5uH Aerospace LISN: How dumb would I be to "throw one together"?

[TimNJ]

September 7, 2020 Update: Please this thread for final design files and discussion: https://www.eevblog.com/forum/projects/lisn-for-rtcado-160-aerospace-emc/

Hi all,

I've been following this lovely thread for a long time: https://www.eevblog.com/forum/projects/5uh-lisn-for-spectrum-analyzer-emcemi-work/

I have two projects that need to comply with RTCA DO-160G (Aerospace) EMC. The end-game is 3rd party testing with "correct" equipment. Like other standards for vehicles, the LISN defined is the 5uH type. I have the option to rent monthly at $600/month for a pair of LISNs. To purchase, it's about $2000-3000. Purchasing is out of the question for the moment. Obviously, the commercial LISNs are already characterized nicely, and since there's already plenty of things that can go wrong with in-house EMC pre-testing, it's nice to keep unknown variables at a minimum. At the same time, I still have a hard time justifying the cost of the commercial LISNs.

For the sake of ballpark pre-compliance testing, how dumb would be to try to DIY the LISN with the simplest possible design? (FYI, I need to pass >12A DC (for project #1) and >3A AC (for project #2). I have a 1.6GHz SA with tracking generator to help characterize. I see most people are using VNAs. I'm not sure the difference in this case.)

-----

A few questions to help me understand whether this is practical or not:

1. The Tekbox 5uH design uses a "segmented" or "staggered" inductor design..with taps that go to RC networks. What gives? Is this to help flatten/control the frequency response? What's the effect of breaking down into segments as opposed to one big "lumped' inductor?

2. Some designs have small value series resistors in line with the 1-10uF capacitors. Same idea as above? Add some damping to the frequency response/prevent oscillation?

3. Generally speaking, are you more likely to have issues with relatively large air-coil inductors or smaller ferrite/powdered core inductors? Seems like air-core is generally preferred, and probably easier to reach at only 5uH, as opposed to 50uH.

4. For designs with series inductors, some have parallel resistors across each. What's the rationale? (Jay_Diddy_B shows 330R across his.)

5. The measurement is taken across a 5K resistor but with a 50R  receiver, so the RC filter is really 0.1uF + (50R || 5K = 49.5R). I suppose the value of that 5K resistor barely even matters then. since it's always paralleled with 50R. (Some LISNs use 1K.) I was planning on copying the HP 11947A transient limiter...Is this generally a "universal design" that can be used with any 50R system? Obviously need to adjust for 10dB attenuation in the receiver.

I know these are pretty naive questions. I think it's obvious I'm still a novice in many of these areas. Maybe that's a sign that now's not the time to dabble in something I'm not too certain about, especially when I have a deadline to meet and customers to please.

Thanks,
Tim

[T3sl4co1l]

The answer to most of these, is: dampen resonances of the inductor.

Play around with this, for example; read the supporting information: http://hamwaves.com/antennas/inductance.html
It uses a helical waveguide model, and a bunch of corrections, to estimate the voltage and current through a single layer, wire solenoid.  Apparently this geometry is remarkably difficult to analyze!

If the calculation succeeds (there is some root-finding that can fail, particularly at small pitch angles), you get the complex impedance of the element -- resonant modes accounted for!

The first (parallel) resonant mode is fine, that's what we want -- a high impedance between DC and EUT sides.  What we don't want is the second (series) resonant mode, which acts in parallel with EUT, shunting signal and distorting the frequency response.

Knowing that standing waves are the culprit, we can employ resistors in strategic locations to dampen those modes.  The series resonance has an antinode in the middle (voltage peak), and nodes (current peak) at the ends, which is why we measure a low impedance at the ends, at that frequency.  A resistor from one end to the middle will act in parallel with that mode, dampening it.  The resistance acts in parallel with the voltage peak, and gets 1/4-wave-transformed into a series equivalent resistance at the terminal.  When R = Zo,* parallel R is transformed to series R, and that's simply the resistance you will measure at the port.

*For an ideal transmission line, Zo is simply Zo.  Helical waveguide however is dispersive (velocity varies with frequency), meaning the Zo varies for each resonance; the resonances also aren't harmonically related.  So it's not quite as simple as knowing the wire length of the coil (though that's close for the 1st mode, I think?), and it may be worth playing with the resonances in a calculator such as above, or testing real hardware.

