EMZER is a company based in Barcelona, Spain which designs and develops innovative EMI receiver with focus on conducted emissions.
<H2> About modal measurements – news approach for conducted emissions
First of all, let’s do a quick reminder on how typical EMC measurements are done. The standard approach is that you measure only one channel at a time whether it’s line or neutral. If it’s three-phase you measure line one, line two, line three and then neutral.
If you use an analog receiver it takes a life to make the measurements, but most of all, it’s not possible to determine the source of the noise. This means that you can measure the total emission of one channel, which can either pass or not. But this emission is composed of common and differential mode and if you don’t have the information on how much of this noise is common and which part differential and in which frequency ranges, it is not possible to correctly design the power line filter.
That’s because the power line filter is composed of components, mainly chokes and capacitors whose duty is to reduce a particular modal emission.
All of this, of course, makes it cost consuming and this is why we proposed a new approach, which is design your filter by analyzing the modal measurements.
<H3> Why Modal Measurements are important
Modal measurements are composed of common and differential mode, if you look at the following two images, you will identify the differential mode as the noise that goes through one conductor and comes back in the other conductor in the opposite direction, while the common mode goes in the same direction for both conductors and it recloses through ground or through stray capacitances.
This is very important for filter design, because each component of the filter acts on each mode and by exactly knowing in which frequency ranges there is the noise, you can pick your components.
Of course, one option is to use the big filters, for example the box filters, but the problem is that, for instance the box might be too big or too expensive or simply doesn’t work so you will need to increase the size of your EUT or take something out to reduce performances somehow.
<H3> EMSCOPE Modal Measurements Capabilities
Now, let’s do a quick reminder on the effect of each component on the modal noises: the common mode choke acts on the common mode noise while the capacitors act more on differential mode, especially the CX capacitor placed between line and neutral.
Our EMSCOPE is composed of two receivers with three detector each which always measures simultaneously the two lines. This is why it is possible to decompose the emission into modal noises.
<H2> EMSCOPE Demos Set-Up
Now, we will present the demo set-up, in the left part, we have our EUT which is a switching mode power supply. This EUT goes through a filter selection box in which we have four positions.
The first one (nº4) goes directly to the receiver, so we measure the emission of the EUT without anything else connected.
Then we have three positions where we placed three different filters. We have a really big filter in the box, a smaller filter, and a PCB where we designed our own filter, thanks to the information that the EMSCOPE can give us.
<H2> EMSCOPE’s Software
This software is embedded inside the instrument. All updates are free, this rarely happens in EMC, you usually have to pay a yearly fee maintenance, but this doesn’t happen with the software inside the EMSCOPE.
As any other software, you can go in device configuration and set all your parameters or update advice for free.
On the left you see a menu, top to bottom, you can use it to configure all the measurement that you have to do. For instance, the frequency band it’s already here so we can just select it and modify still the start and stop frequency.
So, according to the CISPR, we don’t even have to know that the resolution band is 9kH, the software does it for use. We can then set the dual time and sweep mode.
In “amplitude”, you can take into consideration external losses and compensate them, since we are using the internal LISN, we are already compensating it.
Now, let’s see how this trace configuration works. As you can see, if you look at the clear/write option, it has a different color, it means that it is selected.
When we go down, you can see we are also selecting EMI measurement; in this case, we are measuring the line with respect to the ground with the quasi-peak detector.
All this information can be found in the tab’s top left corner, in this case, T1 L-G QPK CLR means: Trace one, Line, quasi-peak, and we are clearing. The blue graphic is our measurement.
Now, next to the top of the tab, we have a small plus sign. When you click on it, it will add another tab, just like would happen with Google. The difference is that you will still see, in the same page, the measurement that you were doing before. This is used both for comparison or for multiple measurements at the same time.
For instance, you would see that here, the software is opening another tab, and in red color we are measuring the line with the peak detector, and we are measuring. Of course, clear means that every second (we have set the dual time at one second) it’s measuring. It’s an effective receiver so the measure it’s really fast. In one second, we can measure from 150kHz to 30 MHz.
Keep in mind that, if you have an analog receiver, you have to do the measurement with peak detector, scan all the band, and then go back with the quasi-peak detector. This doesn’t have to happen with EMSCOPE.
Now, we have trace one and trace two with the two colors, but remember that we can still open other tabs because, as you can see in the following picture, the receiver has two channels, which means that we are measuring the line, the neutral and we are also measuring them simultaneously and with three detectors each.
For instance, we could open another tab, click the neutral and put the quasi-peak detector.
We can even open another tab, add the neutral and the average detector, so in one second, we have the measure on the line neutral, with peak, with quasi-peak and average detector.
Now, going back to the first two tabs, as you can see, we are clearly above the limit. We can also check it with our list of maximum emissions. It will mark all of the emissions, all the points, but we can clearly see that we are failing the EMC test.
At this point, the common thing to do is to use a power line filter. To do so, the first thing we are going to do is save the measurements we have gathered in case we would want to see them later.
Now, we are going to show you what happens when you do what is commonly done in the wrong way in EMC: put a big filter and see if it works.
