Staggered 50uH and 250uH inductor design for LISN

[dazz1]

based on CISPR and the info provided from the good members of eevblog.. i put together this schematic for a bench top LISN rated to 25A.  the top half of the schematic will be free wired and the bottom half will be on a PCB.  i still need to assign part numbers.  schematic is posted as a DRAFT for comment and is not meant for construction.

please feel free to comment 

next i'll be working on the inductor build

Hi
Just some questions.
Why only 4 turns between damping resistors?  Have you modeled 6 or 8 turns between resistors?  It might save some work winding the coils.

Why 2x 45MHz filters in series?  Would one be enough? Without a buffer in between you may end up with an unexpected filter response.  Have you modeled this?

From a reliability (and safety) perspective having 20 caps in parallel reduces reliability by a factor 20.    Reducing the number of caps to 10 will double the reliability and reduce the parts count.

I think it would be prudent to consider adding a fusable link (2x) between the switch S1 and the parallel resistors/caps.  In addition use a high enough power rating on R102-105 to prevent them vapourising before the link blows.  This would also require wider PCB tracks.  Have you modeled the fault current?

Dazz

[Jay_Diddy_B]

slloyd and the group,

I have had a quick look at the schematic in an earlier message. I think the capacitors I marked S should have a good safety rating and the capacitors in the section marked N can be normal capacitors.






Regards,

Jay_Diddy_B

[slloyd]

thanks for the comments dazz & Jay. 

the resistor damping on 4 windings (and 2 and 6) just comes from CISPR16.  someone on this thread explained why there is value in making the two odd-ball winding choices of 2 & 6. 

i didn't design the filter so i can not explain the philosophy of the design choices like 2 filters.  the 2nd filter is like $1 extra in parts so not a big deal.  two stage filter i guess for better filtering.  if they interact with each other that would be bad.  i haven't simulated it myself yet but i will.  jay already did in this thread and it shows flat response, so to a 2nd order approximation it was good.

i'm a little apprehensive about adding components such as a fuse (polyfuse?) in the signal path.  i'd have to simulate what the parasitic elements of this fuse would affect the measurement.  but as a take away from your comment i will add a note "do not operate switches under power". 

Jay, i see your point.  the filter caps are not likely to be under extreme stress unlike C3 which is a little closer to the action and thus more likely to fail

[slloyd]

i was looking at some wire today from which to make the inductor.  in terms of skin effect, maybe the frequency is high enough to warrant using multistrand wire as compared to solid core.  but lots of LISNs made with solid magnet wire. 
 
some #12 cable type THHN  the core is 2.05mm solid copper and it looks like 0.5mm of insulation.   such a winding using only the insulation itself to get the pitch (i.e. winding is tight with no spacers) would have to be wound on 3.5# PVC since it has a favourable outer diameter to get the ~62uH that the reference inductor had (presumably it is 50uH inside the box).  parameters like this:

[T3sl4co1l]

Q matters very little, this is not a power reactance application.  You can use whatever wire you like.

Tim

[dazz1]

i'm a little apprehensive about adding components such as a fuse (polyfuse?) in the signal path.  i'd have to simulate what the parasitic elements of this fuse would affect the measurement.  but as a take away from your comment i will add a note "do not operate switches under power". 

Jay, i see your point.  the filter caps are not likely to be under extreme stress unlike C3 which is a little closer to the action and thus more likely to fail

If one of those caps connected to line fails to a short, something is going to act as a fuse.   Adding a proper fuse will provide protection without adding significant parasitics.  You will get more parasitics through the switch etc. 

I would use the same high voltage caps for all C3s to ensure symmetry though the signal paths, but I would reduce the number to a total of 10. If you model these then add one cap to see what difference is makes to the output. I think you will find that 10 is more than enough.

Dazz

[slloyd]

i'm totally OK with adding fuses up at the front end near the AC connector.  will act faster than my house breaker that probably won't trip before something burns.  maybe a panel mount reset able fuse

[dazz1]

Hi
Do you have 25A breakers on the power circuits in your switchboard?  If not, they are likely to trip before a 25A fuse.  You will need to choose protection that won't trip on inrush but will still provide close protection.   You might end up with  a high current HRC fuse plus a circuit breaker.  Your aim should be to minimise the total energy input after a fault condition. 

I think you need to model the currents with different faults to figure out the fault currents.  There is a risk that a fault to short on a C3 cap won't blow a 25A fuse on the mains input.  It is probable that a leaking C3 cap will cook the various downstream components (including the pcb) without ever tripping a 25A fuse.    The output is accessible to human touch so additional protection is justified.

I haven't done the calcs but I would expect a fuse rated to a few hundred mA would suitable to protect the switches and filter.



Dazz

[slloyd]

what would really be needed is an "L" shaped trip curve with a 200mS delay.  that's long enough to allow any inrush and short enough to trip before anything burns.  that shape of trip curve comes with ground fault protection breakers (i'm not talking about GFCI which will trip due to the leakage of the LISN), this type of breaker is not common for residential.  so.. as a solution and more clean, add a current sensor and some circuitry to open a contactor; circuitry powered from AC input.  could be done, i'd guess would add about $30 to the total cost for one-off pricing from digikey. the contactor would have to be rated for opening under fault current. for anyone that wants to add "advanced" features, this might be a nice option.  another nice option is remote control of the LISN   

for me, i'm trying to keep it simple. 

[floobydust]

the unmeasured line should be 50ohm terminated

Yes, I corrected the Compower LIN-115A LISN schematic for R6. It is 51R termination (qty. 10 of 510R resistors), but after the capacitors/switch.
https://www.eevblog.com/forum/rf-microwave/50uh-and-250uh-inductor-design-for-lisn/msg1454810/#msg1454810

[dazz1]

Hi
For motor VFDs which have similar input to a LISN (large inrush current to charge the hi volt caps), it is common to see a high current HRC fuse in series with an MCB.  The fast fuse handles dead shorts and the relatively slow MCB handles over load.  Regardless of whether the OP has a 25A supply available at the house, if the LISN is rated for 25A, then the protection should match.  A 25A supply will have a low impedance so fault current can be very high.   

You should model inlet currents at normal operation and then model fault and overload currents then match these to the protection curves.

The protection should be aimed to minimise Volts x Amps x time to minimise the total energy injection into the LISN under fault or overload conditions.  This is true at the power inlet and the through the measuring ports where humans may touch something connected through to line voltage.   

If inrush current is a problem then one option is to include a delay relay on the line input that charges the caps from Line through a resistor before switching and shorting the resistor for normal operation.    There are disadvantages and cost that make this option undesirable.

Dazz

[CopperCone]

monitoring inrush is definatly interesting from a systems engineering prospective. fuck the standards they are widdled down to nothing, if they were any tighter companies would probably literally torch whoever is making the standards

but if you are just in it for the money i guess its fine? im already fighting the planet

[T3sl4co1l]

Fuses burn much faster than breakers, yes.

Magnetothermal breakers may operate faster under fault conditions, but likely enough I^2*t will still be delivered to melt the fuse (even if not clearing the fault in the process).

Capacitor inrush is not much of a problem here.  Bulky loads -- with electrolytic capacitors -- consume far greater surges, for a cycle or two.  The film caps equalize in a small fraction of a cycle (exercise: how long is it actually?), and don't charge up (like electrolytics do), they're constantly cycling up and down, drawing nonzero AC current, which is of course nowhere near the magnitude of fault current, which is fine.

Tim

[slloyd]

I have had a quick look at the schematic in an earlier message. I think the capacitors I marked S should have a good safety rating and the capacitors in the section marked N can be normal capacitors.

C3 are "in series with" the mains.  the caps are not across the line and not line-to-earth.  some X2 safety caps are rated for "in series with" the mains and some are not.  i think i'll chose the safety caps here which are rated for "in series with" the mains.

but 25nF is not a common value.  10x27nF or 5x47nF  hmmm...... which to chose

[slloyd]

https://www.niccomp.com/products/catalog/NPXH.pdf  option for C3.  though i'd rather have a digikey option.  datasheet doesn't mention "in series with" but google search the part and it does. 

[slloyd]

I corrected the Compower LIN-115A LISN schematic for R6. It is 51R termination (qty. 10 of 510R resistors),

i wonder why quantity 10?  it is very low power like 10mW on a single 50 ohm resistor

[slloyd]

schematic still in progress as attached... working on assigning the part numbers.

[floobydust]

I corrected the Compower LIN-115A LISN schematic for R6. It is 51R termination (qty. 10 of 510R resistors),

i wonder why quantity 10?  it is very low power like 10mW on a single 50 ohm resistor

I believe using ten smaller parallel resistors or capacitors is to get wider bandwidth as a single large through-hole part has a lower SRF and more ESL.
Some LISN designs use two parallel capacitors (one small+R, one large) to get flatter response.


I use Epcos B32926C3825M000 for the 8.2uF 305VAC X2 capacitor Digi-Key 495-75768-ND huge too at LS=37.5mm W41.5mm x D20 × 39.5mm H but smaller than an AC motor starting cap.

[slloyd]

that makes sense.  but i was wondering specifically about the 50ohm termination resistor.  nothing measuring that one.. maybe you don't care so much the flatness. 

[floobydust]

I find a few LISN resistors can get a huge wallop if you switch in at peak line voltage.

In series with the 0.25uF capacitor and mains, is 50R termination resistor and the first attenuator resistors.
I don't believe a little 0805 can take that, see your R102, R103, R104, R105. This might be why 10 resisistors are used vs one.

If I recall, a gas discharge tube was used by some manufacturers to clamp that, but it is too slow to ionize and pretty much useless.

You can do a LTSPice sim of switching the line/neutral or out/in transient limiter to verify.

[dazz1]

Hi
GDTs are commonly used in radar to protect the sensitive receiver from the high power transmit pulses.   That application requires a fast response and recovery. 

Dazz

[CopperCone]

Yea im stuck on designing the couplers for my isn.

Not sure how i want to protect them. I think bat diodes and a gdt and small fuse.

I want to maintain signal integrity as much as possible and to defend the reciever, but i dont care if the internal electronics explode it is in 3/16 inch steel box

I may make armor of some kind for the difficult to wind inductors but i am not terribly concer ed about some film caps meeting the reaper

[CopperCone]

Also can someone think of a circuit that would ensure the worst possible switch in states so the bad transients can be reliably analyzed rather then having to turn the fucking thing on 500 times to get data?

I thought maybe a zero crossing detector to find the lowest ac voltage, then a delay for 1/4 or 3/4 cycle depending on if neg or pos polarity is required.. But what kind of switch?

Obviously i want to use a trigstron or other tube but what semiconductor is good for this?

Triac? Special diac? Some kind of transmission gate?

If i build it i think i would like to use something simple.

And a polarity switch to see what happens if the neutral or the hot is connected first to test symmetry

Maybe possible to time a relay on a zero crossing? Seems dificult. Dont know what would be most realistic but also accurately timed

[floobydust]

I have seen one LISN where the attenuator resistor got damaged (high value) and everyone thought passing EMC was a piece of cake, readings were quite low

If you look at the switch-in impulse of 50R resistor charging a 0.25uF cap to 170V, it is short duration and very peak high current, I think 160mApk for 20usec; 3.5mJ and 23W.
Any mains transient also hits the resistors.

When you pulse overload a resistor very briefly, it's a gray area because it's 100x rated power dissipation and the element can get damaged. Vishay is the only engineering group that gave me decent answers. I couldn't get pulse-overload data from other resistor manufacturers.

In the old days, you used a large carbon-comp through-hole resistor there and didn't worry about it.

GDT's are slowest, several usec to trigger depending on the voltage rise-time. They are also lowest capacitance, great for RF transceivers and lightning protection.
But here I think the GDT will trigger too late to protect the resistor, and the firing voltage is still substantial at say >90V. That's a bit much on a 50R resistor...

[CopperCone]

Anyone looking for a beast resistor?

What did vishay say? I figure you need a 8w resistor. I thought about using a coaxial dumky load here lol