Posted: Sat Feb 02, 2008 10:07 pm Post subject: Noise Power Ratio measurement
--------------------------------------------------------------------------------
Hello,
I have used the Agilent Signal Studio software to create an NPR test stimulus with the following parameters:
- number of tones = 1001
- noise pedestal width = 80 MHz
- notch width = 800 KHz (10 tones in the notch)
- tone spacing = 80 KHz
- Notch depth = 50 dB
- Fcenter = 20 GHz
- using Agilent E8267D VSG and E4448 PSA
- PSA att = 22 dB
- external att = 30 dB
- Pmax of DUT ~ 40 dBm
- Gain of DUT ~ 25 dB
My questions are in regards to the actual NPR measurement on the PSA:
Note that I am writing software with Agilent VEE pro to measure a DUT's NPR, after the creation of an NPR stimulus with the signal-studio-software.
Note that I'm using the 'band-interval-power' marker function on the PSA
- is 'notch depth' and system NPR the same thing?
- the signal studio software reports a notch depth of 50 dB after correction,
how does signal studio compute this notch depth? When we actually measure the DUT, is it accurate to use the same method to calcuate the DUT's NPR as signal studio did? If so, how does signal studio exactly calculate notch depth?
- How does the width of the band-interval-power affect the NPR measurement?
- I am seeing differences in the calculation of NPR with different sizes of band interval - using the 'band-intvl-pwr' function. Why would we see differences? NPR = (pedestal pwr dBm/ Hz divided by notch pwr dBm/ Hz)
- I am seeing large differences in the calculation of NPR with different res-BW setting on the PSA. Any advice on how to select optimum RSB for measuring synthesized noise - in the notch and in the pedestal?
- I would like to detect as deep into the notch as possible - however, I'm concerned with the fact that the tones are 80KHz apart. What res-BW is reccomended ? What's the best way to determine the optimum res-BW and 'band-interval-power width' for this measurment?
- If we were using a 'true' AWGN generator, I guess that the res-BW and size of the 'band-interval-power' would not matter. However, since the noise pedestal is synthesized, any insight on how to perform NPR measurements with the PSA (with highest possible dynamic range) would be very helpful.
thank you,
Nick
--------------------------------------------------------------------------------
Hello,
I have used the Agilent Signal Studio software to create an NPR test stimulus with the following parameters:
- number of tones = 1001
- noise pedestal width = 80 MHz
- notch width = 800 KHz (10 tones in the notch)
- tone spacing = 80 KHz
- Notch depth = 50 dB
- Fcenter = 20 GHz
- using Agilent E8267D VSG and E4448 PSA
- PSA att = 22 dB
- external att = 30 dB
- Pmax of DUT ~ 40 dBm
- Gain of DUT ~ 25 dB
My questions are in regards to the actual NPR measurement on the PSA:
Note that I am writing software with Agilent VEE pro to measure a DUT's NPR, after the creation of an NPR stimulus with the signal-studio-software.
Note that I'm using the 'band-interval-power' marker function on the PSA
- is 'notch depth' and system NPR the same thing?
- the signal studio software reports a notch depth of 50 dB after correction,
how does signal studio compute this notch depth? When we actually measure the DUT, is it accurate to use the same method to calcuate the DUT's NPR as signal studio did? If so, how does signal studio exactly calculate notch depth?
- How does the width of the band-interval-power affect the NPR measurement?
- I am seeing differences in the calculation of NPR with different sizes of band interval - using the 'band-intvl-pwr' function. Why would we see differences? NPR = (pedestal pwr dBm/ Hz divided by notch pwr dBm/ Hz)
- I am seeing large differences in the calculation of NPR with different res-BW setting on the PSA. Any advice on how to select optimum RSB for measuring synthesized noise - in the notch and in the pedestal?
- I would like to detect as deep into the notch as possible - however, I'm concerned with the fact that the tones are 80KHz apart. What res-BW is reccomended ? What's the best way to determine the optimum res-BW and 'band-interval-power width' for this measurment?