So, when we do put the big filter, we can see that we pass the EMC test, which means that the product is good to go. But it also may happen that the filter is too big and cannot be used or it’s too expensive or whatever problem you may have.
So, what is the correct way to pass conducted emissions tests?
First of all, we are going to go back to the product without filter and, as we were explaining before, perform modal measurements. To do so, we are just going to click in modal, and then click common mode.
So now we have that the trace one is measuring common mode with a quasi-peak detector.
Now, we are going to trace 2, and measure the differential mode with the quasi-peak detector.
As you can see, our emission is composed of both modes. Sometimes it can happen that we have one mode at different frequencies, so we would need different components in the filter. But for now, let’s say that at lower frequency we have a really high peak of common mode, and then differential mode.
Now that we have seen how to correctly test EMC, we will show you another thing the program allows us to do which is the freeze option. To do so, we will freeze both trace 1 and 2.
Next, we will open another tab to measure the differential mode with the quasi-peak detector, and, in another tab, we are going to measure the common mode with the quasi-peak detector.
To sum up, we have frozen two traces, of what the EUT is emitting, and then, we have opened two traces which are continuously measuring.
So, right now, if we’d change the filter and put the big filter, we would see it’s action on the both modes.
Now, if we went back to position one, what we’d see is difficult to understand, so we are using another EMSOCPE function, which allows us to hide traces.
So, for instance, we will hide the differential mode. You will see now why at the beginning we changed the colors to make the common and the differential mode different. As you can see in the image, the blue line is the frozen mode, so EUT without filter. And common mode, represented in red, means WUT with filter number one, the big filter.
We can see that we have a good reduction at high frequencies of common mode, but at lower frequencies not so much. We do have an important reduction but we always have to make sure that this measurement is done in a shielded environment to prevent unwanted couplings, for example.
We are not far from the limit, and measurements of common mode especially are highly affected by the setup and ground connection, so we will do some more adjustments.
To start, we will hide the common mode and show the differential mode.
Here, we have the differential mode in red, the one that is not filtered and, in blue, the filtered one. As you can see, this are really good performances of the filter for differential mode. It’s a really good capacitor.
Now, we are going to hide everything and put the second filter, the smaller one, on. It’s like the metallic box, only smaller, which means less space, and probably less expensive. To check it, we will still use the function to hide the noise.
Let’s compare common modes, so let’s hide the differential modes, and see the difference between the common modes with and without the filter.
We still see a better reduction than with the first filter. As you can see, it has reduced 15 dB at the frequency of 170 kHz which was the point that was giving us problems.
But, if we check the differential mode, we will see something unexpected. The differential mode is higher than the one before, when it was really low, but by changing the filter, it has increased again.
The conclusion, is that these two filters are basically giving the same attenuation but we are not really improving anything. If we were to do a measurement of EMI, (selecting clear/write and putting the line with the QPK detector), we could say that it passes. Remember that this is with the small filter.
Now, if we freeze this measurement, open another tab, and measure the same with the quasi-peak and the big filter, we can see the difference in the overall emissions that these two filters give.
So, we see, apart than the peak at 200 kHz, it’s practically the same, the same happens at 400 700 kHz where it’s giving more attenuation the red one but still, not so relevant.
What we should do here is design a filter using the Wurth filter kit and really solve the problem.
So, for instance, we go in clear/ write and we do the modal measurements with the common mode.
Then, we open another tab for the differential mode with the quasi-peak, and we get the filter out of the configuration again, so that we have the total noise.
Next, we freeze the common mode again, and open two more tabs to measure the same, and we put on the 3rd filter, the PCB, our designed filter.
What we see is that, if we hide and compare one to one, we have a huge reduction in common mode at the lower frequency, we could say a 60 dB reduction.
And also, in differential mode we have a huge reduction too.
We have designed a smaller filter with less components, but with better performances. Let’s check it by measuring it again with the EMI measurements, the quasi-peak detector, and in another tab, we are going to measure it with the average.
As you can see, the results are okay; we have a 40 dB difference, and we are sure that if we were to go to an EMC lab for the final CE stamp on the product it would pass with no surprises. Our filter is perfect for this power supply.
As you have seen, we have designed the most optimized filter both for size, and also for price.
For instance, if your goal is to use the smallest filter in size you can use the modal decomposition to achieve this result.
On the other hand, if your goal is to minimize price, you could settle for components that aren’t as good in performances, but that are cheaper.
The important part is that, when you go to the EMC lab for the final stamp you get a pass. Failing this test means a great loss of time and money and, of course, more time to solve the problem you have. Because if you try to solve an EMC problem without knowing common and differential mode and do a real time comparison, you will enter in a very long loop of trial and error.
Finally, another thing that this software allows you to do is a pdf report which proves your work with your optimized filter and with the results.
Of course, with this software you can do many more things, such as waterfall function. With it, in one click you see the waterfall function and the time frequency diagram, but still, the really major point of EMSCOPE and the advantages it brings, are focused on model noises which are the only components that you need to look at when you design a filter and especially if we wanted to make it small optimized and with good performances.