on a windoze PC to produce a non-overlapping drive signal for the two FETs. Just imagine what happens to the FETs if windoze decides to play "ding-dong" (through both output channels) during transmit ;-)

Cheers,
   Wolf .

Am 21.04.2011 19:46, schrieb Rik Strobbe:

Hello Scott,

 

what kind of MSK do you want to transmit ?

WSPR by any chance ?

 

73, Rik  ON7YD - OR7T

 

On 21 April 2011 01:28, Scott Tilley <sthed475@telus.net> wrote:

Hi Andy

Holy mind bogglingness batman...

Here's the parameters to make this real simple:
- I want to generate a MSK signal out of the transmitter at 8970Hz.
- I'm using Spectrum Lab, which allows me to create a MSK drive and at a baud rate I want at any frequency my sound card can generate.
- SL allows me to output quadrature outputs.

So, If I understand correctly pump said FET driver with the quadrature outputs from the sound card audio (squaring up the sine waves of course) and it should work?

Thanks,
Scott

On 4/20/2011 2:09 PM, Andy Talbot wrote:

Bin too quick to type..

NOt thinking straight...

 

You need to the upconvert either the resulting doubled audio tone by twice the RF carrier, or upconvert first and double the final RF.  Same ideas for frequency doubling can be used.

 

See http://www.g4jnt.com/LFUpconv.pdf  for a direct from audio upconverter.

 

Must fully check before hitting SEND !

 

Andy

G4JNT

On 20 April 2011 21:50, Andy Talbot <andy.g4jnt@gmail.com> wrote:

I assume you are genrating the MSK from soundcard output, in which case two options come to mind:

 

1)  Double the frequency.  Full wave rectify the audio sinusoid, pass the AC part of the result into a comparator and into the divider.  Alternatively, as the signal bandwidth is probably quite low, square up the audio to start with, feed into one inpiut of an XOR gate.  Delay the waveform by 90 deg and feed this into the othe side of the XOR.    Output of XOR to divider as before

 

2)   Remove the  divide by two, and replace with an inverter / buffer pair to generate the out of phase drive signals.   The  input now needs to be close to a 1:1 squarwave at output frequency.   Pass the soundcard audio into a comparator and make sure it gives a 50% duty cycles.

 

 

Andy

www.g4jnt.com

 

ps.

or 3)    Convince the software author the S/W needs modifying to give a 2.F output

 

 

On 20 April 2011 21:32, Scott Tilley <sthed475@telus.net> wrote:

Hi All

I wonder if someone can explain to me how to feed an MSK signal into a typical Class D transmitter?

I mean by typical that the TX has a 4013 type /2 system feeding two FET drivers to drive FETs in a push-pull configuration.

If I understand correctly due to the fact MSK does not have amplitude component to its signal that a non-linear transmitter can be used.

So, how do you do this with a Class-D TX as often used in LF/VLF?

Thanks,
Scott
VE7TIL