Empathy List Archives

DC

Dan CaJacob

Thu, Feb 1, 2018 3:46 PM

Hey guys,

I have put together a noise figure meter application that uses a USRP as
the sensing device. It started off as a way to measure the NF of the USRP
itself. I have a calibrated noise source from an HP 8970B Noise Figure
Meter. To test the NF of the USRP, I connect the head to the USRP input. My
GNURadio flowgraph maximizes the USRP gain and measures a moving average of
the received power while I switch the noise source on and off. The
difference in the received power level, in addition to the ENR table from
the noise source, can then be used to calculate the NF of the USRP itself
using the y-factor method.

Once you have the NF for the USRP at many frequencies (I test every 50 MHz
from 50 MHz - 6000 MHz), you can modify the same procedure to test the NF
of a Device Under Test (DUT) which is connected between the noise source
and the (now calibrated) USRP. You can use the USRP cal table we generated
in the previous step to derive the NF of the DUT corrected for the NF of
the USRP.

In short, this all works incredibly well and garners very repeatable
results. One complication is that you will see wild NF at certain
frequencies due to local interference like LTE and WIFI. I've also compared
the results to that which the HP device measures and they're very
comparable. ... Except below ~ 1GHz.

And here's the issue - I am seeing higher NF for DUTs below about 1GHz and
particularly worse below 500 MHz. I was hoping someone at Ettus might be
able to shed some light on why this might be. Curiously, the USRPs NF
doesn't seem to be too bad, just the DUT.

I'll note that I am nominally using a B200 for these tests, but I also
tried an E312 just in case the filter banks might help out somehow. I
didn't see a difference - they both had the same problem.

I have used several DUTs for this test, including LNA boards we have
designed ourselves and a Mini-Circuits ZX60-P103LN+ (
https://www.minicircuits.com/pdfs/ZX60-P103LN+.pdf). Both seem to exhibit
higher NF when measured with a USRP below 1 GHz. When testing them on the
HP NF meter, the NF is as expected all the way down to 50 MHz.

I have attached the B200 cal data for your enjoyment as well as the
B200-measured ZX60 NF and the HP-measured ZX60. The HP NF meter only goes
up to 1600 MHz, which is why that data file stops there. I was surprised to
see the B200 seemed to have a better NF than the E312, which averaged 8 dB
NF, by the way.

Very Respectfully,

Dan CaJacob

Hey guys, I have put together a noise figure meter application that uses a USRP as the sensing device. It started off as a way to measure the NF of the USRP itself. I have a calibrated noise source from an HP 8970B Noise Figure Meter. To test the NF of the USRP, I connect the head to the USRP input. My GNURadio flowgraph maximizes the USRP gain and measures a moving average of the received power while I switch the noise source on and off. The difference in the received power level, in addition to the ENR table from the noise source, can then be used to calculate the NF of the USRP itself using the y-factor method. Once you have the NF for the USRP at many frequencies (I test every 50 MHz from 50 MHz - 6000 MHz), you can modify the same procedure to test the NF of a Device Under Test (DUT) which is connected between the noise source and the (now calibrated) USRP. You can use the USRP cal table we generated in the previous step to derive the NF of the DUT corrected for the NF of the USRP. In short, this all works incredibly well and garners very repeatable results. One complication is that you will see wild NF at certain frequencies due to local interference like LTE and WIFI. I've also compared the results to that which the HP device measures and they're very comparable. ... Except below ~ 1GHz. And here's the issue - I am seeing higher NF for DUTs below about 1GHz and particularly worse below 500 MHz. I was hoping someone at Ettus might be able to shed some light on why this might be. Curiously, the USRPs NF doesn't seem to be too bad, just the DUT. I'll note that I am nominally using a B200 for these tests, but I also tried an E312 just in case the filter banks might help out somehow. I didn't see a difference - they both had the same problem. I have used several DUTs for this test, including LNA boards we have designed ourselves and a Mini-Circuits ZX60-P103LN+ ( https://www.minicircuits.com/pdfs/ZX60-P103LN+.pdf). Both seem to exhibit higher NF when measured with a USRP below 1 GHz. When testing them on the HP NF meter, the NF is as expected all the way down to 50 MHz. I have attached the B200 cal data for your enjoyment as well as the B200-measured ZX60 NF and the HP-measured ZX60. The HP NF meter only goes up to 1600 MHz, which is why that data file stops there. I was surprised to see the B200 seemed to have a better NF than the E312, which averaged 8 dB NF, by the way. -- Very Respectfully, Dan CaJacob

RC

Robin Coxe

Thu, Feb 1, 2018 10:14 PM

Hi Dan. Both the B200 and the E312 use the Analog Devices AD9361 RF
integrated transceiver. This chip does have an integrated LNA. Perhaps
there's some sort of mismatch between your DUTs and this integrated LNA at
<1 GHz?

