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

[T3sl4co1l]

X caps are safe for mains transients because they self-heal.  Self-heal dumps spikes into the system; not the greatest idea.

Running from a synthesized source?  Even better!  No worry for transients at all.  Hrm, well, beyond hot-plugging I suppose.  Even nominally rated capacitors will do (but a little extra, and/or the use of floating-electrode types, isn't a bad idea).

This may be one of those situations where staggered capacitor sizes and values proves beneficial -- you can get the stray inductance between them low enough that it doesn't make things worse at the inbetween resonant frequencies.  Also may be a good reason to choose PE dielectric (MKT and other types) over PP, the Q is a bit lower.  Anyway, bypassing on the DC side shouldn't be too important thanks to the dominant impedance of the decoupling inductor, which is really the star here.

Tim

[Jay_Diddy_B]

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.

Hi,

The Coilcraft model is only valid over a limited frequency range:

*======================================================================
* SPICE Model generated by Coilcraft
* Coilcraft Part Number : SER1590-102
* Inductance = 1uH
*======================================================================
* Model Parameters:
*   Valid Frequency Range = 0.1MHz-10MHz
*   Ambient Temperature = 25 degC
*   Inductor Frequency Model
*   Use model for Frequency Domain simulations
*======================================================================

They say here:

https://www.coilcraft.com/en-us/models/howto/using-coilcrafts-models-in-ltspice/

When running simulations, note the upper and lower frequency limits for which the model is valid. This information is shown on the Coilcraft SPICE model document for each inductor series.

Which make it pretty useless for looking at the SRF frequency.

The Wurth model seems to be correct. It matched the measurements that I made with a VNA.

Regards,
Jay_Diddy_B

[Jay_Diddy_B]

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.

The datasheet for the MKP1847ac capacitors used by Tekbox says:




I like the idea of using a better cap that is intended for connecting either from L to N (X rated)

or

L to GND (Y rated)


Although you might not need the mains rated caps in your application at 400Hz with the Chroma AC Source, why not build one that is safe as possible to use on the line?

There is some logic in building a small one for the DC application and a bigger one for the AC application.

Regards,
Jay_Diddy_B

[TimNJ]

X caps are safe for mains transients because they self-heal.  Self-heal dumps spikes into the system; not the greatest idea.

Running from a synthesized source?  Even better!  No worry for transients at all.  Hrm, well, beyond hot-plugging I suppose.  Even nominally rated capacitors will do (but a little extra, and/or the use of floating-electrode types, isn't a bad idea).

This may be one of those situations where staggered capacitor sizes and values proves beneficial -- you can get the stray inductance between them low enough that it doesn't make things worse at the inbetween resonant frequencies.  Also may be a good reason to choose PE dielectric (MKT and other types) over PP, the Q is a bit lower.  Anyway, bypassing on the DC side shouldn't be too important thanks to the dominant impedance of the decoupling inductor, which is really the star here.

Tim

Floating electrode as in MLCC with flexible termination? If so, interesting thought, though since these will be hand-soldered, I'd still feel a little queasy about the whole idea. Inductance of those 40x20mm MKP caps is probably not so great at the high frequencies is I guess what you are suggesting. I suppose I could bypass with Y-caps or similar to get some better MHz range performance. I suppose those should be located closer to the AC (or DC) input? Or maybe it's more important to keep the inductance between parallel caps down, exact position of HF bypass caps less critical?

Thanks.

[Jay_Diddy_B]

Floating electrode as in MLCC with flexible termination? If so, interesting thought, though since these will be hand-soldered, I'd still feel a little queasy about the whole idea. Inductance of those 40x20mm MKP caps is probably not so great at the high frequencies is I guess what you are suggesting. I suppose I could bypass with Y-caps or similar to get some better MHz range performance. I suppose those should be located closer to the AC (or DC) input? Or maybe it's more important to keep the inductance between parallel caps down, exact position of HF bypass caps less critical?

Thanks.

The floating electrode, is a type of capacitor construction where there are two (or more capacitors) in series inside the capacitor.
The Wima FKP1 shown here:



Has two capacitors in series.

This reduces the ac voltage across each of the internal capacitors by a factor of two.

Typically one section is used for every 200V of ac voltage rating.

The 10uF capacitor in the EMC application do two things:

1) the define the impedance at low frequencies  say less than 100kHz. Good RF performance is not needed for this part.

2) Act as a filter for noise coming from the input side. This is probably the main reason for having them. Good RF performance is needed for this.
(In your application with the Chroma Source) you might not need this function).

