Spectrum Analyzer and noise

Hi ,

I am doing a test on a switching power supply. You can view the block diagram of the test at the following link

formatting link

I am coupling common mode noise to the AC input lines to see its effect at the DC output of the power supply. I want to see how much common mode is being rejected by the switching power supply. The frequency range of the noise is from 10KHz to 80MHz.

The thing is that I am not sure that the spectrum analyzer is hooked up to the system correctly. It has the input impedance of 50 ohm. The 3db frequency is 318.3KHz.

Is this right way to use spectrum analyzer? or are there other ways to measure common mode.

mel

Reply to
walravenmelissa
Loading thread data ...

gram of the test at the following link

t the DC output of the power supply. I want to see how much common mode is being rejected by the switching power supply. The frequency range of the no ise is from 10KHz to 80MHz.

o the system correctly. It has the input impedance of 50 ohm. The 3db frequ ency is 318.3KHz.

asure common mode.

If you are using an RF spectrum analyzer I would instead go with an oscillo scope which has an FFT function. In this way you can load the power supply with the load you want (The 50 ohm load does not seem to be good) and then measure and compare the FFT of the input noise and the output noise with h igh impedance input on the oscope

Reply to
djlocher56

For USA, you need a LISN (Line Impedance Stabilization Network): For other countries, it's an Impedance Stabilization Network or an AMN (Artificial Mains Network). All these will do both common and differential mode conducted emissions.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

Likewise, check out the relevant standards. IEC 61000-4-6 is the CE requirement for conducted susceptibility. Typically, it's done 150kHz to

80MHz, and only performance (go/no-go) of the product is recorded, not the RF emanations at the far end (which one would reasonably expect to consist of the device's normal noise output summed with whatever gets coupled from the input, which isn't a useful properly of the device).

Tim

--
Seven Transistor Labs 
Electrical Engineering Consultation 
 Click to see the full signature
Reply to
Tim Williams

I am using an RF amplifier and function generator to insert the noise on the AC lines. The noise amplitude in 10Vrms with frequency range from 10KHz to 80MHz. LISN is irrelevant here because LISN measure the common mode noise inserted by the device on the AC line.

In my case it is opposite, I am inserting noise on to the AC line.

I am using the following function generator with an amplifier

formatting link

Can I loose isolation between the input and the output, if I connect one probe of the oscilloscope at the AC input and the other at the DC output of the switching power supply. The switching power supply is based on fly back transformer.

Plus how can I tell that I can measure the input on the AC lines with out damaging the oscilloscope.

mel

Reply to
walravenmelissa

I think you are not really measuring the common-mode at the output. Still, the load probably doesn't respond to common-mode, so what you are measuring might be more relevant.

With your present configuration, you might have to be very careful with how the grounding is done. Any inductance in the ground at the -ve output of the SMPS will likely result in currents flowing through the ground of the spectrum analyzer which might or might not be a problem.

If you really wanted to measure the common-mode at the output of the SMPS, I think you might need to remove the ground from the -ve output terminal, and connect a capacitor from each side of the SMPS to a port of a resistive splitter, with the third port going to the input of the spectrum analyser.

Often in this sort of testing it is a good idea to put a "limiter" right before the input of the spectrum analyzer, which makes it more difficult to accidentally blow up the spectrum analyzer, e.g. with large surges resulting from fast edges when the power is switched on, etc. You can buy limiters, but for under 100MHz it is not difficult to make one from some surface-mount resistors and fast diodes. The one I made is basically a 10dB attenuator with three resistors in a T-configuration. There are two diodes in parallel with the shunt resistor in the middle of the T attenuator, one diode in each direction. I made the limiter as a microstrip or coplanar waveguide structure, on a piece of double-sided copper-clad FR4, with solid ground on one side, and a coplanar waveguide trace, cut with a scalpel on the other side, and plenty of wire-links through the board in the ground areas. SMA launches were attached to each end.

Chris

Reply to
Chris Jones

Common-mode injection requires equal amplitudes being applied between all input ports (in common) and ground. In order to do this, you'd need to stabilize the source line impedances using an LISN for each input line, with respect to ground~earth.

The injection of voltage would use simultaneous capacitive coupling of equal value between the known source impedance grounded generator and each line. These capacitors are situated much as emc filtering 'Y' capacitors might be deployed, and their size and ratings are similarly considered, for the test bandwidth being examined.

Isolated current injection places all input lines as mutifilar secondaries of a high turns ratio transformer.

The intended connections from DUT chassis to ground should also be present ~ as intended in the end-use application. Whether this includes an output terminal will also require knowledge of end-use.

A common immunity/susceptibility standard is IEC61000-4, all chapters. There are similar standards for most application environments, including telecom, aerospace, automotive and military.

RL

Reply to
legg

Hi,

How about using an oscilloscope to measure input and output and compare the FFTs as suggested by snipped-for-privacy@511nyrideshare.org?

mel

Reply to
walravenmelissa

I don't have much experience with the FFT mode of oscilloscopes. I imagine that a real spectrum analyser would have an advantage if you were looking for a very small signal at one frequency in the presence of a very strong signal at a different frequency. Most oscilloscopes only have 8 bit ADCs which means that their dynamic range is quite limited. Since you are probably only looking at one signal at a time, that would not matter for your measurement, so probably you could use an oscilloscope.

I guess the decision might also be determined by any standard that you are testing to. In many cases these call for a specific measurement bandwidth and detector response, which means you would need a specifically designed instrument if you need to replicate the actual compliance test. Often this is not really necessary if you are just comparing things.

Chris

Reply to
Chris Jones

You are injecting CM on the input. It does not make sense to monitor CM on the output. You need to monitor the DM noise on the DC output (that's what you care about anyway)

Hook up a tracking spectrum analyzer. Apply the signal to the input and capacitive couple the output DC to the second channel. That way you lock the input signal to output signal

Cheers

Klaus

Reply to
Klaus Kragelund

Hi,

Would you explain that why does it not make sense to inject common mode noise to the input and to monitor common mode noise at the DC output? I couple the output with the capacitor as shown in the following link

formatting link

mel

Reply to
walravenmelissa

MiniCircuits offers standard splitters and combiners that are useful in separating differential and common mode components from a mixed source, over a defined bandwidth.

RL

Reply to
legg

The FFT can only display the components shown on the scope and cannot differentiate between common-mode and differential mode components. There is no determination of relative phase or coherence between components displayed by two FFT plots.

The oscilloscope/FFT technique is not inherently different from the use of a spectrum analyser. Unfortunately most scopes are not provided with sufficient low-noise signal handling capability, nor are they of sufficient bandwidth to provide the resolution required at the higher end of the conducted spectrum.

It can provide useful information at the lowest end, where many spectrum analysers are not designed to function. Few EMC standards are interested in frequencies below ~150KHz, or in intermittent behavior, where immunity performance artifacts are most often experienced.

RL

Reply to
legg

You probably want to use transformers for isolation (maybe a set of 'em, for that range of frequencies). So, amplifier drives primary of "noise"-injection transformer, and secondary is in series with your AC source. Either pick off the signal at the amplifier terminal (which is transformer-isolated from the AC), or use another transformer (with negligible current) to sense the AC.

Reply to
whit3rd

I was thinking what kind of transformer should I use between the output of the power supply and the spectrum analyzer.

Reply to
walravenmelissa

Why do I feel that there will be a nice spectrum analyzer with a blown first mixer on ebay soon?

Reply to
Tom Miller

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.