[Note that helical waveguide can have quite high impedances, low kohms -- this makes them useful for bias tees for one, but also useful for delay lines in certain applications.

A vintage application was color TV sets: the impedance must be high (low kohms) to suit to the vacuum tubes used in early sets.  The delay is applied to the luma signal (which is simply detected directly from the radio signal), to "catch it up" with the chroma signal, which ends up delayed due to additional filtering and processing stages.  The required delay was about a microsecond.  A typical delay line was a phenolic tube, wound with fine wire, and lined with a strip of foil -- not a complete wrap-around foil lining, that would make a shorted turn, defeating the helical mode; just a narrow strip to give some ground reference for the travelling wave.]


We can also employ loading materials; if we have a lossy ferrite or powdered iron material, we can use it to both increase the inductance (reducing the number of turns required, potentially raising the resonant frequency) and dampen the resonance (by magnetically coupling material losses to the resonances).

The loss has to be appropriate, of course; a high-mu powdered iron will be too conductive, and actually increase the capacitance more than dampen modes.  A low-mu ferrite might have too high of a Q at these frequencies, and just not do much, or maybe make things worse.  It may take some trial and error to perfect.


I haven't done this specifically for solenoids, but I have plotted a few resonances on a 100% coverage, single layer toroid winding.  Here, the boundary condition forces the first mode to be a full wave, and it is series resonant.  Again, the resonant impedance is quite high, which makes it difficult to dampen at the terminals -- if we're using this toroid for a current transformer, we can't afford much burden resistance.  Maybe the resistor needs to be a few ohms for the application, but the correct value for damping would be 300 ohms, or a few kohm even -- highly impractical to dampen.   Knowing that it is a resonant mode, however, it can be shunted by simply distributing the applied current around the core.  Instead of using one loop in one place, use two loops, wired in parallel but positioned in quadrature (at 90 degrees to each other on the toroid).  Or hexature, etc.  (I was using a high-mu toroid, which is rather lossy at the 10s of MHz these modes were showing up, so the 4th mode was already very weak; I didn't try looking any higher.)


Anyway, what we're after, is knocking out series resonances (impedance dips at the EUT/RF port) and replacing them with resistances, hopefully resistances that are high enough not to worry about (i.e., R >> Zo).


If we have discrete inductors that manifest as lumped single RLC networks (i.e., effectively some simple equivalent parallel capacitance, without having ugly modes at higher frequencies), we can employ the same methods, applied to a lumped-constant circuit, without having to worry about waves and modes necessarily.  Here's a bias tee by Picotronics, flat to, as their name suggests, some picoseconds (many GHz):
https://www.seventransistorlabs.com/Images/Picotronics/MVC-349X.JPG
https://www.seventransistorlabs.com/Images/Picotronics/Schematic.JPG
It seems they opted for the parallel damping resistance instead.  I think that can save a little on impedance -- you don't have the lossy cap-to-ground loading it down.  Which one is best probably depends on the inductors used, and desired flatness.

Don't forget, too -- inductors have capacitance between their terminals, and to ground or free space; a 2-terminal model of an inductor or capacitor can be erroneous at these frequencies.  This may also be a factor.


As for some other questions -- air core is preferred over ferrite because ferrite saturates.  Rod cores aren't too bad here, as they can be made with quite high saturation currents, and the advantage is small anyway (mu_eff maybe only 2-5 say), enough to be helpful without getting in the way.

Saturation current should be quite high indeed for a mains LISN or CDN -- loads with poor power factor can draw quite high current peaks, and those current peaks are also likely where most of your EMI is transmitted (the FWB acts as a PIN diode), so you don't want your network pooping out on the peaks!

Tim

[Jay_Diddy_B]

TimNJ and the group,

I can probably help design you a LISN that will work in this application.

Your current specification is higher than the maximum for the LISN that I shared before.

I have a few questions for you:

1) Does the EMC specification that you are testing require one or two LISNs?

2) What is the maximum frequency that is required in the EMC specification for conducted emission?

3) Which EMC specification are using attempting to meet?

Regards,
Jay_Diddy_B

[Jay_Diddy_B]

Hi,

From here: https://www.com-power.com/uploads/pdf/LI-325C-1.pdf

I find this:



So it looks like you need two LISNs.

and this is impedance curve limits:



The limit lines are in red. It looks like 5uH // 50 +/- 20% to 152MHz.

This is all that is needed to design the LISN.