Regards,
Jay_Diddy_B

[TimNJ]

Hi,

The Coilcraft model is only valid over a limited frequency range:

*======================================================================
* SPICE Model generated by Coilcraft
* Coilcraft Part Number : SER1590-102
* Inductance = 1uH
*======================================================================
* Model Parameters:
*   Valid Frequency Range = 0.1MHz-10MHz
*   Ambient Temperature = 25 degC
*   Inductor Frequency Model
*   Use model for Frequency Domain simulations
*======================================================================

They say here:

https://www.coilcraft.com/en-us/models/howto/using-coilcrafts-models-in-ltspice/

When running simulations, note the upper and lower frequency limits for which the model is valid. This information is shown on the Coilcraft SPICE model document for each inductor series.

Which make it pretty useless for looking at the SRF frequency.

The Wurth model seems to be correct. It matched the measurements that I made with a VNA.

Regards,
Jay_Diddy_B

Thanks. I saw this too. Not sure what Wurth is doing to get a reasonably accurate model and why Coilcraft can't (or doesn't care) to model the high frequencies. Perhaps they decided it's outside the "useful" frequency range of the inductor (i.e. few people would try to design a switching power supply above a few megahertz anyway.) Maybe low frequency model accuracy suffers if try to fit the model to high frequencies as well. Not sure.

[TimNJ]

Floating electrode as in MLCC with flexible termination? If so, interesting thought, though since these will be hand-soldered, I'd still feel a little queasy about the whole idea. Inductance of those 40x20mm MKP caps is probably not so great at the high frequencies is I guess what you are suggesting. I suppose I could bypass with Y-caps or similar to get some better MHz range performance. I suppose those should be located closer to the AC (or DC) input? Or maybe it's more important to keep the inductance between parallel caps down, exact position of HF bypass caps less critical?

Thanks.

The floating electrode, is a type of capacitor construction where there are two (or more capacitors) in series inside the capacitor.
The Wima FKP1 shown here:

(Attachment Link)

Has two capacitors in series.

This reduces the ac voltage across each of the internal capacitors by a factor of two.

Typically one section is used for every 200V of ac voltage rating.

The 10uF capacitor in the EMC application do two things:

1) the define the impedance at low frequencies  say less than 100kHz. Good RF performance is not needed for this part.

2) Act as a filter for noise coming from the input side. This is probably the main reason for having them. Good RF performance is needed for this.
(In your application with the Chroma Source) you might not need this function).

Regards,
Jay_Diddy_B

Ah! Didn't know about those capacitors. Learn something new every day. Thanks.

Regarding the capacitance, I don't really know how the AC source synthesizes the AC voltage and what the noise signature of it is. Definitely some sort of switching topology, maybe some sort of DDS thing.

I simulated with 1uF and 10uF in that position. As you said, the low frequency impedance is indeed affected, although when comparing to the DO-160 LISN limits, seems it would still be in the allowable range. Then tried with 4.4uF (2x2.2uF). Impedance is really not that much different between the two.

Yellow = 1uF
Blue = 4.4uF
Red = 10uF

Perhaps the 10uF is somewhat of a historical artifact. Maybe the 400Hz sources of the past were very dirty/noisy. (Perhaps EMC test labes were even using mechanical 400Hz generator?)

Unless a higher impedance below 10KHz is for some reason a problem, seems like a lower capacitance would be okay.

edit: As we know, something's gone wrong with the attachments function. It was showing an image I posted in another thread. Should be fixed now.

[TimNJ]

Hi all,

Here's the planned schematic and layout. Earthed mounting holes mate directly to the die-cast aluminum chassis via M3.5 screws. Planning to use panel mount BNCs with a small 50ohm coax jumper from PCB to BNC. (Too many misaligned holes on previous projects.)

I've changed the inductors to SER2011-122 (1.2uH). 100% inductance to over 30A. I've added the option to populate two ceramic Y-caps on the "mains" side. Undecided if it's a good idea yet. The bottom side will have a matching earth polygon pour.

Any feedback welcome. Thanks.

[Jay_Diddy_B]

Hi,

What capacitors are you using?

I like to use Y caps for the 0.1uF. If they fail they will destroy the analyzer.

You need to use a transient limiter (or at least a 10dB) attenuator with this LISN if you connect the output to a spectrum analyzer.

