Staggered 50uH and 250uH inductor design for LISN

[floobydust]

Again, thanks for real measurements.

I had real trouble finding/seeing a power ferrite past 10MHz so I thought only air-core would be suitable.
Looked at SMT ferrite power inductors from several manufacturers; TDK/Epcos, Murata, AVX, Vishay, Coilcraft, Wurth etc.
Most have no specs for frequency/SRF, the better ones start rolling off L at 10MHz and SRF 30MHz or less.
This is for 4.7uH-10uH and >10A. I'm used to permeability/saturation being prioritized over operating frequency, you want minimum copper (losses).

LTSpice sim with Wurth parts 744333, 332 and their RLC models;
It looks like qty. 7 of 6.8uH WE-HCC 7443320680 Isat 13A SRF=42MHz ferrite is good, as your the analyzer confirms. About $50 for qty. 14. Some Wurth HCC parts had losses at 20-30MHz, attenuation seems limited by the damping resistors.

An air-core part is the size of a can of beer and ugly with no coilforms to keep the windings from springing around. The commercial LISN has a machined-nylon former.
I could only see 1-1/2" or 3" ABS plumbing pipe as low cost but 10-14AWG windings would still move. So mechanically, the air-core is PITA- unless others have suggestions for OP.

[CopperCone]

you can use super glue (sprinkle with baking soda to make it solidify fast), then every 8-12 turns add another dash of glue on the windings to keep it taught. This is what I did.

Not perfect but it sorts works.

I used wood clamps to hold it tight (with the rubber bits) when the glue is drying, and sometimes I had to beat it with a wooden dowel and a hammer (gently) to compress the windings (I am reusing 2.65mm magnet wire from other large inductors I have no use for)

[Jay_Diddy_B]

Hi,

I only measured 6.8uH because I didn't have any 10uH parts.

Nothing particularly bad happens at the self-resonant frequency and beyond, except the impedance starts to drop.

I will do some analysis later to show this.

Regards,

Jay_Diddy_B

[Jay_Diddy_B]

Floobydust and the group,

I looked at the model that you created and there are a few errors.

This is the corrected version:



The DUT under test is connected to the left side of the LISN, represented by the source V1. The line voltage input is normally connected on the right side. This can be open or short for the evaluation, because the 1uF capacitor on the right side is an effective RF short.

There are two major parameters:

1) input impedance

This is obtained by measuring the voltage applied to the LISN and dividing by the current.





2) Transmission

You can measure the transmission from the source V1 to the output. The spectrum analyzer is represented by the 50 Ohm resistor connected to ground.



I have attached my modified model.

There is insignificant difference between all 4 models. Any of them are practical circuits.


Regards,

Jay_Diddy_B

[Jay_Diddy_B]

Hi,

I promised early that I would analyze a LISN with a relative low SRF inductor.

This is the model:



Really there are three models. The top model is used to confirm the model of the inductor. The current source is 1A so the impedance is in Ohms.
The impedance peaks at the self-resonant frequency and then decreases with increasing frequency.

The second model is a LISN built with the 50uH inductor with a 7MHz SRF.

The third model uses an ideal inductor for comparison.

Results



Even with a self-resonant frequency of 7 MHz this results in a LISN which is 10% low (45 ) at 150 MHz.

I have attached the LTspice model.

Regards,
Jay_Diddy_B

[G0HZU]

Those lumped models look to be very basic... Using a model like this for a large 50uH inductor that has a SRF of 7MHz is not going to be adequate if you try and use that model out to something like 150MHz.
I'm assuming here that the 50uH inductor in this case is a physically large solenoid wound with thick wire?

An inductor like that will need to be modelled as a complex transmission line structure or (better still) usea  VNA to obtain a two port s parameter model as this type of inductor will have multiple resonances by 150MHz and your basic lumped model can't capture this.

[Jay_Diddy_B]

Those lumped models look to be very basic... Using a model like this for a large 50uH inductor that has a SRF of 7MHz is not going to be adequate if you try and use that model out to something like 150MHz.
I'm assuming here that the 50uH inductor in this case is a physically large solenoid wound with thick wire?

An inductor like that will need to be modelled as a complex transmission line structure or (better still) usea  VNA to obtain a two port s parameter model as this type of inductor will have multiple resonances by 150MHz and your basic lumped model can't capture this.

G0HZU,

The model is very basic. Most of the discussion has been about whether it is o.k. to use a cored inductor in a LISN and what is the impact of the various parasitic elements on the LISN performance.

The 50uH LISN is (normally) used in the frequency range from 150kHz to 30MHz. So it is the performance in this frequency range that is important. The modelling is been done to higher frequencies, because this is where the parasitic effects show up.

I tested a 6.8uH Wurth HCC ferrite cored inductor on a VNA (HP8714C) and it behaved like the lumped model.

I haven't tested one of the single-layer solenoids normally found in commercial on LISNs on a VNA. I don't expect it will perform all that well.
If I was winding a single layer solenoid, I would space the winding from the coil former and I would ensure that there was some space between the turns to lower the turn-turn capacitance.

