Author Topic: How do I rate a CT for leakage detection use? Also a DC leak detector coil?  (Read 5644 times)

0 Members and 1 Guest are viewing this topic.

Offline 741Topic starter

  • Frequent Contributor
  • **
  • Posts: 406
  • Country: gb
    • Circuit & PCB Design (small PCB quantities OK)
When there is no imbalance in L, N currents (no leakage) the CT sees no net current. For a major fault, current is large but presumably time-limited by MCB.  If I have 230VAC/32A flowing in normal use, must a leakage detect CT be rated over 32A?

How is a DC leakage detect coil specified, is this a stock item? Basically I understand want to see measurable ferrite saturation, hence a change in inductance, at 6mA.

Offline Gregg

  • Super Contributor
  • ***
  • Posts: 1156
  • Country: us
There are a number of stand alone ground fault sensor current transformers similar to this: https://www.nktechnologies.com/ground-fault-protection/
Eaton also makes them but their website sucks; here is an example on fleabay: https://www.ebay.com/itm/Eaton-Ground-Fault-Sensor-EGF3NCACNET3/283507373364  Some have adjustable threshold limits. 
 

Offline 741Topic starter

  • Frequent Contributor
  • **
  • Posts: 406
  • Country: gb
    • Circuit & PCB Design (small PCB quantities OK)
Thank you:
We need to detect both AC (20mA) and DC (6mA) leakage, the latter relying of change of inductance with DC current. I wonder if the same coil can be calibrated to do double duty?

We do not have to measure normal-mode current with the coil, only leakage.

The spec is for an EV using AC charging, 32A and single or 3 phase (96A).

OpenEVSE use a 20A unit, 1000:1 ratio. Then 20mA leakage equates to just 20uA... :-\

Offline duak

  • Super Contributor
  • ***
  • Posts: 1048
  • Country: ca
741, are you aware of the principle behind the Flux-gate Magnetometer? In brief, a magnetic core is excited by a high frequency sine wave and the average current is monitored.  The current to be measured is passed thru another winding.  Because the core is magnetically non-linear, its inductance varies as a function of the inverse of the instantaneous current and so the average current is a facsimile of the current to be measured.  These things can be quite sensitive, easily measuring mA.  Practical units usually have two cores and separated excitation and sensing windings.

LEM makes current sensors, some of which use this principle.  Here's a link to one of their products that explains it better: https://www.lem.com/en/file/5619/download

Since you want to measure leakage current, you'll want to run the line conductors, but not the earth or safety ground conductor through the core.  This way, the operating currents cancel out leaving the leakage current to be measured.  It should be possible to use a common mode choke with 2 or 3 identical windings for the line conductors and rated for the load current with an additional windings for excitation and sensing.  Most common mode chokes use a fairly high mu material that has fairly sharp saturation characteristics.

Note, the excitation and sense circuitry should be designed to handle the surge energy passed thru the core during a fault.
« Last Edit: February 12, 2020, 01:16:50 am by duak »
 

Offline 741Topic starter

  • Frequent Contributor
  • **
  • Posts: 406
  • Country: gb
    • Circuit & PCB Design (small PCB quantities OK)
Thank you for the info aboux Flux-Gate methods.

The LEM diagrams show me how little I understand magnetics! Fig 5 is especially mysterious/impressive with the winding at RHS of continuously increasing diameter  :popcorn:... LEM do say the FG approach can be made very accurate (more so than Hall effect). In my case I'm not quite so bothered about accuracy: I only want 'modest accuracy' leakage current measurement (5%?, eg 6mA +/- 0.3mA), WRT no leakage current of about zero.

I wonder if I can use Hall-effect in this application? One big issue for us is simply price - leakage detection looks set to use a fair proportion of our parts budget.

PS: What happens in a CT when the secondary current circuit is high resistance, do we see sparks as the PD rises to force the secondary current ratio relation? Or does the  primary coil act to oppose the current in the conductor threaded through it? Presumably if both coils are High Z, nothing much happens and it's like there is no CT.