Regards,
Jay_Diddy_B

[TimNJ]

The 3 * 3.3uF caps are X2. Planning to use this one: https://www.digikey.com/product-detail/en/kemet/F861DY335M310ZLH0J/399-17029-ND/8346549

Since posting, I've added a MOV across each, to keep the voltage stress on the caps as low as possible. For my case, should always be powered from a well controlled source but for anyone else who might attempt this, can't guarantee that.

The 0.1uF were also X2. As others and you have mentioned, large physical size and high frequency usually doesn't bode well, so was trying to keep the size and series inductance low. Not sure if it's relevant in that position. However, based on your recommendation, I will plan to use this: https://www.digikey.com/product-detail/en/wima/MKY22W31005D00KSSD/1928-1791-ND/9370699

Regarding the transient limiter I've installed two back-to-back 5V ESD diodes. I know the Tekbox 5uH uses a Bourns GDT rated at 60V. Necessary? Recommended? Anything else? If I add an HP11947 style limiter with 10dB, I'm just a little worried about insertion loss roll-off/deviation with frequency. Presumably I'd have to compensate on the SA somehow.

By the way, I bought a NanoVNA and am excited to learn how to use it. Hopefully I can give an accurate analysis of the impedance characteristics of this LISN once built.

[Pitrsek]

The measurement I've mentioned/described can be done with tracking gen. and SA. It is equivalent to what VNA does as s21, minus the phase. If you are measuring just passives, phase can be calculated out of the measurements. But amplitude is usually all that is needed. 

Ad 10uf capacitors:
some LISNs have the capacitors as add on module - both due to safety reasons, and it is more versatile. There are other regulations that are using 5uH LISN, but do not use the cap. If you are only after the required impedance profile, big foil cap across terminal is all what is needed. If you need to filter out hf crap from outside, as mentioned previously feed through cap is way to go. If you want to implement it on the pcb, and want really the best performance, layout is important.

Ad Layout:
The signal has to be forced through the capacitor pins, ie. fuse - pins of first cap - pins of second cap - pins of nth cap. Just trace, which is narrowed down just to solder pad of the capacitor at the cap pins. If you go with polygons, I'd slot it around the pins, so signal has to go through the pins. Same on the return. I'm usually a big proponent of solid unslotted (ground) planes, but in this case slotting is in order. It will definitely work as it is now, but the effect of the routing will be measurable at the higher end of LISN bandwidth.
 No ground plane under the inductors/traces. I'd put the inductors further away from each other, and provide connections for possible damping resistors/RCs.   

Minicircuits has nice power limiter VLM-33W-2W-S+
You might be also interested in noise splitters and parasitic coupling of components - google for "Characterization and Cancellation of High-Frequency Parasitics for EMI Filters and Noise Separators in Power Electronics Applications" - by Shou Wang

[TimNJ]

The measurement I've mentioned/described can be done with tracking gen. and SA. It is equivalent to what VNA does as s21, minus the phase. If you are measuring just passives, phase can be calculated out of the measurements. But amplitude is usually all that is needed. 

Ad 10uf capacitors:
some LISNs have the capacitors as add on module - both due to safety reasons, and it is more versatile. There are other regulations that are using 5uH LISN, but do not use the cap. If you are only after the required impedance profile, big foil cap across terminal is all what is needed. If you need to filter out hf crap from outside, as mentioned previously feed through cap is way to go. If you want to implement it on the pcb, and want really the best performance, layout is important.

Ad Layout:
The signal has to be forced through the capacitor pins, ie. fuse - pins of first cap - pins of second cap - pins of nth cap. Just trace, which is narrowed down just to solder pad of the capacitor at the cap pins. If you go with polygons, I'd slot it around the pins, so signal has to go through the pins. Same on the return. I'm usually a big proponent of solid unslotted (ground) planes, but in this case slotting is in order. It will definitely work as it is now, but the effect of the routing will be measurable at the higher end of LISN bandwidth.
 No ground plane under the inductors/traces. I'd put the inductors further away from each other, and provide connections for possible damping resistors/RCs.   

Minicircuits has nice power limiter VLM-33W-2W-S+
You might be also interested in noise splitters and parasitic coupling of components - google for "Characterization and Cancellation of High-Frequency Parasitics for EMI Filters and Noise Separators in Power Electronics Applications" - by Shou Wang

Oi! I guess I misinterpreted your post then. Well, I guess I have two options then, though I still have to learn about the measurement principles.

Regarding the 10uF capacitors: It's a good point you make. There's a possibility we'd want to test CISPR25 (auto/vehicle), but I really can't be certain. CISPR25 is 1uF.