Regards,
Jay_Diddy_B

[TimNJ]

The answer to most of these, is: dampen resonances of the inductor.

Play around with this, for example; read the supporting information: http://hamwaves.com/antennas/inductance.html
It uses a helical waveguide model, and a bunch of corrections, to estimate the voltage and current through a single layer, wire solenoid.  Apparently this geometry is remarkably difficult to analyze!

....

The first (parallel) resonant mode is fine, that's what we want -- a high impedance between DC and EUT sides.  What we don't want is the second (series) resonant mode, which acts in parallel with EUT, shunting signal and distorting the frequency response.

Knowing that standing waves are the culprit, we can employ resistors in strategic locations to dampen those modes.  The series resonance has an antinode in the middle (voltage peak), and nodes (current peak) at the ends, which is why we measure a low impedance at the ends, at that frequency.  A resistor from one end to the middle will act in parallel with that mode, dampening it.  The resistance acts in parallel with the voltage peak, and gets 1/4-wave-transformed into a series equivalent resistance at the terminal.  When R = Zo,* parallel R is transformed to series R, and that's simply the resistance you will measure at the port.

*For an ideal transmission line, Zo is simply Zo.  Helical waveguide however is dispersive (velocity varies with frequency), meaning the Zo varies for each resonance; the resonances also aren't harmonically

...
...


Tim

Wow, thanks Tim. As always, you find time to give the most informative answers possible and I appreciate that. I guess the takeaway is that as resistance (or lossy material) added to an LISN (or any similar style network) is to add some damping to keep incidental tank circuits from messing with the "flatness" of the frequency response. From playing with the calculator you attached, and the GCI-Wcalc mentioned in the other thread, I've noticed that making a single 5uH coil with a reasonable diameter (let's say 10mm) might push the SRF to <100MHz. (The conducted EMI measurement goes from 150KHz-152MHz.) But, if you break up the 5uH coil into 4 or 5 coils, then each coil's SRF will be >200MHz. Presumably, if the coils are far enough apart (physically) any proximity effect related issues (that would contribute to parallel capacitance) would be practically gone.

Perhaps I should do some SPICE simulation based on the results from Serge Stroobandt's solver. Even if the SFRs of each coil are above the frequency band of interest, I suppose there could be a way the 4 or 5 coils interact to cause distortion at a lower frequency...maybe? (Talking out the wazoo here.)

By the way, is the "first (parallel) resonant mode" you refer to the "intended" LC interaction as shown in a typical LISN schematic. And the "second (series) resonant mode", that's the interaction between the Cp and the Ls of the coil?

Thank you!

[TimNJ]

TimNJ and the group,

I can probably help design you a LISN that will work in this application.

Your current specification is higher than the maximum for the LISN that I shared before.

I have a few questions for you:

1) Does the EMC specification that you are testing require one or two LISNs?

2) What is the maximum frequency that is required in the EMC specification for conducted emission?

3) Which EMC specification are using attempting to meet?

Regards,
Jay_Diddy_B

DO-160 generally speaking requires 2, one for (L/+) and one for (N/-) if AC or DC respectively. However, some aerospace systems are allowed to return power through a chassis connection, so in some cases you can use a single LISN on the L/+ line, and bond the return direct to the chassis/ground plane.

In my case, these are not typical aerospace metal boxes, and no expectation of returning power through a chassis, so I need two LISNs. The conducted EMI test is from 150KHz-152MHz, which is a departure from the "normal" 30MHz upper limit I'm used to for CISPR.

There is also the radiated emissions test which is, of course, much harder. I have never tried it, but known there are reasonably accurate ways of deriving radiated emissions from CM currents on the cables, as here: https://www.tekbox.com/product/AN_-RF_current_to_electric_field_strength_extrapolation.pdf. I may consider trying it.

These are switching power supplies operating at normal ~100KHz switching frequencies. In my experience so far, there's never usually significant radiation about 250MHz...so if the LISN could at least operate someway consistently up to about 250MHz, that would be ideal.

Thanks for your help and interest.

So, far I've come up with using 15AWG wire (1.45mm) @ 16 turns, on a 3/8" former...yielding 1uH with an SFR of about 250MHz. But not sure what happens when you put 5 in series!

[TimNJ]

If I wanted to do some lumped element modeling in SPICE with coil parameters derived from https://hamwaves.com/inductance/en/index.html#input...