Online coppice

  • Super Contributor
  • ***
  • Posts: 10034
  • Country: gb
Thank you for the info aboux Flux-Gate methods.
Its surprising how many experienced engineers need introducing to the idea of flux-gates. They are actually quite widely used, yet most engineers have never encountered them either in their courses or their professional work.
« Last Edit: February 12, 2020, 07:54:13 pm by coppice »
 

Offline duak

  • Super Contributor
  • ***
  • Posts: 1048
  • Country: ca
I wish I could remember more of my college electromagnetics.  I don't recall encountering many practical magnetic problems and applications in the courses.  Magnetic Amplifiers were commonly used in Aerospace servo drives because they were small and reliable vis-a-vis vacuum tubes and early semiconductors.  hp made a spiffy clamp on current meter (the 428A) from the 50s to the 70s that could read mA full scale.

A Hall effect device could work but the FG principle is probably more sensitive.  IMHO, the AC excitation and demodulation removes a lot of low frequency noise that a direct reading Hall sensor can't.  A Hall sensor requires a custom core cut to introduce the Hall sensor whereas the FG does not.

About your questions.  Yes, when operating properly, there should be zero flux in the core.  LEM sensors work on the nulling principle where the core's flux is driven towards zero by an amplifier and counter winding.  This gets around the core's non-linearities.  At some point though, the amplifier limits and the core flux becomes non-zero.  I suppose that at some point, the measured current is so high that the counter winding develops enough voltage and current to backfeed and damage the amplifier.  To a certain extent, the core protects the electronics from damage by saturating.  It would be like a capacitor going open circuit or just repetitively limiting current when the voltage across it gets too high.  Hmm, I wouldn't mind one of those.  Usually, they do the opposite by shorting out!

One thing to consider in all magnetics that use a ferromagnetic core is that a sufficiently large current can magnetize the core to some extent.  This doesn't damage the core itself, but it can add a DC offset to any DC measurent.  It also reduces the gain slighty by moving the operating point out along the B-H curve.  The core should be chosen to have high permeability to increase sensitivity and low coercivity to minimize remanent flux.  If the core becomes magnetized, degaussing it should restore proper operation.

Cheers,
 

Offline uer166

  • Super Contributor
  • ***
  • Posts: 1026
  • Country: us
That doesn't sound right, in my experience you want a core with the squarest BH loop possible, and it's excited deep in and out of saturation both ways. Degaussing/remnant flux don't make any sense in context since each excitation cycle the core goes deep deep into saturation anyway, removing all history. At least in the 4.5mA DC detection coils that OP is talking about.

You also don't need 2 cores as per LEM app not if you don't care about induced EMF into the line you're trying to measure, which may or may not matter for OP. All commercial sensors designed for EVSEs (circa $20 a piece) have one core. Some sense residual current via a timing principle (how long it takes to saturate core clockwise vs. counter-clockwise). Others have a peak detector to see where the core saturates, and compensate to make clock-wise and counter-clockwise saturation symmetric (closed loop fluxgate, like LEM). My design measures the shift of the entire BH loop along the H axis when residual DC current is applied via some analog circuitry that I probably can't show here..

In any case expect to spend a couple hundred hours doing research and simulation, at least that's how long it took me to have a workable and cheap solution. This post here: https://www.eevblog.com/forum/projects/arbitrary-(saturable)-coupled-inductors-in-ltspice/ is where I started and it was a massive help in my understanding how these things work.

I suggest you get a few saturable cores like https://www.digikey.com/products/en/filters/ferrite-cores-cables-and-wiring/840?k=Toshiba%20MS and start playing with a function generator and a scope (that's all you need). Also get a few commercial EVSE leakage detectors (like Bender) and measure some waveforms.
« Last Edit: February 13, 2020, 12:26:56 am by uer166 »
 

Offline uer166

  • Super Contributor
  • ***
  • Posts: 1026
  • Country: us
Oh, and the flux in the core is never zero, except at B-axis intersections momentarily as flux swings through the zero. If it were zero, you wouldn't be able to measure any residual current, whether DC or AC!
 

Offline uer166

  • Super Contributor
  • ***
  • Posts: 1026
  • Country: us
Also some more answers:
Thank you:
We need to detect both AC (20mA) and DC (6mA) leakage, the latter relying of change of inductance with DC current. I wonder if the same coil can be calibrated to do double duty?

All AC is DC, if you design a fluxgate sensor that can measure current at a couple KHz bandwidth, it can obviously measure 60Hz AC leakage and DC leakage, you just have to decide how to process it in your code. I would challenge the assumption that you need to measure AC, what are you measuring AC for? The standard specifies that you cannot trip on AC fault, that is the RCCB's job upstream, not yours.

