Transmit filtering

I’m generating two NBFM signals, offsetting them plus and minus 25 kHz
with FreqXlatingFilters, and combining the output in an adder which then
drives a USRP sink. It seems sensible to put a bandpass filter on the
result.

I have a basic conceptual confusion. I would think that you’d specify a
bandpass filter around zero with a low cutoff of, say, -40e3 and a high
cutoff of +40e3. However, this doesn’t work – the bandpass filter
seems to require a number >0 for the lowpass frequency.

So, in this situation, where I want to filter a spectrum that is
centered around zero, how do I specify the filter lowpass and highpass
frequencies?

A second question – there is a signal component on the output directly
at the USRP center frequency. It’s at most about 30dB below the desired
signals. I assume this is the equivalent of LO leakage. Is there any
way to filter this out or at least reduce its amplitude?

As I’m writing this, I wonder if the solution to both problems is to put
both my signals on one side of the USRP center frequency, say +25 and
+75 kHz. That would allow a sensible bandpass filter that would remove
the zero-frequency component. But the downside of that seems to be
throwing away half the available bandwidth.

Thanks,

John

On Mon, Sep 05, 2011 at 10:19:54AM -0400, John Ackermann N8UR wrote:

I would think that you’d specify a bandpass filter around zero with a
low cutoff of, say, -40e3 and a high cutoff of +40e3. However, this
doesn’t work – the bandpass filter seems to require a number >0 for
the lowpass frequency.

That would be a low-pass filter with a cutoff at 40e3. With
complex-valued signals, a low-pass filter is essentially a band-pass
filter centered around 0 Hz.

there is a signal component on the output directly at the USRP center
frequency. It’s at most about 30dB below the desired signals. I
assume this is the equivalent of LO leakage. Is there any way to
filter this out or at least reduce its amplitude?

As I’m writing this, I wonder if the solution to both problems is to
put both my signals on one side of the USRP center frequency, say +25
and +75 kHz.

Yes, you could use an external analog band-pass filter if that center
spur is a problem for you.

On 05/09/11 10:19 AM, John Ackermann N8UR wrote:

I’m generating two NBFM signals, offsetting them plus and minus 25 kHz
with FreqXlatingFilters, and combining the output in an adder which
then drives a USRP sink. It seems sensible to put a bandpass filter
on the result.

I have a basic conceptual confusion. I would think that you’d specify
a bandpass filter around zero with a low cutoff of, say, -40e3 and a
high cutoff of +40e3. However, this doesn’t work – the bandpass
filter seems to require a number >0 for the lowpass frequency.
Make sure that you specify complex-taps. I assume you’re using GRC. By
default, it uses the version of
the bandpass filter with real taps. But if you specify complex taps,
you can use negative frequencies.


Principal Investigator
Shirleys Bay Radio Astronomy Consortium

On Mon, Sep 5, 2011 at 12:55 PM, John Ackermann N8UR [email protected]
wrote:

Hi Marcus –

I am selecting complex taps from the drop down, but still get an error at
runtime if either cutoff frequency is 0 or smaller.

Thanks,

John

John,

That sounds like a bug in the GRC block, then. I just tried this as a
complex BPF and it works fine:

gr.firdes.complex_band_pass(1, 1, -0.1, 0.2, 0.1)

However, as was already mentioned, if you are trying to create a
bandpass
filter from -40 to +40 kHz, that’s just a low pass filter with a
bandwidth
of 40 kHz:

gr.firdes.low_pass(1, 100e3, 40e3, 2e3)

Will give you a filter for this purpose. Substitute your own sampling
rate
and transition width, of course.

Tom

Hi Marcus –

I am selecting complex taps from the drop down, but still get an error
at runtime if either cutoff frequency is 0 or smaller.

Thanks,

John

On 05/09/11 12:55 PM, John Ackermann N8UR wrote:

Hi Marcus –

I am selecting complex taps from the drop down, but still get an error at
runtime if either cutoff frequency is 0 or smaller.

Thanks,

John

I just tried it, and it works just fine.

I select: Complex->Complex (complex taps) from the menu, with a complex
I/O, and I’m able to specify
negative frequency components.

What is the error you’re getting at runtime?


Principal Investigator
Shirleys Bay Radio Astronomy Consortium

On 09/05/2011 01:10 PM, Marcus D. Leech wrote:

I just tried it, and it works just fine.

I select: Complex->Complex (complex taps) from the menu, with a complex
I/O, and I’m able to specify
negative frequency components.

What is the error you’re getting at runtime?

Using a band reject filter with low frequency set to -10000 and high
frequency set to 10000, dropdown of complex(decimating), I get this:

Traceback (most recent call last):
File “/home/jra/gnuradio/fm_rx.py”, line 166, in
tb = fm_rx()
File “/home/jra/gnuradio/fm_rx.py”, line 115, in init
1, 200000, -10000, 10000, 1000, firdes.WIN_HAMMING, 6.76))
File
“/usr/local/lib/python2.7/dist-packages/gnuradio/gr/gnuradio_core_general.py”,
line 3437, in band_reject
return _gnuradio_core_general.firdes_band_reject(*args, **kwargs)
IndexError: gr_firdes check failed: 0 < fa <= sampling_freq / 2

Hi…I have an SDRPlay and we have this religious AM station here in DC/MD on 1120kHz and it blocks out alost every signal below 30 MHz. There is also this religious station out of SC where the guy sounds like a troll and just spews right wing propaganda, very annoying to listen to.
I found two sites that sell traps but when you call they are out of business; clifton labs and some ham radio guy.
I have tried everything else but the better the antenna the worse this gets. Even overloads my portable radios. I would make one but cant find the inductors/ dont have the equipment to make one.