ADI publishes the RX S-parameters:
https://ez.analog.com/thread/41208#137929

-Robin

On Thu, Feb 1, 2018 at 7:46 AM, Dan CaJacob via USRP-users <
usrp-users@lists.ettus.com> wrote:

Hey guys,

I have put together a noise figure meter application that uses a USRP as
the sensing device. It started off as a way to measure the NF of the USRP
itself. I have a calibrated noise source from an HP 8970B Noise Figure
Meter. To test the NF of the USRP, I connect the head to the USRP input. My
GNURadio flowgraph maximizes the USRP gain and measures a moving average of
the received power while I switch the noise source on and off. The
difference in the received power level, in addition to the ENR table from
the noise source, can then be used to calculate the NF of the USRP itself
using the y-factor method.

Once you have the NF for the USRP at many frequencies (I test every 50 MHz
from 50 MHz - 6000 MHz), you can modify the same procedure to test the NF
of a Device Under Test (DUT) which is connected between the noise source
and the (now calibrated) USRP. You can use the USRP cal table we generated
in the previous step to derive the NF of the DUT corrected for the NF of
the USRP.

In short, this all works incredibly well and garners very repeatable
results. One complication is that you will see wild NF at certain
frequencies due to local interference like LTE and WIFI. I've also compared
the results to that which the HP device measures and they're very
comparable. ... Except below ~ 1GHz.

And here's the issue - I am seeing higher NF for DUTs below about 1GHz and
particularly worse below 500 MHz. I was hoping someone at Ettus might be
able to shed some light on why this might be. Curiously, the USRPs NF
doesn't seem to be too bad, just the DUT.

I'll note that I am nominally using a B200 for these tests, but I also
tried an E312 just in case the filter banks might help out somehow. I
didn't see a difference - they both had the same problem.

I have used several DUTs for this test, including LNA boards we have
designed ourselves and a Mini-Circuits ZX60-P103LN+ (
https://www.minicircuits.com/pdfs/ZX60-P103LN+.pdf). Both seem to exhibit
higher NF when measured with a USRP below 1 GHz. When testing them on the
HP NF meter, the NF is as expected all the way down to 50 MHz.

I have attached the B200 cal data for your enjoyment as well as the
B200-measured ZX60 NF and the HP-measured ZX60. The HP NF meter only goes
up to 1600 MHz, which is why that data file stops there. I was surprised to
see the B200 seemed to have a better NF than the E312, which averaged 8 dB
NF, by the way.

Very Respectfully,

Dan CaJacob

USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

Hi Dan. Both the B200 and the E312 use the Analog Devices AD9361 RF integrated transceiver. This chip does have an integrated LNA. Perhaps there's some sort of mismatch between your DUTs and this integrated LNA at <1 GHz? ADI publishes the RX S-parameters: https://ez.analog.com/thread/41208#137929 -Robin On Thu, Feb 1, 2018 at 7:46 AM, Dan CaJacob via USRP-users < usrp-users@lists.ettus.com> wrote: > Hey guys, > > I have put together a noise figure meter application that uses a USRP as > the sensing device. It started off as a way to measure the NF of the USRP > itself. I have a calibrated noise source from an HP 8970B Noise Figure > Meter. To test the NF of the USRP, I connect the head to the USRP input. My > GNURadio flowgraph maximizes the USRP gain and measures a moving average of > the received power while I switch the noise source on and off. The > difference in the received power level, in addition to the ENR table from > the noise source, can then be used to calculate the NF of the USRP itself > using the y-factor method. > > Once you have the NF for the USRP at many frequencies (I test every 50 MHz > from 50 MHz - 6000 MHz), you can modify the same procedure to test the NF > of a Device Under Test (DUT) which is connected between the noise source > and the (now calibrated) USRP. You can use the USRP cal table we generated > in the previous step to derive the NF of the DUT corrected for the NF of > the USRP. > > In short, this all works incredibly well and garners very repeatable > results. One complication is that you will see wild NF at certain > frequencies due to local interference like LTE and WIFI. I've also compared > the results to that which the HP device measures and they're very > comparable. ... Except below ~ 1GHz. > > And here's the issue - I am seeing higher NF for DUTs below about 1GHz and > particularly worse below 500 MHz. I was hoping someone at Ettus might be > able to shed some light on why this might be. Curiously, the USRPs NF > doesn't seem to be too bad, just the DUT. > > I'll note that I am nominally using a B200 for these tests, but I also > tried an E312 just in case the filter banks might help out somehow. I > didn't see a difference - they both had the same problem. > > I have used several DUTs for this test, including LNA boards we have > designed ourselves and a Mini-Circuits ZX60-P103LN+ ( > https://www.minicircuits.com/pdfs/ZX60-P103LN+.pdf). Both seem to exhibit > higher NF when measured with a USRP below 1 GHz. When testing them on the > HP NF meter, the NF is as expected all the way down to 50 MHz. > > I have attached the B200 cal data for your enjoyment as well as the > B200-measured ZX60 NF and the HP-measured ZX60. The HP NF meter only goes > up to 1600 MHz, which is why that data file stops there. I was surprised to > see the B200 seemed to have a better NF than the E312, which averaged 8 dB > NF, by the way. > -- > Very Respectfully, > > Dan CaJacob > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > >

DC

Dan CaJacob

Fri, Feb 2, 2018 1:39 AM

That's an interesting thought. The 9361 does have a pretty bad match. I'll
try adding a 50 Ohm attenuator and see if that helps.