...how might I approach it? For example, if I set the design frequency to 1MHz, Cp comes out around 1nF. If I set it to 100MHz, it comes out at 0.3pF. 

Perhaps trying to do some resonant tuning/damping in spice is a futile endeavor due to the fact that a lumped model is inherently only relevant in a limited range of frequencies?

And, I'm having a hard time believing the capacitance could really be 1nF @ 1MHz. At 150KHz, it says it's about 65nF! That's a heck of a lot of capacitance...What am I missing?

[T3sl4co1l]

The Cp equivalent model is a poorer approximation, and as you note, varies with frequency.

The only fundamental thing you can really do, is take the R and L as given -- which are the series equivalent resistance and reactance, reactance being divided by frequency to give inductance -- and plot them over a range of frequencies, then construct an equivalent network that fits those.  Bit of a PITA.

You might build such a model to hit the resonances without putting too much effort into the overall shape; in that case, hunt around for the resonant frequencies (points where reactance goes through zero, from increasingly large L to very small C; and vice versa) and construct the equivalent accordingly.

By the way, is the "first (parallel) resonant mode" you refer to the "intended" LC interaction as shown in a typical LISN schematic. And the "second (series) resonant mode", that's the interaction between the Cp and the Ls of the coil?

No relation to a circuit -- just the two-terminal component itself.  Any small 2-terminal component can be modeled as a parallel combination of R+L+C networks, or a series combination of R||L||C networks; or ladder and other topologies can be used.  (The ladder topology has the advantage that it's relatively easy to calculate: the alternating series and shunt branches are the terms in the continued fraction expansion of the impedance.)

For example, here's a fit to a common mode choke:
https://www.seventransistorlabs.com/Images/CurveFit1.png
https://www.seventransistorlabs.com/Images/CurveFit2.png
you can see the resonant mode(s) at the top end aren't well documented anyway, and aren't tightly fitted.

In this particular case, most of the resistances will be core loss; in particular, the stack of R+L's, in a geometric progression, fits a diffusion characteristic 20-400kHz, skin effect of the wound metal strip core.  It looks like it has some actual capacitance intrinsic to the core as well, presumably the capacitance between layers (C1).

L3 is probably leakage between winding and core, and L4 could be a partial leakage inductance between parts of the winding itself.  The fit breaks down here, partly because the data probably aren't great (and cut off suddenly), and because the impedance up here is due to standing waves in the winding, a transmission line or wave effect not easily modeled by lumped elements.

R5 isn't part of the model, but used to test it as an impedance divider.

The impedance plot of a helical resonator will show multiple peaks and valleys, like the high end in this plot, but more of them.

Tim

[Jay_Diddy_B]

Hi,

I think you could try and build one using off the shelf inductors.

Consider these models:




The models are set up to measure the impedance of the circuit by injecting 1A.

On the left is the standard 5uH//50

In the middle I have a Wurth 7443640470 inductor. This has SRF of 20MHz and a saturation current greater than 40A.
The model in LTspice includes the following lumped parameters:




The right hand model is this inductor in parallel with 50


These are the results:



This is almost good enough the impedance is 37 at 150MHz.

If I replace the inductor with two pieces of Wurth 7443630220 in series I get:



Which is fine.

Regards,

Jay_Diddy_B

[Jay_Diddy_B]

Hi,

I am going to share the results of some quick experiments that illustrate the challenges of measuring the impedance at the higher frequencies.




I am measuring the 5uH LISN that I designed. I am using an HP3577A and a clone of the HP 35676A R/T Test Set. I have calibrated the network with SOL at the reference plane indicated with a red line.



At 30MHz, the marker position the impedance is 50 above 100MHz the impedance rises and at 200MHz the impedance is around 60 .




If I measure two adapters, terminated with a 50 load I get this result:



You can see that the two adapters connected back to back is worse than measuring the LISN.

Alternate Calibration standards

If I use these standards instead of the normal SOL calibration:



So now the network analyzer is calibrated at the banana jacks, I get:



This result is text book.

I just wanted to show how small amounts of inductance can effect the high frequency impedance measurement.

Regards,

Jay_Diddy_B

[TimNJ]

Thank you all. I realize I need to go back to my LISN fundamentals so I’m spending some time with Henry Ott and this rather insightful pdf from Michigan State: https://www.egr.msu.edu/emrg/sites/default/files/content/module11_conducted.pdf

At least where I work, the LISN is a somewhat “mythicized” black box type thing, so this is forcing me to think about it a little more.

