noise figure measurement

The total noise at the measuring instrument has to be insignificant compared to the total noise of the output of the IF, or you have to compensate for the instrument's total noise. I'm not sure why the bandwidth criterion -- are you sure that the app note in question wasn't referring to some specific setup, which is different from what you're doing?

Hello,

As your IF is 45 MHz, you can't use your oscilloscope at BW =3D 20 MHz setting.

Make sure the output of your IF doesn't contain out of band signals. If there is an amplifier after the IF filter, you may need to add filtering (between IF output and oscilloscope) to avoid that the out of band noise from that amplifier reaches your oscilloscope. Put the filter right at the input of the oscilloscope (if possible).

When you connect the IF amplifier to your oscilloscope (without RF input power), your power reading on the oscilloscope should increase >

As IF/CF ratio is small, the output of the IF (without RF input!) is a sinusoidal signal with changing amplitude and phase. This means when you use single sweep, you should see a clean sinusoidal signal (with changing amplitude depending on time base setting).

Now apply RF power and increase it to +3 dB reading on the oscilloscope. You have to divide by the IF filter bandwidth to convert the power setting of the source to power spectral density. Of course if you use dBs, you need to subtract 10log(IFBW).

Make sure that you use sufficient time base length to get stable readings from the oscilloscope.

I hope this helps you a bit.

Wim PA3DJS

Hi

Tanks for your help.

Tim, as far as I can tell, that was a general statement about noise measurement, which I don't understand. I quote: "A general assumption when performing noise measurements is that the device to be tested has an amplitude vs freq characteristic that is constant over the measurement bandwidth. This means that noise measurement bandwidth should be less than the device bandwidth. When this is not the case, an error will be introduced". Then they cite a article that I was unable to ge: "Slater, Carla; Spectrum-Analyzer- Based System Simplifies Noise Figure Measurement; RF Design, Dec.,

Wim, I'm afraid I was not clear. I'm a physicist, so maybe I'm not using the correct jargon....

I'm measuring a module that performs quadrature detection. The chain is as follows: era-5 -> Variab. Attn -> era-5 -> IF filter -> era-5 -> PSD -> output filter The input is IF@45MHz and the output is at baseband. The BW of the module is 4MHz. That's why 20MHz BW on the scope will do (I think!). I get output power around -35dBm. A the output I only have the LO harmonics. The reading on the scope increases 30 dB when I connect the module. I'm using the procedure you mention. Can you point me any faults in this setup?

Tanks

Wimpie was under the impression (as was I, after reading his response and re-reading your post) that you were connecting your scope straight to a 45MHz IF with a 4MHz bandwidth. That wouldn't work, because your scope would be filtering that 45MHz signal pretty heavily.

The app note that you're quoting is more explaining when and why you _shouldn't_ use a measurement device with a narrower bandwidth than the thing you want to measure, although it isn't terribly clear about this.

In your case, if you're getting significant LO energy in the signal, then you should filter it before it gets into the scope. In fact, I would expect that your block marked "output filter" would do just exactly that, and I'm kind of wondering why it's falling down on the job. At any rate, if you're going to use the scope to do noise measurements with, you do want it to be measuring the intended signal out of the system, and just the intended signal out of the system.

Hello Jos=E9,

Regarding the 20 MHz BW, fully clear!

Do you have two outputs (I and Q), or do you have a single baseband output that goes from 0 to 4 MHz? What is the "PSD" block?

If you have two seperate outputs, you can use 1 output, but in that case the image noise also appears in your output signal (as two IF frequencies will produce same baseband output frequency).

Except for the shape of the baseband noise, the measuring procedure doesn't change. If you are unsure about the IF to baseband filter shape, you may determine it by "sweeping" the IF input frequency. Based on this you can calculate the noise bandwidth of the complete IF to baseband chain. This also allows you to measure systems without a brick wall frequency response.

I don't understand the section regarding the LO harmonics and the readings. If the baseband processor removes the harmonics, you should do this also before they reach the oscilloscope (to avoid the out-of- band noise issue)

With kind regards,

Wim PA3DJS

I'm not sure exactly how you are measureing the noise, but here's how I parse the statement.

We assume it's white noise.

You need to know your bandwidth, If the device is narrow band, (compared to the measuring insturment) then you use it's bandwidth, otherwise use the bandwidth of the measurment device.

Assume white noise and know your bandwidth.

George H.