On Thu, Feb 1, 2018 at 5:14 PM Robin Coxe robin.coxe@ettus.com wrote:

Hi Dan. Both the B200 and the E312 use the Analog Devices AD9361 RF
integrated transceiver. This chip does have an integrated LNA. Perhaps
there's some sort of mismatch between your DUTs and this integrated LNA at
<1 GHz?

ADI publishes the RX S-parameters:
https://ez.analog.com/thread/41208#137929

-Robin

On Thu, Feb 1, 2018 at 7:46 AM, Dan CaJacob via USRP-users <
usrp-users@lists.ettus.com> wrote:

Hey guys,

I have put together a noise figure meter application that uses a USRP as
the sensing device. It started off as a way to measure the NF of the USRP
itself. I have a calibrated noise source from an HP 8970B Noise Figure
Meter. To test the NF of the USRP, I connect the head to the USRP input. My
GNURadio flowgraph maximizes the USRP gain and measures a moving average of
the received power while I switch the noise source on and off. The
difference in the received power level, in addition to the ENR table from
the noise source, can then be used to calculate the NF of the USRP itself
using the y-factor method.

Once you have the NF for the USRP at many frequencies (I test every 50
MHz from 50 MHz - 6000 MHz), you can modify the same procedure to test the
NF of a Device Under Test (DUT) which is connected between the noise source
and the (now calibrated) USRP. You can use the USRP cal table we generated
in the previous step to derive the NF of the DUT corrected for the NF of
the USRP.

In short, this all works incredibly well and garners very repeatable
results. One complication is that you will see wild NF at certain
frequencies due to local interference like LTE and WIFI. I've also compared
the results to that which the HP device measures and they're very
comparable. ... Except below ~ 1GHz.

And here's the issue - I am seeing higher NF for DUTs below about 1GHz
and particularly worse below 500 MHz. I was hoping someone at Ettus might
be able to shed some light on why this might be. Curiously, the USRPs NF
doesn't seem to be too bad, just the DUT.

I'll note that I am nominally using a B200 for these tests, but I also
tried an E312 just in case the filter banks might help out somehow. I
didn't see a difference - they both had the same problem.

I have used several DUTs for this test, including LNA boards we have
designed ourselves and a Mini-Circuits ZX60-P103LN+ (
https://www.minicircuits.com/pdfs/ZX60-P103LN+.pdf). Both seem to
exhibit higher NF when measured with a USRP below 1 GHz. When testing them
on the HP NF meter, the NF is as expected all the way down to 50 MHz.

I have attached the B200 cal data for your enjoyment as well as the
B200-measured ZX60 NF and the HP-measured ZX60. The HP NF meter only goes
up to 1600 MHz, which is why that data file stops there. I was surprised to
see the B200 seemed to have a better NF than the E312, which averaged 8 dB
NF, by the way.

Very Respectfully,

Dan CaJacob

USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

--

Very Respectfully,

Dan CaJacob

That's an interesting thought. The 9361 does have a pretty bad match. I'll try adding a 50 Ohm attenuator and see if that helps. On Thu, Feb 1, 2018 at 5:14 PM Robin Coxe <robin.coxe@ettus.com> wrote: > Hi Dan. Both the B200 and the E312 use the Analog Devices AD9361 RF > integrated transceiver. This chip does have an integrated LNA. Perhaps > there's some sort of mismatch between your DUTs and this integrated LNA at > <1 GHz? > > ADI publishes the RX S-parameters: > https://ez.analog.com/thread/41208#137929 > > -Robin > > On Thu, Feb 1, 2018 at 7:46 AM, Dan CaJacob via USRP-users < > usrp-users@lists.ettus.com> wrote: > >> Hey guys, >> >> I have put together a noise figure meter application that uses a USRP as >> the sensing device. It started off as a way to measure the NF of the USRP >> itself. I have a calibrated noise source from an HP 8970B Noise Figure >> Meter. To test the NF of the USRP, I connect the head to the USRP input. My >> GNURadio flowgraph maximizes the USRP gain and measures a moving average of >> the received power while I switch the noise source on and off. The >> difference in the received power level, in addition to the ENR table from >> the noise source, can then be used to calculate the NF of the USRP itself >> using the y-factor method. >> >> Once you have the NF for the USRP at many frequencies (I test every 50 >> MHz from 50 MHz - 6000 MHz), you can modify the same procedure to test the >> NF of a Device Under Test (DUT) which is connected between the noise source >> and the (now calibrated) USRP. You can use the USRP cal table we generated >> in the previous step to derive the NF of the DUT corrected for the NF of >> the USRP. >> >> In short, this all works incredibly well and garners very repeatable >> results. One complication is that you will see wild NF at certain >> frequencies due to local interference like LTE and WIFI. I've also compared >> the results to that which the HP device measures and they're very >> comparable. ... Except below ~ 1GHz. >> >> And here's the issue - I am seeing higher NF for DUTs below about 1GHz >> and particularly worse below 500 MHz. I was hoping someone at Ettus might >> be able to shed some light on why this might be. Curiously, the USRPs NF >> doesn't seem to be too bad, just the DUT. >> >> I'll note that I am nominally using a B200 for these tests, but I also >> tried an E312 just in case the filter banks might help out somehow. I >> didn't see a difference - they both had the same problem. >> >> I have used several DUTs for this test, including LNA boards we have >> designed ourselves and a Mini-Circuits ZX60-P103LN+ ( >> https://www.minicircuits.com/pdfs/ZX60-P103LN+.pdf). Both seem to >> exhibit higher NF when measured with a USRP below 1 GHz. When testing them >> on the HP NF meter, the NF is as expected all the way down to 50 MHz. >> >> I have attached the B200 cal data for your enjoyment as well as the >> B200-measured ZX60 NF and the HP-measured ZX60. The HP NF meter only goes >> up to 1600 MHz, which is why that data file stops there. I was surprised to >> see the B200 seemed to have a better NF than the E312, which averaged 8 dB >> NF, by the way. >> -- >> Very Respectfully, >> >> Dan CaJacob >> >> _______________________________________________ >> USRP-users mailing list >> USRP-users@lists.ettus.com >> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> >> > -- Very Respectfully, Dan CaJacob