I guess my question now is Jay_Diddy, if it’s seemingly so straightforward to put together the network with off the shelf parts, and with reasonable frequency characteristics, why are there so many LISNs with fancy damping techniques? Maybe that’s what’s required to get true flatness in the 50ohm band.

Your plots were very helpful in my understanding. Initially my thinking was that if the inductor’s SRF was anywhere in the measurement band, it was useless..but now it’s obvious that a resonance will mostly not affect the 50 ohm impedance, and mainly affects the HF roll off of the 50 ohm impedance. Still, there certainly can be many other things to which make an inductor look, err..not so much like an inductor, as Tim showed in his complicated model of the common-mode choke. Clearly, there can be several resonant interactions and an impedance which doesn’t quite rise quite in like with 2Pi*f*L.

Seemingly those Wurth inductors were pretty well behaved, ferrite core and all.

Whats SOL by the way?

[2N3055]

Thank you all. I realize I need to go back to my LISN fundamentals so I’m spending some time with Henry Ott and this rather insightful pdf from Michigan State: https://www.egr.msu.edu/emrg/sites/default/files/content/module11_conducted.pdf

At least where I work, the LISN is a somewhat “mythicized” black box type thing, so this is forcing me to think about it a little more.

I guess my question now is Jay_Diddy, if it’s seemingly so straightforward to put together the network with off the shelf parts, and with reasonable frequency characteristics, why are there so many LISNs with fancy damping techniques? Maybe that’s what’s required to get true flatness in the 50ohm band.

Your plots were very helpful in my understanding. Initially my thinking was that if the inductor’s SRF was anywhere in the measurement band, it was useless..but now it’s obvious that a resonance will mostly not affect the 50 ohm impedance, and mainly affects the HF roll off of the 50 ohm impedance. Still, there certainly can be many other things to which make an inductor look, err..not so much like an inductor, as Tim showed in his complicated model of the common-mode choke. Clearly, there can be several resonant interactions and an impedance which doesn’t quite rise quite in like with 2Pi*f*L.

Seemingly those Wurth inductors were pretty well behaved, ferrite core and all.

Whats SOL by the way?

Air core inductors won't saturate.. Cored inductors will also change characteristics as they get loaded. So with cored inductors you have to be mindful of the current going through..

SOL : Short, Open Load calibration of VNA....

[Jay_Diddy_B]

Thank you all. I realize I need to go back to my LISN fundamentals so I’m spending some time with Henry Ott and this rather insightful pdf from Michigan State: https://www.egr.msu.edu/emrg/sites/default/files/content/module11_conducted.pdf

At least where I work, the LISN is a somewhat “mythicized” black box type thing, so this is forcing me to think about it a little more.

I guess my question now is Jay_Diddy, if it’s seemingly so straightforward to put together the network with off the shelf parts, and with reasonable frequency characteristics, why are there so many LISNs with fancy damping techniques? Maybe that’s what’s required to get true flatness in the 50ohm band.

Your plots were very helpful in my understanding. Initially my thinking was that if the inductor’s SRF was anywhere in the measurement band, it was useless..but now it’s obvious that a resonance will mostly not affect the 50 ohm impedance, and mainly affects the HF roll off of the 50 ohm impedance. Still, there certainly can be many other things to which make an inductor look, err..not so much like an inductor, as Tim showed in his complicated model of the common-mode choke. Clearly, there can be several resonant interactions and an impedance which doesn’t quite rise quite in like with 2Pi*f*L.

Seemingly those Wurth inductors were pretty well behaved, ferrite core and all.

Whats SOL by the way?

Hi TimNJ,

You are right there are a lot of myths associated with building LISNs. Which has resulted in a lot of 'traditions' and not a lot of engineering. I believe that that there are many reasons for this:

1) MIL 461 one of the earlier standards was introduced in 1967. In 1967 we didn't have the same tools that we have today.

2) Some of the early EMC specifications described the construction of the LISN with air-cored solenoids. This started the tradition of using air-cored inductors.

Fast Forward today, we have

affordable VNAs
Simulation tools
Modern components
Internet

Concept


The concept of a LISN is simple. It provides a standard impedance to the DUT. In this case 5uH in parallel with 50 .
The requirement is that the LISN looks like 5uH//50

Myth destruction

1) You can use inductors above their self-resonant frequency.