Regards,

Jay_Diddy_B

[Jay_Diddy_B]

Hi group,

I decided to wind a solenoid inductor using PVC coated wire.  The coil is about the same size as a 12oz (355ml) pop can:



I used 2" ABS pipe which 2.37 inches in diameter.

And PVC coated 16Awg wire which is about 0.1 inches diameter.

35 turns, 4 inches long.

I used the calculator found here:

http://electronbunker.ca/eb/InductanceCalc.html

I was aiming for 40 turns which would have resulted in the 50uH inductor. I only wound 35 turns at I got a measured inductance of 37uH. I put my dimensions back into the calculator I get the same answer:



I then connected the inductor in parallel with a 50 resistor to a HP 8714C VNA and measured the impedance.




I can see dips in impedance at 29, 43, 58 and 73 MHz. These are probably caused by some transmission line effects. These are the effects that G0HZU mentioned.

Regards,
Jay_Diddy_B

[G0HZU]

Thanks. I've not done anything with LISNs apart from seeing them get used many times for formal EMC testing at work etc.

But some of the LISNs in the links use huge solenoids for the 50uH inductor and in the absence of the damping resistors I'd expect them to have resonances starting at maybe 7MHz but I'd also expect to see a significant series resonance by maybe 20-25MHz. Then more series resonances every 15-20MHz or so. Presumably the damping resistors are there to try and take this out.

The Tekbox model looks a bit strange. Does it really use 68R damping resistors? This value seems really low. Their 4 section 50uH inductor also looks to be symmetrical which seems odd. I would have expected to see some deliberate asymmetry that aims to break up the various resonance modes. But then I've never tried to make a 50uH LISN like that.

At work I do a lot of ultra wideband RF design work so I have to model inductors very carefully up to frequencies way beyond the first few resonances. Usually I do this with a 2 port VNA model but it can also be done (with limited success) with complex transmission line models.

[G0HZU]

Looks like we double posted...

Yes, those are the series resonances in the solenoid.  The first one crudely corresponds to the solenoid acting as a half wave transmission line. Energy enters one end and there is 180degrees phase shift by the far end. Then it hits the 1uF shorting cap at the far end and there will be a 180degree phase shift in the reflected wave (because it sees a short circuit here). Then there is another 180degrees on the return trip back through the solenoid. So it returns with 180+180+180 which means it returns in anti phase so it looks like a (lossy) short circuit where you want it to look like 50R. I expect that this will typically happen within the 30MHz bandwidth for a huge 50uH solenoid structure like this?

The resistors (between turns) can damp this out but I'd also expect to see undamped sections between the resistors. Also I'd expect to see some deliberate asymmetry in the way the resistors are fitted along the structure. The (symmetrical?) Tekbox solenoids therefore look a bit strange to me but then again I've never tried to make a 50uH solenoid like that.

Try tacking a few resistors along the windings to see how the dips get damped out?

[Jay_Diddy_B]

Hi,
The Tekbox LISN that can be found here:

https://www.tekbox.net/test-equipment/tboh01-5uh-line-impedance-stabilisation-network-lisn-cispr25 is a 5uH LISN. According to the Tekbox documentation they use four 1.25uH air cored inductors connected in series.

It is a little suspicious, because if you  mount the inductors as close together as they seem to they will couple.



Source: https://www.mikrocontroller.net/topic/323322



The manual shows the schematic:



This is different than a 50uH LISN.

Transmission line effects are less of problem in a 5uH LISN because the wire is shorter.

Regards,
Jay_Diddy_B

[G0HZU]

I did a quick youtube video showing an old inductor model I developed at work many years ago. Sorry, there's no sound I don't have a microphone.

It is created using the physical dimensions of the inductor (turns, diameter, length etc) and translated into a distributed model. The red trace data is s2p data taken of a real 400nH inductor.

It works quite well to about three times the frequency of the first 1/4wave resonance at 250MHz. This resonance is the one that the crude/classic lumped model tries to capture. It shows how well the resonances and impedance agree up to about 1GHz. This old model worked really well if a VNA wasn't handy and it could also be used for transient analysis. It works for solenoids big or small and any L/d ratio (within reason). So it should be able to model the big inductors used in LISNs.

But the best way to model an inductor is usually to use a 2 port VNA to extract a 2 port model. But this old inductor model of mine is still very powerful

https://youtu.be/4HSWW672vtc

[dazz1]

Hi
I'd be interested in building a LISN as well. 
I'd also be interested in any group buy of a PCB if that happened.  That is likely to be the most expensive part.

Dazz

[T3sl4co1l]

Hi
I'd be interested in building a LISN as well. 
I'd also be interested in any group buy of a PCB if that happened.  That is likely to be the most expensive part.

Dazz

PCBs are stupid cheap, most of the cost will be in the copper (and cores, if used)!