CISPR25 also uses a more "standard" voltage measurement via BNC/N-type connector on LISN to measurement resistor. DO-160, on the other hand (what I'm currently after), only uses current probe measurement. I suppose it's possible to convert the limits between voltage and current measurements?? But per the standards, that's how they want you to test.

Regarding the layout: I've seen this kind of slotted polygon layout many times before in EMI filter designs, but to be honest, I don't understand why they help. I assumed the lowest inductance/impedance connection would be appropriate, so what's the mechanism here? High frequency currents maybe tend to scatter and bypass the capacitors? (Just seems counter-intuitive from what I know, but I'm sure my current mental models are flawed.)

Is this what you were thinking?



Indeed, no ground plane under the inductors as that will likely affect the SFR among other things.

Regarding the Minicircuits power limiter: This is a new type of component for me. As I understand it, it will pass 100% of the power between 0 and 12dBm. Above 12dBm and below 33dBm, it will limit the power to approx 12dBm. Sound correct? VSWR looks a little high at 200KHz. (Does it matter?) Also, seems like SA damage/problems often occur due to low frequency overdrive, i.e. at the fundamental switching frequency of a power supply...which could be <100KHz. Would this do much to attenuate those signals?

Regardless, this could be an interesting quick solution for me. I just do not have much time at the moment to work on transient limiters, etc, especially since the standard I'm trying to test to doesn't even use that method. If I go for a "DO-160 only" LISN, then I might just remove the BNC outputs altogether, and just put a 49.9R dummy resistor on the PCB and call it a day.

The other option is to maybe include 1uF on the PCB, then allow for connection of ~9uF outside the enclosure. Or no capacitors inside and swap out 1uF or 10uF externally, so that it can be used for CISPR25 or DO-160. But, then I'd need to add provision for voltage measurement internal to the LISN.

Thanks!

[Jay_Diddy_B]

Hi,

Don't work about 10uF versus 1uF for the CISPR 25 LISN. It will be fine, especially for pre-compliance.

The 0.1uF 22.5mm capacitor will have an inductance of about 20nH.

You could tighten up the layout at the grounded end of this capacitor. I use 50 coplanar wave guide.  There is only low voltage higher.

This model shows how the inductance impacts the LISN impedance:



Regards,
Jay_Diddy_B

[TimNJ]

Hi,

Don't work about 10uF versus 1uF for the CISPR 25 LISN. It will be fine, especially for pre-compliance.

The 0.1uF 22.5mm capacitor will have an inductance of about 20nH.

You could tighten up the layout at the grounded end of this capacitor. I use 50 coplanar wave guide.  There is only low voltage higher.

This model shows how the inductance impacts the LISN impedance:

Regards,
Jay_Diddy_B

Thanks again. Feel like I'm getting annoying at this point with the non-stop questions. But, thanks for your help.

What do you mean the grounded end of that capacitor? Which side is that?

The way I've chosen to approach this is to add SMA connectors on the board to allow for a future upgrade to actually use the measured voltage signal. For now, the bottom resistor is 5K (4.99K) per the standard, and then I will screw in a 50R SMA terminator to get the 50R impedance when using current probe measurement. In the future, I may bring the internal SMA connection out to a panel mounted BNC/N/SMA connector. Maybe add one of those Minicircuit limiters in between, or add HP11947A clone on a daughter-board inside. For now, I just don't have the time to think about it.

The trace going to the SMA connector is coplanar waveguide 50R (1.5mm width/0.3mm gap/1.6mm FR4). I don't know if that's helpful or necessary. Maybe I should add a few stitching vias to tie the planes together.
Where's the coplanar waveguide you're talking about go?

Thanks.

[TimNJ]

From a safety standpoint this is questionable, but a fat HV ceramic seems pretty nice in this application.

https://www.digikey.com/product-detail/en/kemet/C2225X104KFRACTU/399-10506-1-ND/4172868

About 800pH series inductance according to Kemet's K-SIM. Unfortunately, even with boardflex/flexible terminations, I'm not sure if this makes sense, especially granted it's position near mounting holes.

Or, use 2 x 0.047uF Y2 film caps in parallel. In theory, it should halve the inductance, but in practice do not know if this winds up making a resonance point or something else.