At high frequencies the LISN impedance is 50 so the requirement is that the impedance of the inductor is 'large' (say 200 ) compared
with 50

2) There always benefit in using a string of different inductors with different SRFs.

This is NOT true.

If you connect inductor in series with different SRFs then they can interact.

3) Inductance changing with DC current

This is not an issue providing the impedance remains 'large' compared to 50

Air cored Inductors


Air cored solenoids typically need many turns to achieve the desired inductance. This long winding may transmission line effects within the target frequency range.
The air cored inductors need to be in a large case, typically 3 diameters of the coil, to avoid coupling the case.

Cored Inductors

Cored inductors are really air cored. They have an air gap or a distributed air gap. The magnetic material is equivalent to a conductor in an electrical circuit.
The winding in a cored inductor is much shorter. This helps eliminate transmission line effects in the target frequency range.
The flux is confined to the core, reducing the effects of coupling.

Inductor Measurements

I am going to measure the Wurth 4.7uH Inductor, 7443640470:



I have solder 2x 100 resistors in parallel and a BNC connector:



This allows the impedance to be measured with my VNA.




This is a log frequency sweep from 10kHz to 200MHz.
The vertical axis is linear 5 /div




This is the LTspice simulation of the same circuit. The LTspice simulation uses lumped parasitic components.

There is excellent agreement between the model and the measured results.

This implies that we can accurately predict the performance of a LISN built this way.

I don't have the proposed inductor 7443630220 in my inventory, or I would have measured it.

SOL is  Short, Open , Load. This one a techniques used to calibrate the VNA.


Regards,
Jay_Diddy_B

[TimNJ]

Thank you very much.  I think I may give it a try with either 2x2.2uH or 3x1.5uH of the Wurth high current inductors. Seems their models are pretty good, which may be important Do these models account for effects at high current? Thank you for running some tests on your VNA!

Any reason I can't verify the same parameters with a spectrum analyzer with tracking generator?

A few more questions related more to the actual test standard/requirements:

1. The standard calls for 2x10uF connected between line and ground plane, and between neutral and ground plane. The setup is as below. For our normal CISPR setup, our LISN is actually earthed to the building's earth wiring. But in this case, if I was to earth the chassis/ground plane, wouldn't that be a ton of earth leakage current, possibly messing up other devices, not to mention potentially dangerous?

I'm trying to figure out if these capacitors are really only to be attached to a "local ground plane", insulated from actual earth. Does that make any sense?

2. On that same note, any reason why I can't put the two LISN halves in the same box? Traditionally, two boxes are used, but I think that's mostly because there are certain cases where you only need one. (Although, from reading the other thread, some people have vocalized opinions about why only using 1 LISN is a bad thing.)

Thanks again.

[Jay_Diddy_B]

Any reason I can't verify the same parameters with a spectrum analyzer with tracking generator?

A few more questions related more to the actual test standard/requirements:

1. The standard calls for 2x10uF connected between line and ground plane, and between neutral and ground plane. The setup is as below. For our normal CISPR setup, our LISN is actually earthed to the building's earth wiring. But in this case, if I was to earth the chassis/ground plane, wouldn't that be a ton of earth leakage current, possibly messing up other devices, not to mention potentially dangerous?

I'm trying to figure out if these capacitors are really only to be attached to a "local ground plane", insulated from actual earth. Does that make any sense?

2. On that same note, any reason why I can't put the two LISN halves in the same box? Traditionally, two boxes are used, but I think that's mostly because there are certain cases where you only need one. (Although, from reading the other thread, some people have vocalized opinions about why only using 1 LISN is a bad thing.)

Thanks again.

Hi TimNJ,

1) I am not aware of an easy way to do the impedance measurement that I did with the VNA with a SA and TG. You can probably get an idea if you have a directional coupler or a return loss bridge.

2) Most of the LISNs that I have built have a capacitor in the position of the 10uF capacitor. It may not be quite as big as 10uF but it is certainly an effect RF short.

3) There is no reason why you can't put two LISNs in one box. Here is a picture of a low current dual 50uH 50 LISN (CISPR 32) that I have built:



It works very well.

Regards,
Jay_Diddy_B

[TimNJ]

Thank you.My ignorance to most things RF is showing. Indeed, VNA and an SA+TG are not intended to measure the same things. Not sure how I got that in my head. Maybe I’ll get a $50 nanoVNA but I haven’t done much research to check if it’s a suitable device.

