Staggered 50uH and 250uH inductor design for LISN

[dazz1]

With a mains LISN, the priority is RF signal integrity, not safety. ...

I am not a lawyer but I believe the health & safety legislation in every civilized country would take exception to that statement.  If there was an accident involving equipment made/supplied by you, that statement would bite you very hard.  The priority should always be safety first.

[CopperCone]

yea some kid might find it if you leave it laying around on the street

[T3sl4co1l]

With a mains LISN, the priority is RF signal integrity, not safety. ...

I am not a lawyer but I believe the health & safety legislation in every civilized country would take exception to that statement.  If there was an accident involving equipment made/supplied by you, that statement would bite you very hard.  The priority should always be safety first.

How could you possibly ever get agency approval for a high voltage generator, a transient generator, an EMI generator, a... ?

Test equipment is exempt from such restrictions for good reason.  They must be used by qualified technicians, in suitable lab conditions, for the same reason!

Example: the local Cooper/Eaton guys have a high voltage test lab, 20kV just sitting out in the air and all that.

How is that acceptable?

It is carefully caged behind a full height chain link fence, with several redundant interlocks.  On top of that, they are very particular about their procedures: measuring voltage twice with a meter, to ensure a de-energized state; shorting anything that needs to be guaranteed safe while handling; wearing heavy rubber gloves; only allowing qualified personnel into the area (i.e., I watched from behind the fence), and so on.

For the LISN, there isn't much you can do that fits within ordinary safety rules.  You do want to address ground leakage and galvanic isolation, using an isolation transformer.  This also allows you to plug it into a RCD/GFCI circuit.  Fusing isn't needed on the network because the isolation transformer or mains circuit has it.  The filtering somewhat addresses mains transients as well (which is where the 2.5kV hazard comes from), and if you provide any additional protection, that helps too.

Tim

[Jay_Diddy_B]

The one thing I haven't seen so far on this thread, and in my view (& experience) something that is essential - a discharge mechanism on the power input.

I built a LISN a few years back for my own lab, when I finished using it I switched it off at the wall socket and pulled the plug.  Whilst handling it to put it away I touched the plug pins and got a real belt off it from the internal capacitors (obviously switched it off at peak cycle).  Since then I fitted a mains relay inside that disconnects the L&N input and shorts the internal L&N to earth via a pair of 7W 4K7 resistors.

Hi,

I have discharge resistor in the line voltage version of the LISN across the input capacitors, from Line to ground and neutral to ground.

Regards,

Jay_Diddy_B

[dazz1]

Example: the local Cooper/Eaton guys have a high voltage test lab, 20kV just sitting out in the air and all that.

How is that acceptable?

It is carefully caged behind a full height chain link fence, with several redundant interlocks.  On top of that, they are very particular about their procedures: measuring voltage twice with a meter, to ensure a de-energized state; shorting anything that needs to be guaranteed safe while handling; wearing heavy rubber gloves; only allowing qualified personnel into the area (i.e., I watched from behind the fence), and so on.

And I am reasonably confident that they will have those procedures documented, site inductions, safety audits and all the other documentation that goes with modern H&S requirements in high risk work environments.  I can guarantee their documentation does not include the statement "safety is not a priority"

For the LISN, there isn't much you can do that fits within ordinary safety rules.  You do want to address ground leakage and galvanic isolation, using an isolation transformer.  This also allows you to plug it into a RCD/GFCI circuit.  Fusing isn't needed on the network because the isolation transformer or mains circuit has it.  The filtering somewhat addresses mains transients as well (which is where the 2.5kV hazard comes from), and if you provide any additional protection, that helps too.

Tim

You and others have described quite a few things that fit within ordinary safety rules.    Typically H&S legislation has a "all practical steps" or similar approach to managing risk.    That is not the same as eliminating risk. 

[floobydust]

Found pics of Comtest TTi1600 LISN.

250uH inductor is large EE ferrite, and the four stacked (to 50uH) air-core parts which remind me of loudspeaker cross-over network inductors, like Solen.

It was a repair, open 2R2 fusible resistor to the GDT. The unit uses lithium batteries to energize signal relays for limiter diode bias.

edit: fixed battery purpose

[Jay_Diddy_B]

Found pics of Comtest TTi1600 LISN.

250uH inductor is large EC ferrite, and the four stacked (to 50uH) air-core parts which remind me of loudspeaker cross-over network inductors, like Solen.

It was a repair, open 2R2 resistor to the GDT. The unit uses lithium batteries to energize signal relays.

The manual, including the schematics, for this LISN can be found here:

http://resources.aimtti.com/manuals/LISN1600_Instruction_Manual.pdf

From the schematic in the manual, the lithium batteries are used to bias the clamping diodes. Biasing the clamp diodes will reduce the capacitance, and therefore the impact on high frequency response.

I believe that the main inductors are single layer solenoids.