Quote
How is a DC leakage detect coil specified, is this a stock item?

According to: "Residual direct current detecting device (RDC-DD) to be used for mode 3 charging of electric vehicles". buy it if you are making an EVSE for europe.
https://global.ihs.com/doc_detail.cfm?document_name=IEC%2062955&item_s_key=00750712
There are plenty of off-the-shelf sensors.
Quote
I only want 'modest accuracy' leakage current measurement (5%?, eg 6mA +/- 0.3mA)
This is not modest, I'd guess this is not achievable without extreme measures and ideal conditions. Throw a car pulling 22kW in real world and best you can get is maybe 6mA +- 1mA if you really try. Read the standard and do no more than that, tripping at 5 +- 1mA is perfectly acceptable.

Quote
I wonder if I can use Hall-effect in this application?
No, hall effect sensors don't really go that low of current rating, they are better suited to measure high DC (1-1000+ amps) currents. Use fluxgate, every commercial EVSE DC leakage detector uses fluxgate.

Quote
One big issue for us is simply price
Expect to spend >$20 per sensor, if you're lucky it will be able to do both the 20mA AC and 6mA DC.

Quote
What happens in a CT
I would forget the word "CT", current transformers are completely inapplicable here and would just cause confusion. I've had suppliers call our designs "CT" when they are fluxgate devices, and it just causes confusion on everyone's part.

Hope this helps and please update us on the journey!
 
The following users thanked this post: 741

Offline 741Topic starter

  • Frequent Contributor
  • **
  • Posts: 406
  • Country: gb
    • Circuit & PCB Design (small PCB quantities OK)
Re: How do I rate a CT for leakage detection use? Also a DC leak detector coil?
« Reply #10 on: February 13, 2020, 11:40:04 am »
Many thanks indeed for this detailed information.

When you say
Quote
I would challenge the assumption that you need to measure AC, what are you measuring AC for? The standard specifies that you cannot trip on AC fault, that is the RCCB's job upstream, not yours.
but also that
Quote
Expect to spend >$20 per sensor, if you're lucky it will be able to do both the 20mA AC and 6mA DC.
that does confuse me a bit. My understanding now is I consciously ignore AC leakage detection as a design goal.

Re:
Quote
There are plenty of off-the-shelf sensors
is that the kind of thing Farnell, Mouser etc stock - and what is the corrent search term?

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 9321
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: How do I rate a CT for leakage detection use? Also a DC leak detector coil?
« Reply #11 on: February 13, 2020, 01:50:06 pm »
How about a large pair of inverse parallel diodes in the ground line with a 10 ohm or so resistor across them as a shunt? A small fault current will create a small voltage drop across the resistor while a large fault current will be shunted by the diodes. It would be very unusual to have a ground fault in an EV that doesn't also fault to the chassis. A separate ground connection can be used to inject a test current to verify the ground fault sensing works and that the ground connection is good. An AC GFCI/RCD would still be present as a backup.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline uer166

  • Super Contributor
  • ***
  • Posts: 1026
  • Country: us
Re: How do I rate a CT for leakage detection use? Also a DC leak detector coil?
« Reply #12 on: February 13, 2020, 10:05:36 pm »
How about a large pair of inverse parallel diodes in the ground line

A residual current fault doesn't generate any current in the device's ground, this won't pick up anything when someone's getting shocked. Only way is to measure current imbalance in the live/neutral, or the phase1/phase2/phase3/neutral lines to pick it out.

To OP: check relevant standard that you're certifying your EVSE to. In north america, we need 20mA AC detection, in EU, 6mA DC detection. A device that does both can be useful if you're selling it in both regions (edit: what I mean is it can have separate GPIO outputs, one for 6mA DC trip, another for 20mA AC, and you use them selectively). In EU you most likely will only need a 6mA detector. Digikey etc won't have these since it's pretty specialized, you'll have to talk to the manufacturers directly, here's the most popular one I've seen in field:
https://www.bender-uk.com/company/news-item/6ma-dc-sensitive-residual-current-monitoring-for-ac-charge-stations

Based on your questions I wouldn't attempt making one yourself yet, but instead integrate an off-the-shelf sensor.
« Last Edit: February 13, 2020, 10:09:46 pm by uer166 »
 
The following users thanked this post: 741

Offline duak

  • Super Contributor
  • ***
  • Posts: 1048
  • Country: ca
Re: How do I rate a CT for leakage detection use? Also a DC leak detector coil?
« Reply #13 on: February 13, 2020, 10:20:48 pm »
Uer166 pointed out some things that I should have been more clear about.  LEM makes a number of current sensors using different principles of operation.  One is the high precision FG type.  Another is a Hall sensor type where an amplifier and compensation winding null out the flux from the sense winding.  It was these I was thinking about in my reply. 