RE

Ron Economos

Fri, Feb 2, 2018 1:45 AM

There's also a balun on the AD9361 input. Unfortunately, the balun part
number for the low frequency path is not on the schematic.

Ron

On 02/01/2018 05:39 PM, Dan CaJacob via USRP-users wrote:

That's an interesting thought. The 9361 does have a pretty bad match.
I'll try adding a 50 Ohm attenuator and see if that helps.

On Thu, Feb 1, 2018 at 5:14 PM Robin Coxe <robin.coxe@ettus.com
mailto:robin.coxe@ettus.com> wrote:

 Hi Dan.   Both the B200 and the E312 use the Analog Devices AD9361
 RF integrated transceiver. This chip does have an integrated LNA. 
  Perhaps there's some sort of mismatch between your DUTs and this
 integrated LNA at <1 GHz?

 ADI publishes the RX S-parameters:
 https://ez.analog.com/thread/41208#137929

 -Robin

 On Thu, Feb 1, 2018 at 7:46 AM, Dan CaJacob via USRP-users
 <usrp-users@lists.ettus.com <mailto:usrp-users@lists.ettus.com>>
 wrote:

     Hey guys,

     I have put together a noise figure meter application that uses
     a USRP as the sensing device. It started off as a way to
     measure the NF of the USRP itself. I have a calibrated noise
     source from an HP 8970B Noise Figure Meter. To test the NF of
     the USRP, I connect the head to the USRP input. My GNURadio
     flowgraph maximizes the USRP gain and measures a moving
     average of the received power while I switch the noise source
     on and off. The difference in the received power level, in
     addition to the ENR table from the noise source, can then be
     used to calculate the NF of the USRP itself using the y-factor
     method.

     Once you have the NF for the USRP at many frequencies (I test
     every 50 MHz from 50 MHz - 6000 MHz), you can modify the same
     procedure to test the NF of a Device Under Test (DUT) which is
     connected between the noise source and the (now calibrated)
     USRP. You can use the USRP cal table we generated in the
     previous step to derive the NF of the DUT corrected for the NF
     of the USRP.

     In short, this all works incredibly well and garners very
     repeatable results. One complication is that you will see wild
     NF at certain frequencies due to local interference like LTE
     and WIFI. I've also compared the results to that which the HP
     device measures and they're very comparable. ... Except below
     ~ 1GHz.

     And here's the issue - I am seeing higher NF for DUTs below
     about 1GHz and particularly worse below 500 MHz. I was hoping
     someone at Ettus might be able to shed some light on why this
     might be. Curiously, the USRPs NF doesn't seem to be too bad,
     just the DUT.

     I'll note that I am nominally using a B200 for these tests,
     but I also tried an E312 just in case the filter banks might
     help out somehow. I didn't see a difference - they both had
     the same problem.

     I have used several DUTs for this test, including LNA boards
     we have designed ourselves and a Mini-Circuits ZX60-P103LN+
     (https://www.minicircuits.com/pdfs/ZX60-P103LN+.pdf). Both
     seem to exhibit higher NF when measured with a USRP below 1
     GHz. When testing them on the HP NF meter, the NF is as
     expected all the way down to 50 MHz.

