Mains LISN¶

This article is about building a mains Line-Impedance-Stabilization-Network (LISN).

Similar to the LV dc-LISN for automotive stuff, this LISN is also "just" a bunch of inductors, capacitor and resistors. It's main purpose is to present a 50Ω impedance to the Equipment-Under-Test (EUT), and to present emissions on the mains to a measurement device. Ususally, you'll add some attenuation to this signal.

In the fundamentals section, I give an overview about conducted emission testing. For the purpos of better understanding the LISN, let me just explain why you're reading about unsymmetric voltages when doing measurements on a LISN. Fundamentally, you can split every signal into some common-mode and some differential-mode component.

Every common-mode signal on the mains, forms a current loop which consists of both, the L and N conductor, the LISN and some ground connection. So this portion of the signal causes a current which has same polarity on L and N.

The differential mode, in contrast, is portion which forces currents on L and N with oposite polarty. Hence, these currents would form a magnetic fields around L and N which cancel each other. Usually, the common-mode-currents cause interference, since the fields produced by these currents would not vanish.

Some Schematics and Insights...¶

First of all, let's look into the most basic LISN.

The first example is the 50µH/50Ω LISN in MP-4 of the FCC. Why? Well, you can (and should!) download this document for free. It's old (1986 vintage), but is give you quite some information on how you could do the measurement and what would be the caveats when setting things up.

Basic LISN

This is identical with the LISN 50Ω/50µH LISN (not the different spelling) in CISPR 16-1-2. For measurements from 150kHz-30MHz you could actually get away with something that simple!

Now things start getting a little more difficult.

There are 3 major parameters for your LISN:

the impedance
the isolation
the Voltage-Division-Factor (VDF)

Impedance¶

The first one, the impedance is pretty straigt forward. For the 50Ω/50µH, the impedance transistions from basically 0 (the mains is a short, right?) to the input impedance of your measurement device. So it's 0-50Ω, where the transition is defined by the size of the capacitor.

Easy, right?

Not so much, since there's somthing called isolation...

Isolation¶

Now we do not want to have the mains impedance mess with the impedance we provide to the EUT. Further, we do not want to have noise from the mains contributing to our measurements. Therefor, CISPR16-1-2 suggest -40dB of isolation would be nice.

Now let's fire up SPICE, and see what we get...

The calibration setup we want to simulate is simle:

Basic LISN

However, the results are a little surprising: Basic LISN

As you can see, we're not going to meet the required -40dB. At the first glance, this is pretty surprising. We have som standardized schematics, but they're not quite sufficient to make a a compliant measurement tool.

Well, yes, and no :)

The problem is, we're building some resonant structur (ie. the 50µH inductor and the 1µF capacitor). They have a resonance frequency of about 70kHz. We need to lower that in order to reach the required isolation. If you're doing your homework, you'll see that starting from 2.2µF, you're golden.

BTW: There's a whole section on the 50µH inductor in the standard...

Voltage-Division-Factor¶

No to something completely different...

The voltage division factor gives you a correction to obtain the voltage level on the EUT port. If you look at the schematic, you see, that there's just a 100nF capacitor between the mains and our measurement port. Hence you can see this a voltage divider. At 150kHz, the 100nF capacitor already has a very low impedance compared to the 50Ω input impedance of you spectrum analyzer. Hence, traditionaly, you don't really bother about the VDF... it's 0dB + whatever your transient limiting device gives you.

As we see later, this changes a lot if we're aproaching the 9-150kHz region.

going 9-150k...¶

After briefly looking at measurements above 150kHz, let's focus on the upcomming requirements below 150kHz.

First of all, you need to have a 50Ω/(50µH + 5Ω) LISN. This means nothing more than having a defined 5Ω impedance for low frequencies. How is this done? Glad you asked...

A 9k-30MHz LISN

As you can see CISPR16-1-2 suggest adding another inductor, another capacitor, and, most importantly, some resistors.

The basic idea is that the 250µH inductor decouples the EUT from the mains pretty efficiently. Further, the 8µF capacitor is already low an impedance at 9kHz. Hence, you'll get 5Ω impedance at the EUT port. Increasing the coupling capacitor to 250nF is just to have the VDF relatively flat down to the lower frequency bound. You could use 100nF as well.

This could be interesting, if eg. you're intending to measure SMPS where you already predict relatively high emissions <150kHz. That ways, those emissions get attenuated while you still have good sensitivity for the higher frequency emissions. Especially, if you're planing to do conducted emission testing using an SDR, this could be an interesting option for you.

caveats¶

So we have again a trivial looking circuit, right? Now what's you expectations for the basic performance parameters we have discussed above? Let's start with the impedance.

Impedance and Isolation¶

Impedance with mains terminals open and short to ground

First, some words to the limit curve. Definitly, you can see some "distortion" at 150kHz. This is due to the combination of the limit lines below, and above 150kHz. At fundamental circuit changes, so the limit values changes as well.

Now for the most obvious: We exceed the limits right at 9kHz. You can also nicely see that this is due to a resonance at around 3-4kHz. If you short the mains, you quickly see, that this problem is caused by the 8µF capacitor and the 250µH inductor. Just do the math (\(|Z|=\frac{1}{2 \pi\sqrt{L C}}\))... you're going to see exactly that resonance at 3.5kHz.

Ok, "nice", what can we do about that? Well, we can't really change the capacitor. This would change the impedance as well. And changing (ie. at least double the inductance) is unfeasible as well.

What's left are the resistor values. I leave out some possible values, since this is kind of a curcial "selling point" of my LISN kit... Btw, yes, I sell a DIY Kit to make your own CISPR 16-1-2 compliant LISN. However, fireing up some SPICE solver should allow you to figure out some possible values within minutes.

The next thing on our agenda is the isolation.

Isolation of a LISN built from the component values in the standard

While this is already way better than the very basic circuit, we still do not get the performance we would have hoped for. In addition, increasing capacitance on the main terminals would create a situation more like a short. Hence, we would get into the unlucky situation with too high a input impedance as shown above.

Now, how to fix this? Again, the answer is in the resistor values. I tend to get a 3dB marin in the isolation, and I try to nail the impedance (ie. get right at the center of the tolerance band). After spending some minuts with resistor values and SPICE, you'll for sure get some good values that fit both, the impedance and the isolation requirements.

My values give me the following results:

Input impedance of my final LISN Isolation of my final LISN Voltage-Division-Factor of my final LISN

For the VDF it may be interesting to increase the input capacitance to, let's say, 1µF. However, getting safety capacitors that big is quite an issue.

That's it for this topic. I put some more information on the measurement uncertainty and the impact of isolation property in a separate article in the "Fundamentals" section. Here, you'll get some math to better understand the requirements in eg. CISPT 16-4-2.

Last updated 2023-02-06