Regarding the 10uF caps, I guess I was just looking for a sanity check if putting 10uF between line/neutral and earth (at 400Hz) really made sense. Obviously the standard calls for it, but sheesh...that’s 40 ohms at 400Hz...and there’s two of them. I’m going to research more in the morning, but just thinking about putting 20uF to earth in a public work area just sounds bad, personally. Interestingly, the DO-160 conducted EMI measurement actually doesn’t use a direct measurement at the LISN. Rather, it uses a current probe. So, in this case, the spectrum analyzer can be isolated from the measurement setup. I may need to move the transient limiter to a small dongle outside the LISN in this case, presuming it still applies for current probe measurements.

In CISPR measurements, the SA earth and LISN earth are one. In this case, I wonder if the LISN chassis can just be referenced to a local plane, instead of a plane referenced back to the building’s electricity. Just spitballing, will update with my findings.

I’ll work on a KiCad project this week with some beefier inductors. Thanks!

[Jay_Diddy_B]

Hi,

Don't worry about the transient limiters being included in the LISN. They are 10dB 50 attenuators in the measurement range. This means that if you forget to terminate the LISN in 50 the maximum return loss is 20dB. It is 20dB because the signal goes through the attenuator, getting attenuated by 10dB and then is reflected going back through the attenuator. You simply add a 50 termination on the BNC connector.

The transient limiters are a good things.

The LISNs I have designed so far were designed using OrCAD. I am moving to KiCad. I have started putting the parts into my KiCad library:




I just threw them on a board so I could see what they look like.
I make the 3D models in Fusion 360.


Coilcraft make an alternative inductor to the Wurth 744364 series that is cheaper and has a higher SRF and is footprint compatible.

For the DC testing the 10uF capacitors don't matter.

What voltage is the AC testing at 400Hz?

Regards,
Jay_Diddy_B

[TimNJ]

Thanks. I mostly understand the function of the 10dB pad + HPF. The CISPR LISN we use (I believe) just has a very low capacitance diode across each BNC. Perhaps I'll just do that to provide some level of protection. For the current probe (https://www.tekbox.com/product/TBCP1_200_Manual.pdf), I think the bandwidth is probably low enough at low frequency to prevent front-end overdrive, although I suppose transients are still possible if something is left unterminated (??) Once again talking out the wazoo.

Thanks for the Coilcraft suggestion. I also found Bourns PQ2614 series inductors. (PQ2614BLA-1R5K). 3 of these in series. Typical resonant frequency is 60MHz so that might push the upper-end bandwidth out a little, might help for radiated emissions pre-compliance testing on the input power cable.

https://www.digikey.com/product-detail/en/bourns-inc/PQ2614BLA-1R5K/PQ2614BLA-1R5K-ND/7695895

The voltage at 400Hz is 115Vac nominal. So, that's on the order of 6A through the earth conductors of the building with 2x10uF. Just seems crazy.
10uF box-style film cap presumably has a fair amount of inductance anyway, so I wonder how effective is in the MHz range anyway. The recommended type per the standard is a feed-thru type with metal shell, with the metal shell bonded to earth/ground plane. Anyway, one of the commercial LISNs says this:

The mounting plate of the LI-325C is left unpainted in order to facilitate connection to earth ground in its installation, which is essential due to high leakage current.

...so I guess it is what it is. We'll make sure the earth connection is very solid, if that's what keeps it safe. Otherwise, I imagine reasonable voltage could even build up on the enclosure.

[Pitrsek]

Depending on your voltages, the textbox lisn might work for you (different box/connectors, bigger wires). I would not be afraid too much if you skirt the impedance limit lines for pre-compliance testing. fan might be needed to cool the inductors.
You can measure impedance by shunt s21 method. Basically you calibrate/normalize for through connection and connect dut between hot and gnd. Google for "2-Port Shunt-Thru". Check your setup against a known impedance to asses accuracy. 

[TimNJ]

Thanks! My company doesn't have a VNA, but as I mentioned, I might consider NanoVNA, assuming it's good enough. (I assume the $5000 Keysight VNAs are expensive for a reason? Thanks for the pointers, as my RF knowledge is at the "don't even know the right things to Google" stage.

Regarding the design, I've decided on 5x Coilcraft SER1590 1uH parts.

https://www.coilcraft.com/getmedia/81a5b014-f8b3-44cc-abe2-de459d23320f/ser1590.pdf

I simulated the design with the provided LTSPICE frequency domain model. The theory is very good...I'm not sure if the upper end frequency range would really behave that nicely in practice, but who knows.