I have experience with the LEM non FG type from the 80s where I used them in a BLDC motor servo drive.  They had sufficient precision and stability to allow the output current to be set to within +/- 10 mA out of 13 A ie. 0.08%, and usually much better than that.  Field service would occasionally recalibrate a servo drive and find that the zero setting was off by 30 to 50 mA but full scale was fine and that adjusting electrical zero restored the spec.  Subsequent tests showed that if the line power to the servo drive was cut while the motor was turning fast, the current  from the motor backfeeding the drive could leave a slight remanent flux in the current sensor core that manifested as an offset.  If the power to the LEM sensors could have been maintained while the motor spun down, their self nulling would have prevented them from being magnetized.  During power loss the drive was designed to brake the servo motor by rectifying the back EMF from the motor back to the DC link while aggressively discharging the DC link with the braking resistor.  In this mode, an asymmetrical pulsed current was applied to the sensors rather than a symmetrical decaying sine wave.

As uer166 reminds me, an FG sensor drives the core into saturation all the time, so the above effect should not occur.  I have a Tek P6042 current probe that can resolve mA.  It is not an FG design and it has a degauss switch.  It can pick up a DC offset just by opening and closing the probe jaws.   I looked at the hp 428B current probe manual and although it is an FG design, it too has a degausser and hp recomends using it after measuring Ampere level signals.  I'm bringing up the possible problem of a remnant field because if the sensor isn't designed correctly, it could hide a DC leakage current that could be a hazard.  Just sayin'.

Uer166 is absolutely right in saying that the flux is never zero except when changing polariy.  I should have said that the average flux over the excitation cycle is zero and is a part of how the LEM sensor works - look at the last line of the 2nd paragraph on p.5 of the LEM document I linked to above.  Other designs may not do this.
« Last Edit: February 14, 2020, 12:52:46 am by duak »
 
The following users thanked this post: uer166

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 9321
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: How do I rate a CT for leakage detection use? Also a DC leak detector coil?
« Reply #14 on: February 14, 2020, 12:20:17 am »
A residual current fault doesn't generate any current in the device's ground, this won't pick up anything when someone's getting shocked. Only way is to measure current imbalance in the live/neutral, or the phase1/phase2/phase3/neutral lines to pick it out.
I'm having a hard time thinking of a realistic fault with an EV that would cause a ground fault not involving the chassis. All the high voltage wiring is well protected for a good reason.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline uer166

  • Super Contributor
  • ***
  • Posts: 1026
  • Country: us
Re: How do I rate a CT for leakage detection use? Also a DC leak detector coil?
« Reply #15 on: February 14, 2020, 05:30:29 am »
I'm having a hard time thinking of a realistic fault with an EV that would cause a ground fault not involving the chassis.

That is irrelevant, it will fail all (almost all) IEC 62955 tests that you have to do to certify the RDC-DD 6mA detection scheme. The entire point of all these leakage detectors is to protect people/equipment when the isolation in the car fails, if it doesn't work in that scenario then what's the point?

Here's a fault not involving the chassis: a person standing on the ground touches a live conductor of the EVSE cord due to insulation failure and gets a shock. The return path is the earth they're standing on. A correctly designed leakage detector will pick that up, anything involving monitoring the GND path in the EVSE will not.

Quote
All the high voltage wiring is well protected for a good reason.
If that was true there would be no need for any leakage detectors ever, and no need for double insulated DC chargers as well, yet all EVSE in europe have both AC and DC leakage detection, and all DC chargers have a double insulated HF transformer. But again all this is irrelevant, any standards body will immediately fail that design without even testing it.
 

