Hello,

I am trying to implement an SSB phasing receiver using the GNU Radio
toolkit [1]. I do not have an RF frontend yet, so I had to depend on
those fortunate folks who had the required hardware to get some real
world samples. They samples are captured as given below:

50.25 mhz
cos
---- x ------- I
|
|
fc=50.30Mhz ADC -
|
|
---- x ------- Q
sin
50.25 mhz

The LSB signal which was transmitted at 50.30 mhz is digitally
downconverted with an IF of 50.25mhz. The complex samples I have is
centered at 50 khz. Now
I do exactly as described in the KK7B article on phasing receiver[2].
The I and Q signals are downconverted to baseband, and lowpass
filtered. The I channel is delayed and Q channel is passed thru hilbert
filter.

I get the audio output, when the I and Q are subtracted. Fine. But I
get audio even when they are summed together!!! i.e There is no
sideband suppression. I am wondering what the problem is. I looked at
the hilbert transformer freq response. It looks okay (allpass, with
nulls at zero and fs/2), could this null at zero cause a problem,
because the input is already at baseband?

Any advise/help will be greatly appreciated.

Thanks
Ramakrishnan

[1] htp://www.gnu.org/software/gnuradio/
[2] http://www.arrl.org/tis/info/pdf/t9301032.pdf

But I get audio even when
they are summed together!!! i.e There is no sideband suppression.

There is no sideband suppression with I/Q demodulation until you get
your carrier synchronized to the transmitter's carrier. If there's no
mechanism for doing this, you don't get sideband suppresion through I/Q
demodulation. (Well, you may experience rapid fading if you're within
a few Hz, but I don't think you want that!)

There are some homebrewers who do not-completely-suppressed carrier
with synchronous detection. It's sort of SSB but not traditional ham
radio SSB. In traditional ham radio SSB you try to suppress as much of
the carrier as you can, and this won't help you synchronously
demodulate it.

Without that synchronization, just use a bandpass (IF+300 Hz to IF+3000
Hz for example for USB).

Tim.

"Tim Shoppa" ) writes:
But I get audio even when
they are summed together!!! i.e There is no sideband suppression.

There is no sideband suppression with I/Q demodulation until you get
your carrier synchronized to the transmitter's carrier. If there's no
mechanism for doing this, you don't get sideband suppresion through I/Q
demodulation. (Well, you may experience rapid fading if you're within
a few Hz, but I don't think you want that!)

There are some homebrewers who do not-completely-suppressed carrier
with synchronous detection. It's sort of SSB but not traditional ham
radio SSB. In traditional ham radio SSB you try to suppress as much of
the carrier as you can, and this won't help you synchronously
demodulate it.

Without that synchronization, just use a bandpass (IF+300 Hz to IF+3000
Hz for example for USB).

Tim.


But he's not talking about synchronized demodulation, he's talking about
SSB demodulation.

Phasing demodulation was around before "synchronous detection" was described
in CQ in the late fifties. Where the carrier comes from is irrelevant, it's
about the proper combination of phase shifts to select which sideband
is being received. Add the outputs, and you get one sideband, subtract
them and you get the other.

Synchronous detection merely adds a means of locking the locally generated
"carrier" to the incoming signal. And of course, one can have sync
detection without selectable sideband.

But he's talking about SSB, and there's neither a carrier, nor two
sidebands to sync to.

But if he is doing this right, he should not receive the incoming sideband
if he has the combinations set up for receiving the other sideband.

I have no idea what's wrong, but this has nothing to do with synchronous
detection.

Michael VE2BVW

On 23 Feb 2005 07:37:41 -0800, "Tim Shoppa"
wrote:

There are some homebrewers who do not-completely-suppressed carrier
with synchronous detection. It's sort of SSB but not traditional ham
radio SSB. In traditional ham radio SSB you try to suppress as much of
the carrier as you can, and this won't help you synchronously
demodulate it.

I got the impression he was using an I-Q system to implment a hilbert
transform. The result is an image suppressing system that can
demodulate SSB with a baseband IF.

Essentually what a KK7B R2/miniR2/R2pro(RX) and the companion T2
(transmit) using aboth a 90degree I-Q RF path and an +45 and -45
degree audio processing path.


Allison

On 23 Feb 2005 02:27:53 -0800, wrote:

The LSB signal which was transmitted at 50.30 mhz is digitally
downconverted with an IF of 50.25mhz. The complex samples I have is
centered at 50 khz. Now

As someone running a 6m analog phasing rig what your describing makes
little sense to me.

Will not work unless you down covert again to baseband.

The signal you should be looking at is SSB rf source at 50.250 with
a 90 degree I-Q local osc at 50.250 and an audio baseband. Then
sample at greater than 20Ks/S with two channels and process them by
placeing phase delay of 90 degrees on one path and the and summing
them.

I do exactly as described in the KK7B article on phasing receiver[2].
The I and Q signals are downconverted to baseband, and lowpass
filtered. The I channel is delayed and Q channel is passed thru hilbert
filter.

No it is not.

I get the audio output, when the I and Q are subtracted. Fine. But I
get audio even when they are summed together!!! i.e There is no
sideband suppression. I am wondering what the problem is. I looked at
the hilbert transformer freq response. It looks okay (allpass, with
nulls at zero and fs/2), could this null at zero cause a problem,
because the input is already at baseband?

Unless your considering baseband signal desired to be the 50khz signal
then what you've done is created an image reject mixer with a 50khz
output and not an SSB detector. The image reject mixer case means
your signal is in the output passband at LO+ 50khz but not at
LO-50khz. You'd still have to repeat the process again to get from
50khz to recovered audio with a 50khz digitally derived LO and mixing
process. You can do that. In the end your doing something that
needs only to be done once. Look at EMDRF [Experimental Methods in RF
design] doing SSB in DSP is coverd there.


Allison

wrote:

Unless your considering baseband signal desired to be the 50khz
signal
then what you've done is created an image reject mixer with a 50khz
output and not an SSB detector. The image reject mixer case means
your signal is in the output passband at LO+ 50khz but not at
LO-50khz. You'd still have to repeat the process again to get from
50khz to recovered audio with a 50khz digitally derived LO and mixing
process. You can do that. In the end your doing something that
needs only to be done once. Look at EMDRF [Experimental Methods in
RF
design] doing SSB in DSP is coverd there.

Thanks for the ideas. I will try to acquire EMFRD and look for more
details.

I am not having an RF frontend, the samples I have are in a file
captured by someone else and they are centered around 50 khz. I am
doing the mixing again to baseband, so yes, I am unnecessarily doing it
twice instead of once, but since I have no access to hardware now, I
have no other option. I will verify the waveforms at various points
again and get back.

Thanks again for the helpful comments. It was a bit confusing for me
initially to work with the complex digitally downconverted samples. I
think a bit of confusion still persists. I will go through the whole
thing again.

Thanks
Ramakrishnan

Look at the data at

sdradio.org

73, Dave, N3HE

"rkrishnan" wrote in message
ps.com...
wrote:

Unless your considering baseband signal desired to be the 50khz
signal
then what you've done is created an image reject mixer with a 50khz
SNIP