[Jay_Diddy_B]

Hi,

The SER1590 inductors look like they will work. They may be slightly low on saturation compared to some of the other parts. The price is right.

The 10uF feed-thru capacitor is bit harder. This is what the 'official' parts look like:



I found three manufacturers of these but they all looked similar.

I found this listing:

http://www.icwltd.co.uk/products/ft.php

They show this style:



I don't really know much about the RF performance of these capacitors.

I would be tempted to try 10 pieces of a PHE850ER7100MR03R06L2 from Kemet. These are Y2 rated.

This would be one of the two LISNs:



I included the larger SER2918 Inductors.
I did not place the parts in the transient limiter.

Regards,
Jay_Diddy_B

[TimNJ]

Just did a sanity check on the Coilcraft Model...The SFR is not showing at all. Might explain why it's so flat...although 115MHz SFR sounds good regardless. You're right about the saturation point. Perfect inductance is really only maintained to 10-12A which isn't a huge upgrade over a standard Tekbox model. On the other hand the SER2011/SER2013 series are bigger and more expensive. I'll think about it.

Here's the model. Not sure what everything is exactly. Wurth model in the same setup shows SFR just fine.

.subckt SER1590-102_freq port1 port2 PARAMS: Cpar=27.7pF Ind=1uH
X1 port1 port2 Model1A PARAMS:
+ R1=535
+ R2=0.02
+ C= {Cpar}
+ K1=1E-09
+ K2=0.341
+ K3= {Ind}
+ K4=0.012
+ K5=0.00001
+ L=0
+ Is=0
+ a=0
+ L_Z0=0
+ L_EL=0
+ L_F0=0E6
+ PkZ=186.791125
.ends SER1590-102_freq

Regarding the 10uF capacitance...you are right that Y-class capacitors are technically appropriate here. I was looking for Y-class caps in large value today, but somehow did not come across this series, so I decided X2 might be okay for *laboratory use only*. The cost of Y2 film caps is very high, not to mention the size is enormous...close to 10x the size of X2.

I am definitely hesitant to say any of this on a public forum.

Warning to anyone reading: X-class capacitors are built to be safer than normal film capacitors, but still are not intended to go from line/neutral to earth. If a capacitors in that position fails, it has the potential to put the chassis at a high potential. X capacitors are not advised in the position I mentioned.

At the same time, for a properly earthed system, an internal short to earth should blow the internal or building fuse/breaker. I'm a little bit on the fence.

[TimNJ]

This doesn't necessarily make it right but the Tekbox TBLC08 LISN (teardown EEVBlog #993) uses Vishay MKP1847

 https://www.vishay.com/docs/28172/mkp1847ac.pdf

...not even X2. Just a normal film cap.

The LISN at work (Atten brand) also only uses X2 capacitors. I wonder what a Rhode and Schwarz uses.

[T3sl4co1l]

Honestly, I'd feel safer with MOVs on the mains side, and regular whatever-voltage-rated caps in the network, not X or Y.  630V or up is probably a reasonable choice.

Also, at some point we went from happy little DC LISNs, to mains LISNs.  Probably better to make one specific to each, so you don't have these problems with scale.  (The mains one probably won't do so well above 30MHz, either.)

The Coilcraft model can be plotted here, and also converted into a format usable in any SPICE engine:
https://www.seventransistorlabs.com/Calc/Coilcraft1.html

Ahh fuck, they updated their website, didn't they.  All my links are broken... 

These are the parameters required: https://www.coilcraft.com/en-us/models/spice/ I don't know what the other parameters are, in the above post.

Tim

[TimNJ]

Thanks. Any reason not to use X-class?

Sorry if it wasn't clear but I think I stated that it was an AC (115V/400Hz) + DC (14V/28V) LISN from the start, but maybe that's buried in the mounds of text now in this thread.. In my case, I'll be using an isolated programmable AC source to produce the 400Hz (Chroma 61503) and an isolated supply for the DC. So...maybe not mains in that sense since there's some isolation.

I guess you're just talking about issues >30MHz due to sizes of parts for the mains rated version, if there was one? I agree it would probably make sense. Unfortunately, don't have a ton of time so I'd really rather only have one...if it was "good enough".

Thanks for pointing to your website, shall give it a look.

Thanks again.