     I have attached the B200 cal data for your enjoyment as well
     as the B200-measured ZX60 NF and the HP-measured ZX60. The HP
     NF meter only goes up to 1600 MHz, which is why that data file
     stops there. I was surprised to see the B200 seemed to have a
     better NF than the E312, which averaged 8 dB NF, by the way.
     -- 
     Very Respectfully,

     Dan CaJacob

     _______________________________________________
     USRP-users mailing list
     USRP-users@lists.ettus.com <mailto:USRP-users@lists.ettus.com>
     http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

--
Very Respectfully,

Dan CaJacob

USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

There's also a balun on the AD9361 input. Unfortunately, the balun part number for the low frequency path is not on the schematic. Ron On 02/01/2018 05:39 PM, Dan CaJacob via USRP-users wrote: > That's an interesting thought. The 9361 does have a pretty bad match. > I'll try adding a 50 Ohm attenuator and see if that helps. > > On Thu, Feb 1, 2018 at 5:14 PM Robin Coxe <robin.coxe@ettus.com > <mailto:robin.coxe@ettus.com>> wrote: > > Hi Dan. Both the B200 and the E312 use the Analog Devices AD9361 > RF integrated transceiver. This chip does have an integrated LNA. > Perhaps there's some sort of mismatch between your DUTs and this > integrated LNA at <1 GHz? > > ADI publishes the RX S-parameters: > https://ez.analog.com/thread/41208#137929 > > -Robin > > On Thu, Feb 1, 2018 at 7:46 AM, Dan CaJacob via USRP-users > <usrp-users@lists.ettus.com <mailto:usrp-users@lists.ettus.com>> > wrote: > > Hey guys, > > I have put together a noise figure meter application that uses > a USRP as the sensing device. It started off as a way to > measure the NF of the USRP itself. I have a calibrated noise > source from an HP 8970B Noise Figure Meter. To test the NF of > the USRP, I connect the head to the USRP input. My GNURadio > flowgraph maximizes the USRP gain and measures a moving > average of the received power while I switch the noise source > on and off. The difference in the received power level, in > addition to the ENR table from the noise source, can then be > used to calculate the NF of the USRP itself using the y-factor > method. > > Once you have the NF for the USRP at many frequencies (I test > every 50 MHz from 50 MHz - 6000 MHz), you can modify the same > procedure to test the NF of a Device Under Test (DUT) which is > connected between the noise source and the (now calibrated) > USRP. You can use the USRP cal table we generated in the > previous step to derive the NF of the DUT corrected for the NF > of the USRP. > > In short, this all works incredibly well and garners very > repeatable results. One complication is that you will see wild > NF at certain frequencies due to local interference like LTE > and WIFI. I've also compared the results to that which the HP > device measures and they're very comparable. ... Except below > ~ 1GHz. > > And here's the issue - I am seeing higher NF for DUTs below > about 1GHz and particularly worse below 500 MHz. I was hoping > someone at Ettus might be able to shed some light on why this > might be. Curiously, the USRPs NF doesn't seem to be too bad, > just the DUT. > > I'll note that I am nominally using a B200 for these tests, > but I also tried an E312 just in case the filter banks might > help out somehow. I didn't see a difference - they both had > the same problem. > > I have used several DUTs for this test, including LNA boards > we have designed ourselves and a Mini-Circuits ZX60-P103LN+ > (https://www.minicircuits.com/pdfs/ZX60-P103LN+.pdf). Both > seem to exhibit higher NF when measured with a USRP below 1 > GHz. When testing them on the HP NF meter, the NF is as > expected all the way down to 50 MHz. > > I have attached the B200 cal data for your enjoyment as well > as the B200-measured ZX60 NF and the HP-measured ZX60. The HP > NF meter only goes up to 1600 MHz, which is why that data file > stops there. I was surprised to see the B200 seemed to have a > better NF than the E312, which averaged 8 dB NF, by the way. > -- > Very Respectfully, > > Dan CaJacob > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com <mailto:USRP-users@lists.ettus.com> > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > > > -- > Very Respectfully, > > Dan CaJacob > > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

RC

Robin Coxe

Fri, Feb 2, 2018 2:05 AM

On p.8 of B200 schematic:
T801 is Macom ETC1-1-13TR (RF2)
T800 is Minicircuits TC1-1-43A+ (RF3)
U802 is Anaren BD3150L50100AHF (RF1)

On Thu, Feb 1, 2018 at 5:45 PM, Ron Economos via USRP-users <
usrp-users@lists.ettus.com> wrote:

There's also a balun on the AD9361 input. Unfortunately, the balun part
number for the low frequency path is not on the schematic.

Ron
On 02/01/2018 05:39 PM, Dan CaJacob via USRP-users wrote:

That's an interesting thought. The 9361 does have a pretty bad match. I'll
try adding a 50 Ohm attenuator and see if that helps.

On Thu, Feb 1, 2018 at 5:14 PM Robin Coxe robin.coxe@ettus.com wrote:

Hi Dan. Both the B200 and the E312 use the Analog Devices AD9361 RF
integrated transceiver. This chip does have an integrated LNA. Perhaps
there's some sort of mismatch between your DUTs and this integrated LNA at
<1 GHz?

ADI publishes the RX S-parameters: https://ez.
analog.com/thread/41208#137929

-Robin

On Thu, Feb 1, 2018 at 7:46 AM, Dan CaJacob via USRP-users <
usrp-users@lists.ettus.com> wrote:

Hey guys,

I have put together a noise figure meter application that uses a USRP as
the sensing device. It started off as a way to measure the NF of the USRP
itself. I have a calibrated noise source from an HP 8970B Noise Figure
Meter. To test the NF of the USRP, I connect the head to the USRP input. My
GNURadio flowgraph maximizes the USRP gain and measures a moving average of
the received power while I switch the noise source on and off. The
difference in the received power level, in addition to the ENR table from
the noise source, can then be used to calculate the NF of the USRP itself
using the y-factor method.

