Building an RF Front end for DSP FPGA Kits with ADCs

GNU Radio
1

Hello,

I am looking to buy an Altera DSP FPGA Kit (DSP Development Kit, Stratix
III
Edition) for school research but I’m not sure how i am going to build an
RF-Front End for it. I am looking to receive signals up to +30 dBm from
30
MHz - 2 GHz with bandwidths up to 50 MHz.

This kit comes with two 150 MSPS 14-bit ADCs and 250 MSPS 14-bit DACs.
My
question is it possible to buy a Commercial off the Shelf Front-End that
provides the necessary inputs?

Is there something special that I have to do in order to receive signal
levels up to +30 dBm so that my ADCs do not saturate?

My goal would be to build it myself, but if I am pressed with deadlines
I
will have to get COTS product.

View this message in context:
http://old.nabble.com/Building-an-RF-Front-end-for-DSP-FPGA-Kits-with-ADCs-tp30662773p30662773.html
Sent from the GnuRadio mailing list archive at Nabble.com.

2

Why not just buy a USRP + WBX?

2011/1/13 sirjanselot [email protected]

3

I have one already, but you can’t receive anything above - 10 dBm. Plus
I’d
like to gain FPGA programming experience.

  • Jan

On Thu, Jan 13, 2011 at 10:33 AM, Euripedes Rocha Filho <

4

On 01/13/2011 10:33 AM, Euripedes Rocha Filho wrote:

Edition) for school research but I'm not sure how i am going to
question is it possible to buy a Commercial off the Shelf
Front-End that
provides the necessary inputs?

Is there something special that I have to do in order to receive
signal
levels up to +30 dBm so that my ADCs do not saturate?

Attenuators. At +30dBm, the concern isn’t saturation, but device
destruction–it would be a shame
to destroy your $2800.00 Altera development board.

5

Thanks Marcus.

The reason why I do not want to attenuate is because I want to receive a
high-powered signal and low-powered signal at the same frequency.

If I attenuate then the low-powered signal will be reduced and if I go
beyond the noise floor, I might see it. The goal was to stretch the
difference in power between the high-powered signal and the low-power.
But
now if I think about it no matter what I will be limited to my Dynamic
Range.

Marcus D. Leech wrote:

I am looking to buy an Altera DSP FPGA Kit (DSP Development Kit,
This kit comes with two 150 MSPS 14-bit ADCs and 250 MSPS 14-bit

destruction*–it would be a shame

Discuss-gnuradio mailing list
[email protected]
Discuss-gnuradio Info Page

–
View this message in context:
http://old.nabble.com/Building-an-RF-Front-end-for-DSP-FPGA-Kits-with-ADCs-tp30662773p30666803.html
Sent from the GnuRadio mailing list archive at Nabble.com.

6

On 01/13/2011 11:23 AM, jan acosta wrote:

I have one already, but you can’t receive anything above - 10 dBm.
Plus I’d like to gain FPGA programming experience.
I’ll observe that with a 30dB attenuator in front of your receiver, you
can easily do what you want
to do. I suspect that you don’t know that much about the RF world,
otherwise the $5.00 attenuator
solution would immediately occur to you as a more economically-sound
solution than a
$2800.00 system replacement.

Further, the FPGAs on the USRPs are programmable, and the souce Verilog
is publically available
for you to modify, etc, etc.

Note that I’m not saying dont buy your Altera development board–it’s
your $2800.00, do what you
want with it, but if the reasons are just the attenuator and “I want
to program my own FPGA”, then
you already pretty-much have what you need.

–
Principal Investigator
Shirleys Bay Radio Astronomy Consortium

7

Yes, given that perfectly-acceptable signals can arrive at the front-end
at -90dBm or lower, and you want to be able
to “see” +30dBm at the same time, you’ll need roughly 120dB of
dynamic range. But +30dBm is about 7.07VRMS, or
about 10VP. There’s no way on gods little green earth that you’re
going to find an ADC that can sample at 100Msps or
more, and have a peak input voltage of +10VP (+7VRMS), and have a
120dB dynamic range. Let alone a receiver
front-end RF chain that can tolerate anything more than about +15dBm
or so.

For 120dB dynamic range, you’d need a 20-bit ADC, and it would need to
run at a fast sample rate, and it would need to be
able to handle +10VP input. Not gonna happen.

I’m curious about why you need to inject +30dBm into a receiver. You’re
getting no more information out of that signal than if
it arrived at -30dBm.

–
Marcus L.
Principal Investigator
Shirleys Bay Radio Astronomy Consortium

8

Moeller: Yes I am trying to solve the near-far problem. I have used
the
USRP radios to test the concept, and it works really well I am getting
about
55 - 60 dB suppression which is essentially equal to the practical
dynamic
range of the ADCs for the USRP1. I would insert about 1 MHz of white
Gaussian noise at about 50 dB higher than a NBFM signal on the same
frequency band. There are still some problems with it though. I
believe
for highly-nonstationary interfering signals (signals that are on and
off
constantly with a short duration in between) it might not work.

However, my solution doesn’t fix the problem where the ADC is completely
saturated, which according to wikipedia is the case for CDMA. Perhaps
as I
continue studying, at some point I will extend my project to this level.

Marcus: The USRP most likely will not have enough logic cells to handle
the
amount of programming I need to do =D. I have USRP1, and the 12-bit ADC
is
also a limitation. It would also be more expensive for me to go with
the
higher USRPs because I need two receive inputs and a transmit port all
working at the same time while as for the Altera FPGA already comes with
2
ADC inputs and 2 DACs, and it just needs an RF frontend, which would be
a
nice homebrew project. :wink:

Moeller wrote:

similar to the near far effect, the high powered signal will create
interference. Your success will then also depend on the spreading factor
(correlation lengths).

Discuss-gnuradio mailing list
[email protected]
Discuss-gnuradio Info Page

–
View this message in context:
http://old.nabble.com/Building-an-RF-Front-end-for-DSP-FPGA-Kits-with-ADCs-tp30662773p30705084.html
Sent from the GnuRadio mailing list archive at Nabble.com.

9

On 18.01.2011 23:37, JP234 wrote:

Moeller: Yes I am trying to solve the near-far problem. I have used the
USRP radios to test the concept, and it works really well I am getting about
55 - 60 dB suppression which is essentially equal to the practical dynamic
range of the ADCs for the USRP1. I would insert about 1 MHz of white
Gaussian noise at about 50 dB higher than a NBFM signal on the same
frequency band. There are still some problems with it though. I believe
for highly-nonstationary interfering signals (signals that are on and off
constantly with a short duration in between) it might not work.

The near-far problem is known for DSSS (direct sequence spread
spectrum).
I never heard of it in the context of NBFM. How can you reconstruct NBFM
signals from below the noise floor?
For on-off jammers or high-power TDMA interferers there are chances to
get the weak signal in the quite moments (gated attenuation?).

However, my solution doesn’t fix the problem where the ADC is completely
saturated, which according to wikipedia is the case for CDMA. Perhaps as I
continue studying, at some point I will extend my project to this level.

I think you should read the appropriate literature, like
Simon, Omura, Scholz “Spread Spectrum Communications Handbook”
Wikipedia is not very detailed, a very good Encyklopedia, but not a
substitute
for a text book.

Marcus: The USRP most likely will not have enough logic cells to handle the
amount of programming I need to do =D. I have USRP1, and the 12-bit ADC is
also a limitation. It would also be more expensive for me to go with the

I think it needs some other tricks than just adding a few bits of
resolution.
At first you should exactly define what you want.

higher USRPs because I need two receive inputs and a transmit port all
working at the same time while as for the Altera FPGA already comes with 2
ADC inputs and 2 DACs, and it just needs an RF frontend, which would be a
nice homebrew project. :wink:

What’s the bandwidth of the Altera ADC inputs? I guess it’s not very
wide.
But with an open design RF frontend it would be easier to modify it,
possibly with a circuit to protect it from saturation or destruction
(gated attenuation for high-power or on-off jammers).

10

On 13.01.2011 22:59, sirjanselot wrote:

The reason why I do not want to attenuate is because I want to receive a
high-powered signal and low-powered signal at the same frequency.

Can you be more specific? This is an interesting topic.
In most cases the signals cannot be separated any more.
You have certain chances if you want to seperate spread spectrum
signals or TDMA signals. For TDMA you can operate above the saturation
level, below the destruction level. For direct sequence CDMA this is
similar to the near far effect, the high powered signal will create
interference. Your success will then also depend on the spreading factor
(correlation lengths).