| Author |
Message |
littlemmx
Joined: 14 Jun 2003
Posts: 29
|
 28 Aug 2003 12:01 how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
i want to measure a x-band lna use the 8970a,i don't know whether the method that i used right or not.
i use the 1.3 mode,with a mixer and a LO source, then i calibrate the device,after that ,i connect the lna between the noise source and the mixer's rf port,the mixer's if port connect to 8970a.
that all.
the connector and the mixer giving me another error,isn't it? how much is it ?
|
|
| Back to top |
|
 |
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 28 Aug 2003 12:25 |
|
|
|
Measurement mode 1.3 is wrong!
You must use 1.1 or 1.2 (it does depend on the type of oscillator you have, variable or fixed respectively).
If you have a variable local oscillator (meas mode 1.1), use SF 3.0 to set the IF frequency.
Use a LNA in front of the mixer to reduce the NF of the measurement system.
External filtering (after the LNA) is needed for SSB measurements.
Connect the IF port of the mixer to the input port of the 8970B.
Connect the noise source output to the input of your meas system (input of the LNA if you have , or input of the mixer) and then calibrate (ALL) the meas system.
Then, remove the source and insert the LNA and measure the corrected NF of the DUT.
p.s. the NF of the mixer is taken in account from the calibration
|
|
| Back to top |
|
|
littlemmx
Joined: 14 Jun 2003
Posts: 29
|
| 28 Aug 2003 15:27 |
|
|
|
Thanks u very much.
i still have a question,you said i should connect a external filter after the LNA if i'm in SSB measurement.
what is the use of the filter?why don't connect the filter after the mixer?
and another one,when i can use the mode 1.3
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 28 Aug 2003 15:37 |
|
|
|
| littlemmx wrote: |
Thanks u very much.
i still have a question,you said i should connect a external filter after the LNA if i'm in SSB measurement.
what is the use of the filter?why don't connect the filter after the mixer?
and another one,when i can use the mode 1.3 |
The measurement mode 1.3 may be used when the DUT is a frequency traslating device (mixer).
The filter rejects the image frequency that otherwise goes into the mixer; if you have not the filter make a DSB (use special function 2.0) measurement and set the IF frequency as low as possible (15MHz for example).
|
|
| Back to top |
|
|
htforever
Joined: 01 Jul 2003
Posts: 74
Location: China
|
| 29 Aug 2003 4:15 |
|
|
|
Hi,I also want to know How to use HP 8970A to measure NF of my LNA at 2GHz.I have some questions to ask.
I use mode 1.2, I use RMS-25MH as mixer,A RF filter(BW:1920-1980MHz) placed at the RF port of the mixer. The CONVERSION LOSS of mixer is 7-9dB, and the CONVERSION LOSS of filter is about 3dB. I measured the filter and mixer, they do work properly.
I set the start,stop,step freq as 1920,1980,10. then 3.1 SF 1400, the freq of fixed LO is 1400MHz. But when I calibrate, E27 was apperaed, It's means overflow while calibration. Even sometimes, after calibration,no E27 was appeared, but the NF is still big, and is changing all the time.
What's the reason? what should I do in order to measure the NF? Need I add a amplifier in front of the mixer when do the calibration? If so, what kind of amplifier should I use? Low Noise, high Gain?
Waiting for your reply anxiously. Thanks very much.
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 29 Aug 2003 7:26 |
|
|
|
Ok. FIrst af all, use SF 32.0 to set the input gain stage inside the instrument to the highest value (default value). You may obtain this also pushing on PRESET.
Probably this will not solve your problem because the losses in front of the instrument seem to be too much; use a LNA (low noise as much as possible, and at least 20 dB gain) and E27 will not appear anymore.
p.s. use the LNA in front of the filter, as first stage (better if you also add, in front of this last one, an isolator).
To reduce the jitter in the display reading increas the smoothing.
regards
|
|
| Back to top |
|
|
htforever
Joined: 01 Jul 2003
Posts: 74
Location: China
|
| 30 Aug 2003 9:05 |
|
|
|
To marcomdd:
Why the gain of the amplifier in front of the filter should be more than 20dB?
The loss of the filter and mixer is about 10-12dB. 20dB need two stages amplifier.
All I need to measure the NF of a amplifier at 2GHz is a amplifier with high gain and low noise, followed by a filter and then a mixer?
Is there anything should pay attention to?
Thanks very much!
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 30 Aug 2003 10:12 |
|
|
|
With a 20dB gain amplifier you are able to reduce the noise of the passive blocks (filter and mixer) one hundred times, so that the noise factor of all the instrumentation is, more or less, equal to the noise factor of the (low noise) amplifier.
Low noise instrumentation means lower uncertainty !
