Measurement of low noise amplifier

[eason.RFIC]

i have design a 45GHz low noise amplifier ,we have a PNA N5247A ，what else instrument should i buy when i want to measure the noise figure of the LNA ,thanks

 

[drkirkby]

Well, since that covers 10 MHz to 67 GHz, I doubt you need any more than a calibrated noise source.

I note one of the options is 029 (Fully-Corrected Noise-Figure Measurements to 50 GHz), so I'm not sure exactly what that does. Perhaps it might be worth adding that option - you would need to speak to Agielnt. But that option will cost you $84,000. That is some seriously expensive noise figure meter you have there - do you want to swap it for my vintage AVO meter?

I don't know the device, although I have used lower cost (i.e. older NF meters). But there are very good Agilent forums for test and measurement equipment. I find posting a question on there gets answers very fast. Many of the people reading and replying on the forums are Agilent employees. "Dr_joel" seems the guru on vector network analyzers, but I'm sure there will be someone who is a guru on noise figure meters.

But I assume if you have spent over $200k on a noise figure meter, Agilent will offer you excellent support.

Dave

 

[jiripolivka]

It depends how detailed knowledge of your 45 GHz LNA you need.
Assuming that it will be used as a 45 GHz receiver front end, you need the complete receiver to evaluate its noise figure.
Then you can use the Y-factor method and a RF or IF power meter, to read power-level change when the ambient AND a higher or lower than ambient noise temperature is connected to LNA input.
The ambient and the other input noise temperatures can be injected to the LNA input by a small horn. You point the horn to a room wall to get the ambient temperature, and use a microwave absorber or a wet paper pack or a wet wooden board instead. If you heat or cool the absorber to a known temperature (boiling water for ~400 K or boiling nitrogen for 77 K), you can measure he corresponding output RF or IF power and calculate receiver NF from it. If you need a frequency plot f the NF, you will need an expensive calibrated instrumentation as above.

 

[drkirkby]

Since he already has a 67 GHz NF meter costing over $200k, should he need any more than a calibrated noise source?

 

[D.A.(Tony)Stewart]

Consider 3 methods; **broken link removed**
Gain and Y factor method uses common spectrum analyzer.

Consider Calibrated Noise source
https://noisecom.com/products/calibrated-sources

 

[jiripolivka]

Since he already has a 67 GHz NF meter costing over $200k, should he need any more than a calibrated noise source?

I would expect that such NF meter comes with calibrated noise sources...)

 

[D.A.(Tony)Stewart]

You may have to consult with Agilent on expensive optional software Fully-Corrected Noise-Figure Measurements to 50 GHz

 

[drkirkby]

I would expect that such NF meter comes with calibrated noise sources...)

I doubt it.

The noise source is unlikely to be part of the analyzer. The correct noise source has to be chosen depending on the frequency range of interest and the excess noise ratio (ENR).

The noise sources are not cheap either. On the used market you can pick up an HP 8970A noise figure meter for less than a used HP noise source.

Dave

- - - Updated - - -

You may have to consult with Agilent on expensive optional software Fully-Corrected Noise-Figure Measurements to 50 GHz

I wondered what that option was.

I recently bought a hand-held VNA from Agilent, and found the cost of the options doubled the cost of the basic instrument. ALL these are software options. Just to enable the time domain reflectometer, which is just an FFT of the frequency response, is over $5000.

I recently bought a used 20 GHz benchtop VNA, which had the TDR added by the cal lab before it was put on eBay. I somehow suspect there are places that know how to add these options, but of course it is not a satisfactory solution if your equipment is under warranty. In fact, it did cross my mind if Agilent woud remove an "unauthorisd" options if an instrument was sent to them for calibration.

Dave

 

[eason.RFIC]

i have found that the options 029 is too expensive to me .so i must find other method to measure the noise figure ,there is a N8975A in my lab , but its maximum frequency is 26.5G ,so i must add blockdown converter ,N8975AZ-K40/K50 is a blockdown converter for N8975A by agilent ,but it is also expensive . more than $50.000 . our lab cann`t afford this . i want to measure the NF using money blow $5.000. so do you have any suggestion for me to achieve this .thanks

- - - Updated - - -

Well, since that covers 10 MHz to 67 GHz, I doubt you need any more than a calibrated noise source.

I note one of the options is 029 (Fully-Corrected Noise-Figure Measurements to 50 GHz), so I'm not sure exactly what that does. Perhaps it might be worth adding that option - you would need to speak to Agielnt. But that option will cost you $84,000. That is some seriously expensive noise figure meter you have there - do you want to swap it for my vintage AVO meter?

I don't know the device, although I have used lower cost (i.e. older NF meters). But there are very good Agilent forums for test and measurement equipment. I find posting a question on there gets answers very fast. Many of the people reading and replying on the forums are Agilent employees. "Dr_joel" seems the guru on vector network analyzers, but I'm sure there will be someone who is a guru on noise figure meters.

But I assume if you have spent over $200k on a noise figure meter, Agilent will offer you excellent support.