L2 and L3 don't look right on the schematic, the value is too low.

Regards,

Jay_Diddy_B

[floobydust]

Yes they are low value- L2, L3 limiter inductors possibly 1mH not 1uH each?
Seem to be Toku 8RBSH series, no datasheets but 8RBS range is 0.1mH-15mH and here I am again seeing ferrite-core parts run past SRF of 2.1MHz (#262LY-102K) to 30MHz.

[CopperCone]

how would you guys position the 1.8mH inductor?

I'm not sure what to do. I got ready made 1mH inductors, I could put two of em in series. But, should I put them up right, or in parallel with the 50uH long air inductors? If I place them upright, then they will be in parallel with the 250uH inductors, but physically kinda far away. Or I can put them up on an odd angle, but this seems stupid.

And with the 1.8mH inductor, is there any reason to combine two of them? I figure this would be counterproductive to the self resonant frequency, and its better to just use two separate ferrite cores (vs removing the cap of the inductor with heat, and re-winding it with a single winding.. this would have a lower SRF... so other then it looking nice, I don't see a benefit, right?).

Should I maybe shield them? They are the bobbin type.

[Jay_Diddy_B]

how would you guys position the 1.8mH inductor?

snip...

Should I maybe shield them? They are the bobbin type.

Can you share your schematic? Where is the 1.8mH inductor in the LISN schematic?

The 50uH inductor is the main inductor that helps define the low frequency impedance, the 250uH inductor is essentially a line filter.

Regards,
Jay_Diddy_B

[CopperCone]

look at the tekbox schematic earlier in this thread. It is a optional switched in inductor for ground isolation between the tests so you don't have a group loop between the equipment and the ground connection of the LISN. It is directly between the ground output of the LISN and the DUT. I think it might be a useful sanity check, especially since I don't have a test lab with a very good earth ground, well throught out mains wiring, shielded room, etc.. neighbors can always buy some kinda jamming device (by this I mean some shitty PSU) on alibaba so it might be nice just to switch it in to see if there is some kinda problem going on.

https://www.eevblog.com/forum/rf-microwave/50uh-and-250uh-inductor-design-for-lisn/?action=dlattach;attach=404645;image

I am adding it but with a low voltage GDT in parallel with it so it can hopefully trigger the breaker if the chassis on the DUT becomes electrified with HF due to some kind of failure mode (at 60KHz it is already 2000 ohm impedance so it wont blow a breaker or significantly discharge its self. The main danger is a switching converter heatsink or some shit touching the DUT box (like we all know ATX supplies have live heatsinks.... ) .  I don't want some 20 amp device going through the BNC/SMA cable into my expensive spectrum analyzer! or to be zapped during hookup if I accidentally have the inductor switched in, though it should be shorted when something is being plugged in for safteys sake, but mistakes happen).. of course this is less of a concern in a test lab where there is dedicated equipment probobly always plugged into the LISN at its own station, but I can't really justify that kind of shit and it will be used sparsely and probobly put away on a shelf after every use due to space concerns so there is alot of room for error).

[floobydust]

Lifting the EUT (RF) ground, I wouldn't implement that. Many reasons, but the EUT will then seek capacitive coupling to anything like your bench, field wiring etc. and readings would be flaky, if a 3-prong (grounded) power cord to EUT vs. double -insulated device.
Instead I would incorporate a splitter to look at CM verses DM. This is what I use for designing/evaluating SMPS input filters where I need weigh X and Y cap, CM choke values.

[T3sl4co1l]

Measurements are, of course, meant to be done in a reliable and consistent manner, so that capacitive coupling doesn't matter, it's just part of the test (or not).

Given that CISPR 14 isn't very specific or consistent on how certain things are handled, like power cables...

Usually, a nonconductive table is used, so that there is relatively little capacitance to ground, except through the cable.  And what capacitance there is, is (relatively) strongly coupled to the antenna, so will show up on the vertical polarization measurement at low frequencies (~30MHz), while higher frequencies are just whatever because of random antenna elements formed from cables and such.

Tim

[CopperCone]

Your not really lifting rf ground with most instruments though. Most oscilloscopes and spectrum analyzers and emi recivers will have chassis grounded coaxial connectors.

So if you dont isolate the ground, you have a ground loop. The test does not measure 60hz, so the ground loop here is ok, but the blocming impedance begins to become significant where the test begins.

I think thats the rational for it. There is no real point to running something with a lisn without instrumentation connected, so i think it will be ok.

As for cm measurement i saw plans online how to make a network that would plug into the neutral and live sampling ports and combine them to give you a common mode response.

I think the 1.8mH inductor is more of a sanity check for mains grounding issues?

[CopperCone]

Also, as for using magnetic cores for the inductors, wont they act like nonlinearmixers, particulalry when current draw is heavy? When they are near saturation there might be significnant imd. So things like power up transients might have garbled frequency information, that is if you care.

[Jay_Diddy_B]

Also, as for using magnetic cores for the inductors, wont they act like nonlinearmixers, particulalry when current draw is heavy? When they are near saturation there might be significnant imd. So things like power up transients might have garbled frequency information, that is if you care.

This picture will partially answer the question. Two pairs of LISNs are being used to measure the same power supply. The power supply has spread spectrum modulation. One of the measurements is using a pair of Com-Power LI-550A LISNs and the other measurement is using a pair of Jay_Diddy_B 5uH LISNs:



The peak measurements are particularly close.
The cored inductors include an air gap or a distributed air gap, so they are fairly linear, and therefore there is little IMD. Since the emissions are normally 'rich' in harmonics, the small contribution from the inductor non-linearity is tiny.

Remember the purpose of the inductor is to isolate the DUT from the power source. So long as the impedance of the inductor is 'large' compared to the impedance of the emissions the inductor will have little or no effect on the measurement.

I would do a comparison of my line voltage LISN, 50uH, but I don't have a commercial LISN to compare against. I could measure a standard load, say an Agilent DSOX 3k scope and other people with access to a commercial LISN post a comparison?

Regards,
Jay_Diddy_B

[CopperCone]

Maybe a stupid request, but can you hook up two non synchronized spread spectrum power converters to thkse lisn?  Reason i ask is because i am curious if two higher power signals will show evidence of mixing.

I would expect this problem to be worse if something is going on during something like a load switch, i.e. dcdc converter 1 is operating normally and a heavy capacitive load is switched in where there is a capacitor being switched in by a load switch or relay.

I dont know anything about magnetic mixing though, maybe its more present in frequencies over 3MHz?

I figure there has got to be a pretty good reason why people go through the trouble of winding these air core behemoths.. No one would wanna increase product size and stuff without good reason..

It seems that they at least make the 50uH bit air cored with more leway on the 250.

Did you shift the measurements you took for clariyy or does your lisn have a different noise floor in the low frequency range?

[Jay_Diddy_B]

Snip...

Did you shift the measurements you took for clariyy or does your lisn have a different noise floor in the low frequency range?

The Jay_Diddy_B 5uH was just the PCB, not enclosed, this is probably why the noise floor is higher. It doesn't matter unless the noise floor is close to the limit.

Regards,

Jay_Diddy_B

[CopperCone]

Dynamic range is always nice.

[dazz1]

Hi
I think the significant unknown here is how well the selected ferrite coils perform as the load current approaches or exceeds saturation, especially with peaky loads.
When they saturate, do they significantly affect LISN readings.

The key advantage of air cores is that they don't saturate. I presume CISPR circuit was designed taking into account the air coil parasitics.

Dazz

[CopperCone]

Well the 250uH coil is not well defined. We have seen product examples going both ways, with big air coils and large EI core transformers. All commercial products do seem to use air coils for the 5 and 50uH.

I also read some where that the 5uH coils are sometimes used for very heavy loads. I think this means that if you need to run something at ridiculous like ?200 amps, you use a 5uH lisn because the series resistance of the LISN is small vs the large voltage drop across 250/50uH... This was on some military website though.

[CopperCone]

What variation on impedances on the output coupling network did you guys see?

The designs in this thread are 470nF, but I see the military standard is like 250nF.

I figured smaller capacitor =  better but this shit has gotten pretty complicated already.

So long you calibrate it yourself, it does not really matter does it?

[Jay_Diddy_B]

Hi,

For typical 5uH LISNs the coupling capacitor is 100nF. If you have a 50 source and a 50 load, 100 total.

The -3dB point is 1/( 2 x Pi x RC) = 15.9kHz

In practice you have a low impedance source, the emissions are low impedance, the -3dB point is 31.8kHz

For a lot of 50uH LISNs the coupling capacitor is 250nF resulting:


1/ (2 x Pi x RC) = 1/( 2 x Pi x  100 x 25E-9) = 6.36kHz 

making it suitable for the military specification.

If you are doing commercial equipment where the emission measurements start at 150 kHz you can use a smaller capacitor.

This is NOT calibrated out. You build the LISN according to the relevant EMC specification and you use it.

The only calibration is to make sure that impedance is correct and there is typically a +/- 20% window around the nominal.

You can measure transmission if the LISN has a 10dB attenuator or filter.

Regards,

Jay_Diddy_B

[charliedelta]

Why guess about LISN construction? The are precisely defined and   full details are given in the CISPR documents about their construction. Same goes for the MIL and IEEE standards.

The most important detail that is often missed in the construction is the 430 ohm resistors across  every other few turns on the 50uh inductor. These are clearly required in the CISPR designs.

[CopperCone]

I think its better to put inductors in series with the parallel resistors then to go to an undampened coil but I need to make sure