Regards
p.s. if you have only 10dB gain .....better than nothing!
|
|
| Back to top |
|
|
htforever
Joined: 01 Jul 2003
Posts: 74
Location: China
|
| 13 Sep 2003 10:20 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
marcomdd,Thanks very much!
Now I can use mode 1.2 to measure the NF of LNA at 1.95GHz using HP 8970A.But Now I have another question to ask.Because of the diamatch between the LNA(DUT) and the mixer module, The Gain and NF measured by the Noise Figure Meter 8970A is not accurate.
The Gain of the mixer module is about 15dB. The gain of the DUT(LNA) is about 13dB, But when the LNA and mixer are connected together, the whole Gain is not 15+13=28dB, but only 22dB.
The measurement result of the Noise figure meter is 7dB(Gain)(not 13dB) and 2dB(NF).
I guess the gain and NF presented by 8970A are calculated(22-15~=7).
The output VSWR of LNA is 1.5 and the VSWR of mixer module is about 1.5.
So how can I make accurate measurement?
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 13 Sep 2003 12:54 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
What do you mean with diamatch ? Perhaps mismatch ?
Please with only the mixer module in place , connect the noise source and measure the noise figure of the instrumentation. Say me the value.
Then , which is the noise source you have (model) ? More important, do you insert the exact ENR values in the noise figure meter ?
I need more informations to help you.
|
|
| Back to top |
|
|
htforever
Joined: 01 Jul 2003
Posts: 74
Location: China
|
| 14 Sep 2003 12:50 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
I think it's perhaps mismatch,as the gain of the LNA+mixer is not just 13+15dB.
I just used mode 1.4 to measure the mixer model, The LO freq is 2.2GHz,power is 12dBm.
The IF is 220-280MHz(So the RF is 1.92-1.98GHz), the result is
IF Freq(MHz) Gain(dB) NF(dB)
280 19.7 10
270 15.8 9.3
260 17.6 8
250 17.6 7.5
240 16.2 7.8
230 15.1 8.4
220 14.6 8.2
the Noise source is HP 346A,freq rang is 10MHz～18GHz,I do input the ENR correctly.
Do you need any more information about my measurement? Thanks very much.
|
|
| Back to top |
|
|
littlemmx
Joined: 14 Jun 2003
Posts: 29
|
| 14 Sep 2003 15:01 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
To marcomdd:
first,thank you for all your reply very much.
and,i also want to know the maxim nf figure the Noise Figure Meter 8970A can measure.
for example,if i wanted to measure a device which nf figure was about 20db,8970A is ok?
|
|
| Back to top |
|
|
littlemmx
Joined: 14 Jun 2003
Posts: 29
|
| 14 Sep 2003 15:12 |
|
|
|
[quote="marcomdd"]
The measurement mode 1.3 may be used when the DUT is a frequency traslating device (mixer).
The filter rejects the image frequency that otherwise goes into the mixer; if you have not the filter make a DSB (use special function 2.0) measurement and set the IF frequency as low as possible (15MHz for example).[/quote]
when i set IF frequency lower than 30mhz , it display a wrong,do you konw what means it is ?
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 14 Sep 2003 19:16 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
| htforever wrote: |
I think it's perhaps mismatch,as the gain of the LNA+mixer is not just 13+15dB.
I just used mode 1.4 to measure the mixer model, The LO freq is 2.2GHz,power is 12dBm.
The IF is 220-280MHz(So the RF is 1.92-1.98GHz), the result is
IF Freq(MHz) Gain(dB) NF(dB)
280 19.7 10
270 15.8 9.3
260 17.6 8
250 17.6 7.5
240 16.2 7.8
230 15.1 8.4
220 14.6 8.2
the Noise source is HP 346A,freq rang is 10MHz～18GHz,I do input the ENR correctly.
Do you need any more information about my measurement? Thanks very much. |
1- Repeat the measurement with an isolator at the input of the LNA+mixer . This will isolate the instrumentation from the change of the input reflection coefficient between the calibration step and the measurement one.
2- do you have a single block that aplifies and down converts the signal? How do you measure it ? With the NFM ? Being a translating device you have two use two different ENR values , the IF value for the calibration step and the RF value for the measurement step.
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 14 Sep 2003 19:23 |
|
|
|
| littlemmx wrote: |
| marcomdd wrote: |
The measurement mode 1.3 may be used when the DUT is a frequency traslating device (mixer).
The filter rejects the image frequency that otherwise goes into the mixer; if you have not the filter make a DSB (use special function 2.0) measurement and set the IF frequency as low as possible (15MHz for example). |
when i set IF frequency lower than 30mhz , it display a wrong,do you konw what means it is ? |
I have a HP 8970B and the minimum frequency is 10 MHz!