[Pitrsek]

Paralleling same value capacitors works well. Paralleling same capacitors actually it works much better than the 1uf 100n 10n triad of resonance...
Make yourself a impedance test pcbs so you can see for yourself (Or buy them https://www.sv1afn.com/en/products/rf-experimenter-s-pcb-panel-of-8-pcs-diy-kit.html).
Each piece of trace adds an inductance. Depending on the inductance of the capacitor and added inductance this might have big or negligible effect. If you have power plane/ground plane pair with very tight spacing, and caps connected directly to the planes, added inductance is marginal. In case of 2L board, spacing is not tight and added inductance might not be negligible at all. Ie. If you add 2cm of trace you will multiply inductance of a small mlcc capacitor. It's a same here, just capacitor inductance is rather big to start with, so the effect is not as huge.

I did a quick test - small pcb, top was signal, bottom ground capacitor(330n/630v mkt, rather big package) connected between the two. SMAs on both ends. 2 samples - one with solid planes on both sides, second with plane slots so it forced flow through pins of capacitor. Difference in attenuation at HF was cca 1.5db. This is not much but: big package - with smaller package the effect would be bigger, and if you have multiple capacitors/multistage filter and on each cap you can gain 1.5db of attenuation....
In your case it is a detail, but depending on components size and layout, it might have bigger effect.

Imagine you are electron traveling from the fuse to the first small cap. In you original design, you go through the trace(inductance) and through the capacitor. If you want to go the output, you do not travel through the same inductance. You have basically added extra ESL to you capacitor.

Actually there will be a HUGE resonance between the 470pF and 10uF. At some point it will be marginally better than just 10uF, and at some much, much worse than just 10uF. For more of the topic I can recommend something from Steve Sandler about flat output impedance. Or Istvan Novak - distributed matched bypassing.

But this all is more of curiosity/academic discussion, unless you have extremely noisy power supply your lisn will be fine will be fine either way.                     

[T3sl4co1l]

I don't think quibbling about nH is all that useful here.  The body of the capacitor will be some pF to surroundings, thus having a LCL equivalent with a cutoff at some point, and whatever impedances.  Perhaps more intuitive is to consider it as three TL segments: the leads and body.

About all you can do about the body is remove ground under it.

Tim

[TimNJ]

Thanks to everyone for their input. I've ordered a few PCBs based on some recommendations here. I will try to characterize the design once I receive the NanoVNA (and once I watch enough W2AEW tutorials to figure out how to use the thing).

Pitrsek, regarding your explanation:

Imagine you are electron traveling from the fuse to the first small cap. In you original design, you go through the trace(inductance) and through the capacitor. If you want to go the output, you do not travel through the same inductance. You have basically added extra ESL to you capacitor.

Which capacitor? I guess I just can't seem to wrap my head around how a cutting the polygon pour (in my head, increasing the  average distance traveled) makes the inductance lower.

---

Side question: The main difference between a 5uH and 50uH LISN is the low frequency impedance, below roughly 5MHz...If the noise source is modeled as an ideal current source, then the voltage at the RF port of the LISN in theory should always be worse with a 50uH LISN than with a 5uH LISN. I thought about it for a while and figured there was a hypothetical situation where the wiring from the DUT, its input EMI filter, etc. may resonant with the LISN's impedance in such a way that the EMI level is worse with 5uH than it is with 50uH.

However, after trying a number of LTSPICE simulations, the only time I could get resonant peaking (with higher RF port voltage on 5uH vs 50uH) was with a relatively large common-mode capacitance (line to earth), in the range of ~5nF and above. I feel this would generally be a rare scenario, especially for mains power supplies or anything with earth leakage requirements.

Perhaps a situation still exists, but my question is: Is it generally true that a system passing EMC standard on 50uH should most likely pass on 5uH?

Thanks.

[Jay_Diddy_B]

Hi,

Let me address the differences between the 5uH and 50uH LISNs.

Historically the 5uH LISN were used in small vehicles like cars and planes where the wiring is short. 50uH LISNs were used for line voltage and ship board applications where the wiring is longer.

At high frequencies the impedance of both LISNs is the same, it is 50 . So for high frequencies the results will be identical.

LTspice Model

I am going to turn to LTspice to illustrate the differences between the LISNs. I am stepping the value of the inductor between 5uH and 50uH:



Results



The bottom graph shows how the impedance changes with frequency.

The middle graph shows the voltage output if the emission has a 50 source impedance.

The top graph shows what happens if the emission source is low impedance. I have chosen 1 . It could be lower than 1 . In the case the choice of LISN has very little effect on the voltage at the output of the LISN.

