Noise figure measurement using an SA44B with Spike software

I successfully measured noise figures (NF) by the Y factor method using the measurement receiver utility (MRU) in the version 2.18b SA44B software. The results being similar to those obtain with a professional noise figure meter.

However using the Spike version 3.0.21 the MRU gives a Y factor about 3dB less than the expected value. In contrast measuring off a sweep I get the expected value.

What am I doing wrong?

My Firmware version is 2.03.

Richard

Hi Richard,

This sounds like it could be related to the image rejection algorithm for the SA44B. Do you get expected results when you disable ‘Spur Reject’ in the ‘Settings’ file menu?

Regards,
A.J.

Andrew
Thank you for your comments.

No the problem is not related to image rejection, switching off ‘Spur Reject’ has no effect, still reading 3dB low.

The Spike measurement receiver utility (MRU) fails to synchronise to zero with white noise always reading a few dBs negative. The version 2.18 software in contrast synchronises fine with white noise.

I did a test using the Spike MRU where I inputted a pure sign wave, it synchronised fine and I then added a 6dB attenuator and got a reading of -5.981dB. So the Spike MRU works fine with a sign wave but not with noise even after you correct for the synchronisation offset.

I notice the Spike MRU lacks the options available in the old software e.g. bandwidth adjustment,FM and AM detection and gives unstable readings with noise suggesting it’s using a very narrow bandwidth. Is this the problem?

Richard

Hi Richard,

I apologize, I misread your original question. The MRU functions differently in Spike than it does in 2.18. In Spike the Measuring Receiver utility in the file menu is used for measuring generators/attenuators/etc. and as you guessed, requires a CW signal and measures at a very narrow bandwidth. The audio modulation analysis portion of the 2.18 MRU was moved to zero-span in the Spike software. I’m not familiar with the noise figure measurements you are making. Email support at signalhound dot com with your question and Justin will be able to answer your question early next week.

Regards,
A.J.

Andrew
Thank you for the reply.

I did try the audio modulation and zero span options but could not find away of averaging noise over a suitable period.

To measure a noise figure under the Y factor method you need to measure a change in the noise floor when you switch on a calibrated noise source. I use a 5dB excess noise reference source so I am looking for about a 6dB change with a low noise amplifier and less with a noisy one. Spike gives the expected result using an ordinary sweep with suitable bandwidth but it is not as easy to use as the measurement receiver utility in version 2.18 software which reports the change directly.

Has anybody else used a SA44B to measure noise figures and if so how?

Richard

Hi Richard,
I’ve been successfully measuring noise figure with the SA44. I am using the Y-factor method; I have 30Hz RBW and 3Hz VBW which gives about 0.5dB of noise on the spectrum analyzer screen. I take a trace of noise with diode off and on and estimate the average noise either by putting my cursor in the middle of the noise or exporting the file to excel and averaging. I get values which seem to be sensible – NF of 1.2 for a device with a noise figure of 1.1.

What would be helpful, though, is an on-screen average line, or even a line that you can move to the centre of the noise. I notice that the cursor position reading seems to be about 0.6dB out.

Hi all,
i just tried to measure the Noise Figure of SA44b at its signal input. The result is about 40 dB at 1100 MHz.

The double output power method is used, not an Y-factor one; the spectrum analyzer tests a noise signal from external generator (Rohde&Schwarz SMBV100A), the level of signal at the input is known and adjusted to get the 3 dB higher result in comparison with an unperturbed noise floor when RF signal is off. The software is Spike 3.2.0; the averaging detector, power units, averaging trace were switched on. The results for power were read with enabled “Channel power”, bandwidth – 1 MHz.The generator gives a noise in 10 MHz band. The 3 dB rise in analyzer readouts was at the input power about -73 dBm/1 MHz; that gives NF=174dBm/Hz-10*log(1MHz)-73dBm/1MHz=41 dB.

The same procedure gives NF almost 20 dB for Rohde&Schwarz FSW13.

Regards, Dimitry, UA3AVR.

The best NF at SA44B input I have got is about 10 dB for 1000 MHz and somewhat better for 145 MHz (8.5 dB). The double power method was used too. The LabVIEW tools were used (not the Spike software), special procedures of noise measurements were composed, needed number of samples for predefined averaging accuracy was calculated a priori. The reference level set was -60 dB for best results received.

Unfortunately, an attempt to set attenuation or to switch on the preamp by LabVIEW function failed.

Regards,
Dimitry, UA3AVR.

Hi,
Iam trying to do noise figure meter using spike software and signal hound SDR (SA44B).my problem is SDR is detecting by the software but not connecting, why? and also please say the complete procedure to do noise figure meter

Thanks in advance,
likhil

[Author]

Posts