Which is the minimum frequency for your 8970A?
Which is the error code you read on the display ?
|
|
| Back to top |
|
|
littlemmx
Joined: 14 Jun 2003
Posts: 29
|
| 15 Sep 2003 1:33 |
|
|
|
the minimum frequency by the instruction book of the 8970a is 10mhz!
but i used the mode 1.2 or 1.1,when i set IF frequency lower than 30mhz,i doesn't work in the process of calibration,it displays E27.
and,another,the maximum NF figure the 8970A can measure.
for example,if i wanted to measure a device which nf figure was about 20db,8970A is ok?
_________________
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 15 Sep 2003 13:17 |
|
|
|
| littlemmx wrote: |
the minimum frequency by the instruction book of the 8970a is 10mhz!
but i used the mode 1.2 or 1.1,when i set IF frequency lower than 30mhz,i doesn't work in the process of calibration,it displays E27.
and,another,the maximum NF figure the 8970A can measure.
for example,if i wanted to measure a device which nf figure was about 20db,8970A is ok?
_________________ |
The min NF my 8970B can measure is 30 dB.
E27 means too much loos in the calibration path. Put a LNA at the input of the NFM.
|
|
| Back to top |
|
|
littlemmx
Joined: 14 Jun 2003
Posts: 29
|
| 16 Sep 2003 1:35 |
|
|
|
| when i calibrate,should i connect the LNA in front of the mixer?
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 16 Sep 2003 7:29 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
| Yes.
|
|
| Back to top |
|
|
littlemmx
Joined: 14 Jun 2003
Posts: 29
|
| 24 Sep 2003 1:42 |
|
|
|
When i measure amplifier with the noise figure meter 8970,i found the Gain was less than others instrument,such as Vector Network measure system,Spectrum Analyzer,etc.
why?
|
|
| Back to top |
|
|
htforever
Joined: 01 Jul 2003
Posts: 74
Location: China
|
| 24 Sep 2003 4:11 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
I have two independent blocks: a LNA at 1.9GHz, a mixer(containing a amplifier and a RF filter).
I measure the LNA+mixer with HP 8970A using mode 1.4, mode 1.4 needs to input IF range and fixed LO freq.
from the Friis formula:Ftotal=F1+(F2-1)/Ga1,
F2 means the NF of mixer, Ftotal means the NF of LNA+mixer. F1 is the NF of LNA which is unknown and want to be measured.
I can get F2 and Ftotal using mode 1.4 of HP 8970A. Now the problem is how to get Ga1, Ga1 is the available Gain of the LNA. Generally speaking, we can't measure Ga1 directly, but it can be calculated from its formula.In the formula, S parameter can be measured from the VNA, but there is a "source reflection coefficient", what does it mean? how to measure it?
Thx!
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 24 Sep 2003 8:09 |
|
|
|
| littlemmx wrote: |
When i measure amplifier with the noise figure meter 8970,i found the Gain was less than others instrument,such as Vector Network measure system,Spectrum Analyzer,etc.
why? |
How much lower ?
The Noise Figure Metere makes scalar measurements, as well as the spectrum analyzer; therefore, there would be some differences with the VNA results if the LNA is not well matched.
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 24 Sep 2003 8:19 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
| htforever wrote: |
I have two independent blocks: a LNA at 1.9GHz, a mixer(containing a amplifier and a RF filter).
I measure the LNA+mixer with HP 8970A using mode 1.4, mode 1.4 needs to input IF range and fixed LO freq.
from the Friis formula:Ftotal=F1+(F2-1)/Ga1,
F2 means the NF of mixer, Ftotal means the NF of LNA+mixer. F1 is the NF of LNA which is unknown and want to be measured.
I can get F2 and Ftotal using mode 1.4 of HP 8970A. Now the problem is how to get Ga1, Ga1 is the available Gain of the LNA. |
Rigth ! F2 is the (linear) noise factor of your down converting block !
F1 is the LNA unknown noise factor and Ftot is the LNA+mixer noise factor.
You may obtain the Gins (insertion gain ) of the LNA directly from the NFM by rationing the measured power with the LNA in place with the measured power without the LNA .
| htforever wrote: |
Generally speaking, we can't measure Ga1 directly, but it can be calculated from its formula.In the formula, S parameter can be measured from the VNA, but there is a "source reflection coefficient", what does it mean? how to measure it?
Thx! |
The source reflection coefficient is that of the noise source.
But pay attention ! The noise source has two different states and two different reflection coefficients.
Try using the Gins, available directly from the NFM.
Last edited by marcomdd on 24 Sep 2003 17:12; edited 1 time in total |
|
| Back to top |
|
|
htforever
Joined: 01 Jul 2003
Posts: 74
Location: China
|
| 24 Sep 2003 15:09 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
But in the Friis formula, Ftotal=F1+(F2-1)/Ga1,
The Ga1 means the available Gain of the LNA, but not the insertion gain(Gi).
In order to calculate the F1,I need to know Ga1, How can I get available Gain of the LNA(Ga1)?
There is a formula to calculate Ga1, in the formula we need to know S11,S12,S21,S22 and
the source reflection coefficient of the noise source while hot. How can I get the source reflection coefficient of the noise source while hot?
Thx!
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 24 Sep 2003 17:11 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
| htforever wrote: |
But in the Friis formula, Ftotal=F1+(F2-1)/Ga1,
The Ga1 means the available Gain of the LNA, but not the insertion gain(Gi).
In order to calculate the F1,I need to know Ga1, How can I get available Gain of the LNA(Ga1)?
There is a formula to calculate Ga1, in the formula we need to know S11,S12,S21,S22 and
the source reflection coefficient of the noise source while hot. How can I get the source reflection coefficient of the noise source while hot?
Thx! |
Obviously Gav is not Gins.
But if the noise source, the LNA and the Noise Figure Meter are well matched --> Gav is more or less equal to Gins.
To measure the reflection coefficient of the noise source you need a vector network analyzer (use the special function 9.2 to set in the hot state the noise source).
|
|
| Back to top |
|
|
godz
Joined: 22 Sep 2003
Posts: 34
|
| 25 Sep 2003 9:31 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
Hi,
I am measuring the NF of a monolithic mixer at 900MHz and IF frequency of 200MHz, the objective is to measure the SSB NF, after calibration with the noise source only i inserted RF BP filter centered around the 900MHz before the mixer RF port and an IF filter of center frequency 177MHz at the output of the mixer, then i entered the values of loss compensation tables for the RF filter before the DUT and the IF filter after the DUT, the NFA type i am using is HP N8973A, now to my surprise i found gain very large 26dB while expected 5.5dB( from spectrum analyzer) and NF of -8dB!!!. When i disabled the RF filter loss compensation table and substituted it with a fixed value of 3dB( which is the filter IL at center frequecny 900HMz) the results are reasonable and close to expected (G=5.4dB and NF=13.8dB). I tried another thing, i deleted all the values of the RF filter loss comp. table that lie in the stop band of the filter and i find the results getting more reasonable!!.
Does anyone has explanation?!
|
|
| Back to top |
|
|
htforever
Joined: 01 Jul 2003
Posts: 74
Location: China
|
| 25 Sep 2003 9:59 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
I have another question to ask: what's the difference between uncorrected NF and corrected Gain&NF?
SF 9.2 means Noise Source On(uncorrected), why don't use SF 9.4 Noise Source On(corrected)?
Ftotal(the NF of LNA+mixer) and F2(the NF of mixer) can be measured using HP 8970A with mode 1.4, but mode 1.4 consist of two kinds of NF,uncorrected NF and corrected NF,which one is the Ture Ftotal and F2?
Thanks very much!
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 25 Sep 2003 11:09 Re: how to measure the noise up to 1.6ghz use the HP 8970A |
|
|
|
| htforever wrote: |
I have another question to ask: what's the difference between uncorrected NF and corrected Gain&NF?
|
The uncorrected NF is the NF of the DUT plus the noise figure of the instrumentation (Noise figure measurement system) you use to perform the measurement (your Ftotal).
The corrected noise figure is just the DUT noise figure (your F1).
| htforever wrote: |
SF 9.2 means Noise Source On(uncorrected), why don't use SF 9.4 Noise Source On(corrected)?
|
It's the same. You use this special function to switch (steady state) on the noise source, then you perform the measurement through a VNA
| htforever wrote: |
Ftotal(the NF of LNA+mixer) and F2(the NF of mixer) can be measured using HP 8970A with mode 1.4, but mode 1.4 consist of two kinds of NF,uncorrected NF and corrected NF,which one is the Ture Ftotal and F2?
Thanks very much! |
The uncorrected noise figure with the DUT in place is Ftotal, the uncorrected noise figure without the DUT is your F2.
|
|
| Back to top |
|
|
littlemmx
Joined: 14 Jun 2003
Posts: 29
|
| 27 Sep 2003 6:08 |
|
|
|
| if i don't have system LO,i use Signal Generator as the mixer's LO,will it effect the NF of the amp?
|
|
| Back to top |
|
|
marcomdd
Joined: 11 Jan 2003
Posts: 158
Helped: 3
Location: Rome
|
| 27 Sep 2003 20:14 |
|
|
|
| littlemmx wrote: |
| if i don't have system LO,i use Signal Generator as the mixer's LO,will it effect the NF of the amp? |
What do you mean?
|
|
| Back to top |
|
|