The circuit is so simple, it's hardly worth making a PCB IMHO.  Point-to-point is more than adequate here.

Tim

[floobydust]

I think there's reluctance to make a PCB available, due to safety and liability concerns for a mains-powered design.
Although, an air-core inductor ungluing or going sproing and shorting to something would be the worst. Or a point-to-point design not staying together, during an earthquake 
Commercial LISN's have no agency approvals, no fuse. Just an aluminum box to contain things. These things are used in a gray area, the lab.

[dazz1]

PCBs are stupid cheap, ...

Not where I live.  I am designing 2 boards at present that will be the most expensive component for each project.

The circuit is so simple, it's hardly worth making a PCB IMHO.  Point-to-point is more than adequate here.

Are you including the protection circuitry for the port in your statement?

[T3sl4co1l]

Not where I live.  I am designing 2 boards at present that will be the most expensive component for each project.

PCB Shopper shows five mfg's that will do a typical board for this sort of project (100 x 150mm, 2 layers, 2 oz, nothing else special), <= 10 AUD/ea at qty 10?  Is that not cheap enough?

Are you including the protection circuitry for the port in your statement?

Yes, and filtering -- it's just a ladder network, easily done with SMTs on copper clad.  Would only take me a few hours to scratch one out; or a few bucks first, if I didn't have the utility knife and copper clad on hand.

Now, the enclosure, that's the real expensive item, if you don't have one handy.

Tim

[dazz1]

PCB Shopper shows five mfg's that will do a typical board for this sort of project (100 x 150mm, 2 layers, 2 oz, nothing else special), <= 10 AUD/ea at qty 10?  Is that not cheap enough?

Which is exactly my first point.  A PCB is a good idea and if 10 or more people want boards, the cost per board is very reasonable.
So if there is enough interest to do a group buy on a LISN PCB, I'd be interested in joining.

[CopperCone]

How should i setup my 4 section 50uH inductor for tuning on the vna?

Do i connect em all, put it in parallel with 50 ohms, measure, add the dampening resistors and hf coils then measure again?

I am a vna noob

[Jay_Diddy_B]

I think there's reluctance to make a PCB available, due to safety and liability concerns for a mains-powered design.
Although, an air-core inductor ungluing or going sproing and shorting to something would be the worst. Or a point-to-point design not staying together, during an earthquake 
Commercial LISN's have no agency approvals, no fuse. Just an aluminum box to contain things. These things are used in a gray area, the lab.

Floobydust nailed it. I am happy to share the 5uH LISN because it is normally used for low voltages. I am a little reluctant to share the 50uH PCB because of liability.  I have sent the Gerber files to a few individuals on a case by case basis.

The board is currently 4.5 x 4.5 inches. This didn't matter when I made the prototype for my own use using my LPKF Protomat c60.  If I shrink the board to 100 x 100mm it opens up the possibility of using the very low cost manufacturers.

It doesn't make sense to do a group buy if the board is 100 x 100mm. The distribution cost and effort is higher than ordering them direct.

If you believe a LISN can built with cored inductors, build it as I did. If you are a non-believer and want to do air cored, use the same PCB and wire in the air cored inductors. Doing this you will benefit from the transient limiters and filters located on the board.

If you are the fence, you can build both and compare them. You can also build the 5uH LISN on the 50uH PCBs.

Does this make sense?

Regards,

Jay_Diddy_B

[Jay_Diddy_B]

How should i setup my 4 section 50uH inductor for tuning on the vna?

Do i connect em all, put it in parallel with 50 ohms, measure, add the dampening resistors and hf coils then measure again?

I am a vna noob

Which VNA do you have?

I have been measuring the inductor in parallel with a 50 Ohm resistor. But if you tell me which VNA I might be able to help more.

Regards,

Jay_Diddy_B

[CopperCone]

 :-//I have a 300MHz version of the E5100A


Btw, I read (and suspected) that the parallel resistors may add a kind of HF shunt, and i read that filter designers recommend adding inductors in series with the dampening resistors if degraded hf responce is noted.

https://www.google.com/url?sa=t&source=web&rct=j&url=http://ecee.colorado.edu/~rwe/papers/APEC99.pdf&ved=2ahUKEwj_yOjGsY_aAhUDTd8KHe8EAooQFjAMegQIBhAB&usg=AOvVaw22rdcGWHhoqkIZ98eQiuaw

[floobydust]

Damping resistors every 4 turns, then one oddball at the end on the CISPR air-core inductor.

[G0HZU]

That looks to be along the lines of what I'd expect to see. There are undamped sections between the resistors and some deliberate asymmetry in the structure

[Jay_Diddy_B]

:-//I have a 300MHz version of the E5100A

I am not familiar with the E5100A analyzer. I had a quick look at the manual and could not see how to get into the mode required.

On the HP 8714C



I set the desired frequency range
I do a one port calibration with OSL standards.
I set the VNA for reflection.
The format to impedance
and adjust the scaling.

And you can see the display that I get.

Regards,

Jay_Diddy_B