Once you have the NF for the USRP at many frequencies (I test every 50
MHz from 50 MHz - 6000 MHz), you can modify the same procedure to test the
NF of a Device Under Test (DUT) which is connected between the noise source
and the (now calibrated) USRP. You can use the USRP cal table we generated
in the previous step to derive the NF of the DUT corrected for the NF of
the USRP.

In short, this all works incredibly well and garners very repeatable
results. One complication is that you will see wild NF at certain
frequencies due to local interference like LTE and WIFI. I've also compared
the results to that which the HP device measures and they're very
comparable. ... Except below ~ 1GHz.

And here's the issue - I am seeing higher NF for DUTs below about 1GHz
and particularly worse below 500 MHz. I was hoping someone at Ettus might
be able to shed some light on why this might be. Curiously, the USRPs NF
doesn't seem to be too bad, just the DUT.

I'll note that I am nominally using a B200 for these tests, but I also
tried an E312 just in case the filter banks might help out somehow. I
didn't see a difference - they both had the same problem.

I have used several DUTs for this test, including LNA boards we have
designed ourselves and a Mini-Circuits ZX60-P103LN+ (
https://www.minicircuits.com/pdfs/ZX60-P103LN+.pdf). Both seem to
exhibit higher NF when measured with a USRP below 1 GHz. When testing them
on the HP NF meter, the NF is as expected all the way down to 50 MHz.

I have attached the B200 cal data for your enjoyment as well as the
B200-measured ZX60 NF and the HP-measured ZX60. The HP NF meter only goes
up to 1600 MHz, which is why that data file stops there. I was surprised to
see the B200 seemed to have a better NF than the E312, which averaged 8 dB
NF, by the way.

Very Respectfully,

Dan CaJacob

USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

--

Very Respectfully,

Dan CaJacob

USRP-users mailing listUSRP-users@lists.ettus.comhttp://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

On p.8 of B200 schematic: T801 is Macom ETC1-1-13TR (RF2) T800 is Minicircuits TC1-1-43A+ (RF3) U802 is Anaren BD3150L50100AHF (RF1) On Thu, Feb 1, 2018 at 5:45 PM, Ron Economos via USRP-users < usrp-users@lists.ettus.com> wrote: > There's also a balun on the AD9361 input. Unfortunately, the balun part > number for the low frequency path is not on the schematic. > > Ron > On 02/01/2018 05:39 PM, Dan CaJacob via USRP-users wrote: > > That's an interesting thought. The 9361 does have a pretty bad match. I'll > try adding a 50 Ohm attenuator and see if that helps. > > On Thu, Feb 1, 2018 at 5:14 PM Robin Coxe <robin.coxe@ettus.com> wrote: > >> Hi Dan. Both the B200 and the E312 use the Analog Devices AD9361 RF >> integrated transceiver. This chip does have an integrated LNA. Perhaps >> there's some sort of mismatch between your DUTs and this integrated LNA at >> <1 GHz? >> >> ADI publishes the RX S-parameters: https://ez. >> analog.com/thread/41208#137929 >> >> -Robin >> >> On Thu, Feb 1, 2018 at 7:46 AM, Dan CaJacob via USRP-users < >> usrp-users@lists.ettus.com> wrote: >> >>> Hey guys, >>> >>> I have put together a noise figure meter application that uses a USRP as >>> the sensing device. It started off as a way to measure the NF of the USRP >>> itself. I have a calibrated noise source from an HP 8970B Noise Figure >>> Meter. To test the NF of the USRP, I connect the head to the USRP input. My >>> GNURadio flowgraph maximizes the USRP gain and measures a moving average of >>> the received power while I switch the noise source on and off. The >>> difference in the received power level, in addition to the ENR table from >>> the noise source, can then be used to calculate the NF of the USRP itself >>> using the y-factor method. >>> >>> Once you have the NF for the USRP at many frequencies (I test every 50 >>> MHz from 50 MHz - 6000 MHz), you can modify the same procedure to test the >>> NF of a Device Under Test (DUT) which is connected between the noise source >>> and the (now calibrated) USRP. You can use the USRP cal table we generated >>> in the previous step to derive the NF of the DUT corrected for the NF of >>> the USRP. >>> >>> In short, this all works incredibly well and garners very repeatable >>> results. One complication is that you will see wild NF at certain >>> frequencies due to local interference like LTE and WIFI. I've also compared >>> the results to that which the HP device measures and they're very >>> comparable. ... Except below ~ 1GHz. >>> >>> And here's the issue - I am seeing higher NF for DUTs below about 1GHz >>> and particularly worse below 500 MHz. I was hoping someone at Ettus might >>> be able to shed some light on why this might be. Curiously, the USRPs NF >>> doesn't seem to be too bad, just the DUT. >>> >>> I'll note that I am nominally using a B200 for these tests, but I also >>> tried an E312 just in case the filter banks might help out somehow. I >>> didn't see a difference - they both had the same problem. >>> >>> I have used several DUTs for this test, including LNA boards we have >>> designed ourselves and a Mini-Circuits ZX60-P103LN+ ( >>> https://www.minicircuits.com/pdfs/ZX60-P103LN+.pdf). Both seem to >>> exhibit higher NF when measured with a USRP below 1 GHz. When testing them >>> on the HP NF meter, the NF is as expected all the way down to 50 MHz. >>> >>> I have attached the B200 cal data for your enjoyment as well as the >>> B200-measured ZX60 NF and the HP-measured ZX60. The HP NF meter only goes >>> up to 1600 MHz, which is why that data file stops there. I was surprised to >>> see the B200 seemed to have a better NF than the E312, which averaged 8 dB >>> NF, by the way. >>> -- >>> Very Respectfully, >>> >>> Dan CaJacob >>> >>> _______________________________________________ >>> USRP-users mailing list >>> USRP-users@lists.ettus.com >>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >>> >>> >> -- > Very Respectfully, > > Dan CaJacob > > > _______________________________________________ > USRP-users mailing listUSRP-users@lists.ettus.comhttp://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > > > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > >

DC

Dan CaJacob

Fri, Feb 2, 2018 2:16 AM

I was gonna say, there's actually three of them ;)

On Thu, Feb 1, 2018, 9:06 PM Robin Coxe via USRP-users <
usrp-users@lists.ettus.com> wrote:

On p.8 of B200 schematic:
T801 is Macom ETC1-1-13TR (RF2)
T800 is Minicircuits TC1-1-43A+ (RF3)
U802 is Anaren BD3150L50100AHF (RF1)

On Thu, Feb 1, 2018 at 5:45 PM, Ron Economos via USRP-users <
usrp-users@lists.ettus.com> wrote:

There's also a balun on the AD9361 input. Unfortunately, the balun part
number for the low frequency path is not on the schematic.

Ron
On 02/01/2018 05:39 PM, Dan CaJacob via USRP-users wrote:

That's an interesting thought. The 9361 does have a pretty bad match.
I'll try adding a 50 Ohm attenuator and see if that helps.

On Thu, Feb 1, 2018 at 5:14 PM Robin Coxe robin.coxe@ettus.com wrote:

Hi Dan. Both the B200 and the E312 use the Analog Devices AD9361 RF
integrated transceiver. This chip does have an integrated LNA. Perhaps
there's some sort of mismatch between your DUTs and this integrated LNA at
<1 GHz?

ADI publishes the RX S-parameters:
https://ez.analog.com/thread/41208#137929

-Robin

On Thu, Feb 1, 2018 at 7:46 AM, Dan CaJacob via USRP-users <
usrp-users@lists.ettus.com> wrote:

Hey guys,

I have put together a noise figure meter application that uses a USRP
as the sensing device. It started off as a way to measure the NF of the
USRP itself. I have a calibrated noise source from an HP 8970B Noise Figure
Meter. To test the NF of the USRP, I connect the head to the USRP input. My
GNURadio flowgraph maximizes the USRP gain and measures a moving average of
the received power while I switch the noise source on and off. The
difference in the received power level, in addition to the ENR table from
the noise source, can then be used to calculate the NF of the USRP itself
using the y-factor method.

Once you have the NF for the USRP at many frequencies (I test every 50
MHz from 50 MHz - 6000 MHz), you can modify the same procedure to test the
NF of a Device Under Test (DUT) which is connected between the noise source
and the (now calibrated) USRP. You can use the USRP cal table we generated
in the previous step to derive the NF of the DUT corrected for the NF of
the USRP.

In short, this all works incredibly well and garners very repeatable
results. One complication is that you will see wild NF at certain
frequencies due to local interference like LTE and WIFI. I've also compared
the results to that which the HP device measures and they're very
comparable. ... Except below ~ 1GHz.

And here's the issue - I am seeing higher NF for DUTs below about 1GHz
and particularly worse below 500 MHz. I was hoping someone at Ettus might
be able to shed some light on why this might be. Curiously, the USRPs NF
doesn't seem to be too bad, just the DUT.

I'll note that I am nominally using a B200 for these tests, but I also
tried an E312 just in case the filter banks might help out somehow. I
didn't see a difference - they both had the same problem.

I have used several DUTs for this test, including LNA boards we have
designed ourselves and a Mini-Circuits ZX60-P103LN+ (
https://www.minicircuits.com/pdfs/ZX60-P103LN+.pdf). Both seem to
exhibit higher NF when measured with a USRP below 1 GHz. When testing them
on the HP NF meter, the NF is as expected all the way down to 50 MHz.

I have attached the B200 cal data for your enjoyment as well as the
B200-measured ZX60 NF and the HP-measured ZX60. The HP NF meter only goes
up to 1600 MHz, which is why that data file stops there. I was surprised to
see the B200 seemed to have a better NF than the E312, which averaged 8 dB
NF, by the way.