- If we were using a 'true' AWGN generator, I guess that the res-BW and size of the 'band-interval-power' would not matter. However, since the noise pedestal is synthesized, any insight on how to perform NPR measurements with the PSA (with highest possible dynamic range) would be very helpful.
thank you,
Nick
The bulleted answers are below your individual questions:
Hello,
I have used the Agilent Signal Studio software to create an NPR test stimulus with the following parameters:
- number of tones = 1001
- noise pedestal width = 80 MHz
- notch width = 800 KHz (10 tones in the notch)
- tone spacing = 80 KHz
- Notch depth = 50 dB
- Fcenter = 20 GHz
- using Agilent E8267D VSG and E4448 PSA
- PSA att = 22 dB
- external att = 30 dB
- Pmax of DUT ~ 40 dBm
- Gain of DUT ~ 25 dB
My questions are in regards to the actual NPR measurement on the PSA:
Note that I am writing software with Agilent VEE pro to measure a DUT's NPR, after the creation of an NPR stimulus with the signal-studio-software.
Note that I'm using the 'band-interval-power' marker function on the PSA
- is 'notch depth' and system NPR the same thing?
No, notch depth is level difference between the average pedestal level and the average notch level.
NPR is the ratio of the power of the spectral components in the system passband to the power of those in the notch.
- the signal studio software reports a notch depth of 50 dB after correction,
how does signal studio compute this notch depth? When we actually measure the DUT, is it accurate to use the same method to calcuate the DUT's NPR as signal studio did? If so, how does signal studio exactly calculate notch depth?
The waveform consists of a number of tones each at a certain amplitude and spacing. For a given notch depth and notch width ideally signal studio figures out both the number of tones and the level.
Signal studio along with the PSA or X-series spectrum analyzer can then be use to measurement and adjust the notch depth by minimizing each tone in the notch.
Signal Studio does not actually calculate NPR but rather configures a wideband noise signal with a user-defined notch positioned within the noise bandwidth. An advanced predistortion algorithm is used to increase the notch dynamic range and improve noise flatness of the NPR test stimulus.
- How does the width of the band-interval-power affect the NPR measurement?
NPR is the ratio of the power of the spectral components in the system passband to the power of those in the notch. This means that for the system passband, the width of the band-interval-power must only include the actual power contained in the DUT’s passband and also for the notch, the width of the band-interval-power must only include the actual power contained in the notch’s passband. Any errors in these measurements will directly result in NPR errors.
- I am seeing differences in the calculation of NPR with different sizes of band interval - using the 'band-intvl-pwr' function. Why would we see differences? NPR = (pedestal pwr dBm/ Hz divided by notch pwr dBm/ Hz)
The band intervals need to always match those of the test stimulus signal. Your signal for example, would need to be measured in an 80MHz bandwidth for the DUT’s passband and 80KHz bandwidth for the notch’s passband. The notch is a noise signal, and as you increase the band interval, you couple in more power to your measurement, and therefore the result will change.
- I am seeing large differences in the calculation of NPR with different res-BW setting on the PSA. Any advice on how to select optimum RSB for measuring synthesized noise - in the notch and in the pedestal?
Decreasing the RBW increases the sensitivity of the analyzer; however, as you decrease the RBW you will get to a point where decreasing the RBW no longer lowers the integrated power measured in the notch. At this point you are measuring the notch power. If you continue to decrease the RBW, the measurement will take longer to make without any improvement to the results.
- I would like to detect as deep into the notch as possible - however, I'm concerned with the fact that the tones are 80KHz apart. What res-BW is reccomended ? What's the best way to determine the optimum res-BW and 'band-interval-power width' for this measurment?
In general the RBW chosen would be approximately one fiftieth of the tone spacing (provided you don’t get into too narrow of RBWs – i.e. 10 Hz or less). You basically want to have narrow enough of RBW to not couple power for the adjacent tones but you want to have wide enough of RBW to have fast repeatable results.
- If we were using a 'true' AWGN generator, I guess that the res-BW and size of the 'band-interval-power' would not matter. However, since the noise pedestal is synthesized, any insight on how to perform NPR measurements with the PSA (with highest possible dynamic range) would be very helpful.
An RBW of about one-fiftieth of the measured span is a good rule of thumb. Decreasing the RBW would not affect the integrated power if you are just measuring noise, but decreasing the span increases the measurement time and if you already have sufficient dynamic range to measure the DUT’s performance, it doesn’t make sense to keep decreasing the RBW.
Regards -