Dave

i have found that the options 029 is too expensive to me .so i must find other method to measure the noise figure ,there is a N8975A in my lab , but its maximum frequency is 26.5G ,so i must add blockdown converter ,N8975AZ-K40/K50 is a blockdown converter for N8975A by agilent ,but it is also expensive . more than $50.000 . our lab cann`t afford this . i want to measure the NF using money blow $5.000. so do you have any suggestion for me to achieve this .thanks

 

[D.A.(Tony)Stewart]

YOu need 2 things and 1 test method using Y parameters.

WR-15 Noise Source
• To perform Noise Figure Measurement of 60GHz, Noise Figure Analyzer with Block Down Convertor are required.

N8975AZ-K67 Block Down Convertor (58-67GHz)

http://noisecom.com/products/calibrated-sources/60-ghz-noise-figure-test-set

Using Spectrum Analyzer without Noise Figure Software is harder. but possible to learn.

This R&S manual may or may not help.

**broken link removed**

 

[drkirkby]

I can't be bothered to read all the specs and options on the Agient web site, but from what I can see from a quick glance, your noise figure meter works to 67 GHz. So what exactly is stopping you using just a noise source and the meter you already have?

Dave

 

[eason.RFIC]

my noise figure is N8975A,its maximum frequency is only 26.5GHz , so i need a blockdown converter ,agilent has some converter can achieve the request ,but it is so expensive ,so i want to find some block converter which price small than $5000 ,can you give me some advise .thanks

I can't be bothered to read all the specs and options on the Agient web site, but from what I can see from a quick glance, your noise figure meter works to 67 GHz. So what exactly is stopping you using just a noise source and the meter you already have?

Dave

- - - Updated - - -

YOu need 2 things and 1 test method using Y parameters.

WR-15 Noise Source
• To perform Noise Figure Measurement of 60GHz, Noise Figure Analyzer with Block Down Convertor are required.

N8975AZ-K67 Block Down Convertor (58-67GHz)

http://noisecom.com/products/calibrated-sources/60-ghz-noise-figure-test-set

Using Spectrum Analyzer without Noise Figure Software is harder. but possible to learn.

This R&S manual may or may not help.

**broken link removed**

my LNA is work at 45GHz ,agilent block down convertor N8975AZ-K50 can achieve my request , but i find it is so expensive(more than $50,000) ,i want to find some block down converter blow $5,000 ,can you give me some advice ,thanks

 

[tony_lth]

There is a coarse method I had used for test mm-wave noise figure.
Use a low freq, say 18G noise figure meter and a 18G noise figure source. First cal with these two components. You should know calibration is to calibrate the noise figure meter.
Then replace the 18G noise figure source with 45GHz noise figure source, DON'T do calibration again. And just input 8mm ENR of 45GHz noise figure source.
You can measure now.
My previous company used this method to measure more than 100 receivers. And I estimate the error is very low to ignore.

 

[drkirkby]

Do you by any chance have one of Agilent's PNA-X series VNAs? If so, I just found out that they can do very accurate vector corrected noise figure measurements. It might again be an option - I have no idea, but it might be worth investigating. I've not read it myself yet, but take a look at

**broken link removed**

I think when you own a noise figure meter costing over $200,000, any options for it from Agilent are going to cost you more than $5000. I paid about $3000 to have a TDR option added to my portable $8000 portable VNA.

Perhaps you should look on eBay and see if you can find a mixer which will work at the frequency you want.

I've not read **broken link removed** myself, but I am going to take a read, as the idea of using a VNA for noise figure measurements is not something I have ever thought about, but even without reading it, I can see how one might go about it, if S21 is measured sepparately with the noise on and noise off, and the standard deviation of them is compared.

Hopefully the above will at least give you some ideas.

Perhaps you could also tell your boss that $5000 is not a lot of money to allocate to a project of a 47 GHz LNA.

Dave

 

[eason.RFIC]

There is a coarse method I had used for test mm-wave noise figure.
Use a low freq, say 18G noise figure meter and a 18G noise figure source. First cal with these two components. You should know calibration is to calibrate the noise figure meter.
Then replace the 18G noise figure source with 45GHz noise figure source, DON'T do calibration again. And just input 8mm ENR of 45GHz noise figure source.
You can measure now.
My previous company used this method to measure more than 100 receivers. And I estimate the error is very low to ignore.

you say use a 18G noise figure meter measure a 45G LNA directly ? i want to ask Does this method need any mixer ?

 

[eason.RFIC]

Do you by any chance have one of Agilent's PNA-X series VNAs? If so, I just found out that they can do very accurate vector corrected noise figure measurements. It might again be an option - I have no idea, but it might be worth investigating. I've not read it myself yet, but take a look at

**broken link removed**

I think when you own a noise figure meter costing over $200,000, any options for it from Agilent are going to cost you more than $5000. I paid about $3000 to have a TDR option added to my portable $8000 portable VNA.

Perhaps you should look on eBay and see if you can find a mixer which will work at the frequency you want.

I've not read **broken link removed** myself, but I am going to take a read, as the idea of using a VNA for noise figure measurements is not something I have ever thought about, but even without reading it, I can see how one might go about it, if S21 is measured sepparately with the noise on and noise off, and the standard deviation of them is compared.

Hopefully the above will at least give you some ideas.

Perhaps you could also tell your boss that $5000 is not a lot of money to allocate to a project of a 47 GHz LNA.

Dave

it not only $5000 , the option 029 for nosie figure measurement may cost more than $70,000