Very Respectfully,

Dan CaJacob

USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

--

Very Respectfully,

Dan CaJacob

USRP-users mailing listUSRP-users@lists.ettus.comhttp://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

USRP-users mailing list
USRP-users@lists.ettus.com
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

--
Very Respectfully,

Dan CaJacob

I was gonna say, there's actually three of them ;) On Thu, Feb 1, 2018, 9:06 PM Robin Coxe via USRP-users < usrp-users@lists.ettus.com> wrote: > On p.8 of B200 schematic: > T801 is Macom ETC1-1-13TR (RF2) > T800 is Minicircuits TC1-1-43A+ (RF3) > U802 is Anaren BD3150L50100AHF (RF1) > > On Thu, Feb 1, 2018 at 5:45 PM, Ron Economos via USRP-users < > usrp-users@lists.ettus.com> wrote: > >> There's also a balun on the AD9361 input. Unfortunately, the balun part >> number for the low frequency path is not on the schematic. >> >> Ron >> On 02/01/2018 05:39 PM, Dan CaJacob via USRP-users wrote: >> >> That's an interesting thought. The 9361 does have a pretty bad match. >> I'll try adding a 50 Ohm attenuator and see if that helps. >> >> On Thu, Feb 1, 2018 at 5:14 PM Robin Coxe <robin.coxe@ettus.com> wrote: >> >>> Hi Dan. Both the B200 and the E312 use the Analog Devices AD9361 RF >>> integrated transceiver. This chip does have an integrated LNA. Perhaps >>> there's some sort of mismatch between your DUTs and this integrated LNA at >>> <1 GHz? >>> >>> ADI publishes the RX S-parameters: >>> https://ez.analog.com/thread/41208#137929 >>> >>> -Robin >>> >>> On Thu, Feb 1, 2018 at 7:46 AM, Dan CaJacob via USRP-users < >>> usrp-users@lists.ettus.com> wrote: >>> >>>> Hey guys, >>>> >>>> I have put together a noise figure meter application that uses a USRP >>>> as the sensing device. It started off as a way to measure the NF of the >>>> USRP itself. I have a calibrated noise source from an HP 8970B Noise Figure >>>> Meter. To test the NF of the USRP, I connect the head to the USRP input. My >>>> GNURadio flowgraph maximizes the USRP gain and measures a moving average of >>>> the received power while I switch the noise source on and off. The >>>> difference in the received power level, in addition to the ENR table from >>>> the noise source, can then be used to calculate the NF of the USRP itself >>>> using the y-factor method. >>>> >>>> Once you have the NF for the USRP at many frequencies (I test every 50 >>>> MHz from 50 MHz - 6000 MHz), you can modify the same procedure to test the >>>> NF of a Device Under Test (DUT) which is connected between the noise source >>>> and the (now calibrated) USRP. You can use the USRP cal table we generated >>>> in the previous step to derive the NF of the DUT corrected for the NF of >>>> the USRP. >>>> >>>> In short, this all works incredibly well and garners very repeatable >>>> results. One complication is that you will see wild NF at certain >>>> frequencies due to local interference like LTE and WIFI. I've also compared >>>> the results to that which the HP device measures and they're very >>>> comparable. ... Except below ~ 1GHz. >>>> >>>> And here's the issue - I am seeing higher NF for DUTs below about 1GHz >>>> and particularly worse below 500 MHz. I was hoping someone at Ettus might >>>> be able to shed some light on why this might be. Curiously, the USRPs NF >>>> doesn't seem to be too bad, just the DUT. >>>> >>>> I'll note that I am nominally using a B200 for these tests, but I also >>>> tried an E312 just in case the filter banks might help out somehow. I >>>> didn't see a difference - they both had the same problem. >>>> >>>> I have used several DUTs for this test, including LNA boards we have >>>> designed ourselves and a Mini-Circuits ZX60-P103LN+ ( >>>> https://www.minicircuits.com/pdfs/ZX60-P103LN+.pdf). Both seem to >>>> exhibit higher NF when measured with a USRP below 1 GHz. When testing them >>>> on the HP NF meter, the NF is as expected all the way down to 50 MHz. >>>> >>>> I have attached the B200 cal data for your enjoyment as well as the >>>> B200-measured ZX60 NF and the HP-measured ZX60. The HP NF meter only goes >>>> up to 1600 MHz, which is why that data file stops there. I was surprised to >>>> see the B200 seemed to have a better NF than the E312, which averaged 8 dB >>>> NF, by the way. >>>> -- >>>> Very Respectfully, >>>> >>>> Dan CaJacob >>>> >>>> _______________________________________________ >>>> USRP-users mailing list >>>> USRP-users@lists.ettus.com >>>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >>>> >>>> >>> -- >> Very Respectfully, >> >> Dan CaJacob >> >> >> _______________________________________________ >> USRP-users mailing listUSRP-users@lists.ettus.comhttp://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> >> >> >> _______________________________________________ >> USRP-users mailing list >> USRP-users@lists.ettus.com >> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> >> > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > -- Very Respectfully, Dan CaJacob

DC

Dan CaJacob

Sat, Feb 3, 2018 3:41 PM

I saw no improvement when including a 3dB 50 Ohm attenuator as part of the
B200 NF meter. I guess I could try higher attenuator values.

On Thu, Feb 1, 2018 at 9:16 PM Dan CaJacob dan.cajacob@gmail.com wrote: