Help: measuring noise flr/ density and noise output of a buck using spectrum analyzer

[alimjoco]

I just want to know the settings used in spectrum analyzer. In my case, I used first to measure the spectral peak. So peak detect is ON,while I lower the RBW of the analyzer, in this case I will see the switching frequency and including all its harmonics(especially low level because of the lower RBW)..Is that correct?

Second, how can I measure the exact noise floor of the converter? Is it that i just turn OFF the peak detect and all will be displayed is random noise?

3rd, how can I accurately calculate the noise spectral density? does it depends on the RBW or the frequency span used???

 

[RCinFLA]

Re: Help: measuring noise flr/ density and noise output of a buck using spectrum ****

Depends on type of spectrum analyzer. Low frequency spectrum analyzer nowdays are often FFT design. For a sweeping spectrum analyzer you have to understand that its sweep and frequency point it happens to be looking at is likely asyncronous to spurious frequency created by a switching power supply. Peak detection can be used to capture these asyncronous events.

Noise floor measurement is based on resolution bandwidth. dB/Hz is -10log (Rbw) Understand the noise floor of the spectrum analyzer and stay at least 10 db above it. Spectrum analyzer have pretty poor noise figure, typically 20 db unless they have a front end low noise amp.

 

[alimjoco]

Re: Help: measuring noise flr/ density and noise output of a buck using spectrum ****

Thanks.. So how can i measure the effective noise floor of the switcher? Will it be on non peak detect mode?Then compare it to a terminated to ground input of spectrum analyzer noise floor? Usually in peak detect , negative peaks are eliminated so i guess it cannot measure accurate noise floor,,,how about the resolution bandwidth?will i set to a very narrow BW for more accurate results or I just leave as auto BW coupling...

 

[FvM]

Re: Help: measuring noise flr/ density and noise output of a buck using spectrum ****

I doubt, that the quantities noise spectral density and noise floor are of much use to characterize a switch mode supply.

 

[alimjoco]

Re: Help: measuring noise flr/ density and noise output of a buck using spectrum ****

Ok thanks..So you mean to that a more appropriate mean to characterize noise of a switching power supply is to measure peak of the spectral energy associated on the switching including it's harmonics and if there are also possible wide band peak on the higher frequency range. Is that it? Thanks!!!!

 

[RCinFLA]

Re: Help: measuring noise flr/ density and noise output of a buck using spectrum ****

I doubt, that the quantities noise spectral density and noise floor are of much use to characterize a switch mode supply.

They are if you are using them to power a high dynamic range audio circuit or RF transceiver circuitry.

Using switching power supply is only way to get reduced power drain when there are multiple supply voltage requirements for a system that runs on critical battery power, like a cellphone application. Most cellphones have four or five switching power supplies to supply various circuits.

 

[alimjoco]

Re: Help: measuring noise flr/ density and noise output of a buck using spectrum ****

Yes I'm planning to power up ADC using switchers but I need to start of on how to correctly characterize noise floor, noise density and noise ripple output of a switcher. ADC is very noise sensitive and it can even digitized noise in its supply input.

Anyone knows the correct basic set-up on measuring the noise floor or noise density of a switcher?Idea perhaps on measuring it using a spectrum analyzer?

 

[FvM]

Re: Help: measuring noise flr/ density and noise output of a buck using spectrum ****

They are if you are using them to power a high dynamic range audio circuit or RF transceiver circuitry.

So at which circuit node you would measure these quantities? In my understanding, they are signal parameters. But you aren't referring to a signal rather than suppply rails connected to the switch mode supply. So what I would do is to measure the magnitude of the dominant spectral components, e.g. using the peak table feature of the analyzer. The load impedance and DC current must be well defined, of course.

If you refer to industry standard SMPS modules, you won't find much information except for a "Vpp ripple and noise" specification with a corresponding filter bandwidth, similar to what you can easily determine with an oscilloscope. If you interrogate the real waveform, the ripple is typically comprised of multiple high order harmonics of the switcher frequency, or possibly a kind of ringing.

But it's no problem to reduce this ripple by several orders of magnitude by LC filters until it reaches the supply pins of a critical component.

 

[alimjoco]

Re: Help: measuring noise flr/ density and noise output of a buck using spectrum ****

Thanks this is helpful information. Measuring Pk to Pk of spectral components should have an enabled peak detect mode using analyzer.
My switching frequency is around 620Khz. I guess with a 20Mhz BW set in oscilloscope is appropriate for measuring Vpk to pk for time domain.

I am also thinking to make the set up in scope as full BW but more noise will be introduced thus degrading my measuremnts.

But is there anyone who attempt to measure the noise floor using a spectrum analyzer?

 

[FvM]

Re: Help: measuring noise flr/ density and noise output of a buck using spectrum ****

But is there anyone who attempt to measure the noise floor using a spectrum analyzer?

Selection of measurement instruments should be driven by the expectable signal or interference generated by the respective device under test. A switch mode converter can be expected to have a fundamental switching frequency and harmonics, possibly of high order, e.g. oscillations from rectifier diode reverse recovery. A spectrum analyzer can be useful to get an overview about the emitted frequency range, but I don't think that it will be often used to characterize the device. The main problem is, that the absolute magnitude will strongly depend on small external circuit variations, e.g. bypass capacitors or layout related changes.

Unless, the inverter falls into an unstable operation mode, I won't expect a noise floor alike spectrum.

A spectrum analyzer can be even more important to identify the origin of interfering signals observed in the application circuit.

 

[biff44]

Re: Help: measuring noise flr/ density and noise output of a buck using spectrum ****

a "buck" means a switching dc power supply??? I hope you realize how easy it would be to blow up your spectrum analyzer to a DC voltage source! I would us a DC block and then a 30 dB coaxial pad before the spectrum analyzer!

 

[alimjoco]

Re: Help: measuring noise flr/ density and noise output of a buck using spectrum ****

Yes, I used HP 10240 DC blocking capacitor between 50 ohm input coupled Spectrum to DUT Vout. So i guess only the AC ripple will be seen by the spectrum

 

[alimjoco]

Re: Help: measuring noise flr/ density and noise output of a buck using spectrum ****

Hi biff44? what is the 3dB coaxial pad used for? Can it be only DC blocker?