Regards,
Jay_Diddy_B

[TimNJ]

Thanks, as always, for your insightful responses. I understand 5uH may closer approximate a more compact distribution system , and 50uH may closer approximate the wiring in a building. That part makes sense from a historic/rationale perspective. But if the response to EMI is so similar..why were two impedances ever created?

What if we actually measure the EMI current going into the LISN, instead of the voltage across the 50R resistor? (This is as measured in DO-160 and for signal wires in CISPR25). In this case, the 5uH inductor presents a lower impedance and thus the current below ~5MHz will be higher for a given noise voltage signal. No? By contrast, if measuring at the RF out port, since more EMI current is now shunted through 5uH, then less signal reaches the RF out port.

So, by that logic, seems you will show lower EMI with a 5uH LISN if using RF measurement port, but you will show higher EMI if using current transformer method. Once in the 50R region, then both methods should show equivalent readings. Am I making sense?



[ Specified attachment is not available ]

[TimNJ]

A few sanity checks about safety:

The LISN is an unusual piece of test equipment due to it's normal high earth leakage current. 60Hz LISNs may leak close to 1A (depending on the standard). For 400Hz DO-160, the earth leakage current is close to 3-6A (!!!)

However, if my analysis is correct, with an isolation transformer (or isolated AC source), a high current no longer flows in the earth wiring of the building since the conductive path is broken by the galvanic isolation of the transformer. So, in these cases, no more 3-6A flowing in earth wiring which could possibly cause a ground bounce issue that might affect nearby equipment and/or elevate the voltage on nearby earthed surfaces to something unpleasant or dangerous. The earth leakage current through the 10uF caps should be limited by the capacitive reactance/insulation resistance of the isolation barrier (transformer)...which will be on the order of 100s of KOhms at least.

Is that correct?

For an ideal isolation transformer, grabbing either the live or neutral secondary wire is not a risk because there's no way for current to flow through your body. But, with two 10uF caps now tied to ground, a loop can be formed on the secondary side of the transformer. It's is nearly "earth-referenced" with such a huge capacitance between earth and one of the secondary. So, grabbing one of the hot wires allows a current to flow out of the transformer, through your body, through the earth, up through one of the capacitors, and back into the transformer. Danger! Still, this requires actually touching a hot wire, which is not such a likely scenario.

If the LISN is left unearthed with an isolation transformer, touching the housing of the LISN should still be safe since you're still galvanically isolated from mains. From this, it seems that an isolated source provides a good deal of safety.

On the other hand, with a non-isolated source, there seems to be much less room for error. If the LISN is left unearthed, the housing floats to 1/2 the mains voltage with respect to earth. Touching the housing will be nasty, especially since the source impedance will be quite low through the 10uF. And, now, since there is a conducting path back to the mains supply, a large steady state current flows in the building earth wiring/structure.

So, it sounds like isolated supply is really the way to go for two reasons. Does this seem accurate?

[Jay_Diddy_B]

Hi

If the noise source is current, not voltage, you get this result:



If you inject and current and measure current you get a flat response from the CT and the expected result from the LISN impedance.

Jay_Diddy_B

[TimNJ]

Thanks. Sorry, my circuit image is not showing. I modeled as voltage source with 1R source resistance, same as you did before. But then measuring the current into the LISN. In reality, I don’t know if a “typical noise source” (if one existed) is more closely a current source or voltage source.

[chris_11]

Try to give you some practical advice. The first inductor data sheet SRF is the parallel resonance of the inductance with its own parallel capacitance. This is in a LISN a high impedance point (against 50 Ohm) and of no concern. The problem are the real series resonance frequencies (low impedance points) where partial parts of the inductor inductance resonate with partial serial capacitance. Since those points are not given in any data sheet you need to characterise them with your own measurements (VNA or Tracking generator with SA). You are better off with several low inductors in series (same values) because their resonances are higher than larger inductances in the same package. I got good results up to 1Ghz with 6x 800nH SER2009-801 Coilcraft inductors in series. Do not mount them on a PCB, since the connections will add capacitance and move the partial series resonances lower. Solder those 6 inductors in the air in a long strip. You might support them with some low capacitance material on the core if you like too, but due to the flat wire with over 50A capability it is usually not an issue.

br
Christian

[TimNJ]

Hi everyone,

Photo of the assembled PCB is below (sitting unmounted in the Bud AN-1322-A enclosure.) I'll report back with NanoVNA measurements soon. If everything's okay, I'll share the KiCAD project files after.

Thanks for all of your help!