Offline max_torque

  • Super Contributor
  • ***
  • Posts: 1327
  • Country: gb
    • bitdynamics
Re: How do I rate a CT for leakage detection use? Also a DC leak detector coil?
« Reply #16 on: February 14, 2020, 01:01:10 pm »
A lot of the EVSE monitoring capability is mandated to protect against upstream failures (ie the supply side) rather than with the vehicle itself!

https://professional-electrician.com/technical/ev-charging-equipment-the-new-rules-explained/





The choice between developing your own fault current detection systems and buying an OTS component really fall to the choice between cost and liability.  When you buy an OTS solution, you carry across the certifications of that solution, so yes, it costs more, but you have some legal protections.  If you develop your own solution, then you entirely "own" the liabilities for that solution.

So, if you're going to sell hundreds of thousands,then rolling your own looks sensible, as you can afford to amortise the test and certification for your device over many more units, but if you are only makes hundreds or low thousands, then an OTS solution may be a better idea overall
 

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 9321
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: How do I rate a CT for leakage detection use? Also a DC leak detector coil?
« Reply #17 on: February 14, 2020, 01:33:03 pm »
Here's a fault not involving the chassis: a person standing on the ground touches a live conductor of the EVSE cord due to insulation failure and gets a shock. The return path is the earth they're standing on. A correctly designed leakage detector will pick that up, anything involving monitoring the GND path in the EVSE will not.
That would get picked up by the AC RCD.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline 741Topic starter

  • Frequent Contributor
  • **
  • Posts: 406
  • Country: gb
    • Circuit & PCB Design (small PCB quantities OK)
Re: How do I rate a CT for leakage detection use? Also a DC leak detector coil?
« Reply #18 on: February 17, 2020, 12:31:35 pm »
The video teardown in post #16 is narrated by two UK electricians. I took notes, this is what struck me as important -

(1) "OLEV" - A UK grant scheme from the "Office for Low Emission Vehicles". User wants compatibility. A new requirement of the scheme, is that "as from July 2019 EVSE MUST be able to connect to the internet".


(2) TN-C-S is the most common earthing arrangement in the UK  //See also https://medium.com/@Voltimum/the-principles-of-protective-multiple-earthing-pme-c068f2f433ac
Wiring Regulations (18th edition, section 722) FORBID TN-C-S for EVSE outdoor stations - other than for three specific exceptions.
 
  "A PME earthing facility shall not be used as the means of earthing" //Wiring regulations, 18th ed regs, part 722
  However, TN-C-S is an example of a PME system.
      https://www.voltimum.co.uk/articles/do-you-need-install-earth-rod-ev-charge
      https://eocharging.zendesk.com/hc/en-us/articles/360024737931-18th-Edition-Wiring-Regulations
         
Normal practice for outdoor EVSE is to add an earth rod, creating a local 'TT' system. Zappi 2 uses an exception to the TN-C-S rule to save the need for an earth rod.
     
My understanding is TN-C-S has the following features:   
  (a) From the source supply and up to the supplier fuse, Earth & Neutral are one single conductor ('Terre' & Neutral are Combined hence 'TN-C'). 'Terre' = 'Earth'.
  (b) Along the route of this "Combined" conductor, there are "multiple earths". So TN-C-S is a 'PME' system, "Protective Multiple Earth". NOTE: The actual combined Earth/Neutral conductor itself is termed 'PEN conductor' or 'CNE conductor'.

Why is TN-C-S normally forbidden for outdoor EVSE? It is to allow for the case where the supplier's PME combined conductor (called 'PEN' or 'CNE') breaks outside your property. The current return path for both 'line to earth' faults and line to neutral (fault or normal use) is via the combined PEN conductor.

But if the PEN breaks, then even normal-use Neutral flow will go the easiest way to 'Terre' (Earth). Maybe a person is touching the car. The current cannot return via the failed PEN cable though. So it will opportunistically jump via the person into the floor that person is standing on.

* Question: Why is there no requiremnet for a house-local earthing rod in all installations, regardless of outdoor equipment? - the same scenario can occur indoors too. 

Of the 3 exceptions mentioned above, two are pretty rare situations. The 3rd exception allows TN-C-S if the safety RCD cuts "Earth" as well as Live. Zappi 2 uses this method, hence saving the installer worrying how good a local earth they can create, especially regarding dry soil. The narrators like this feature, apparently MyEnergi/Notts University have notes explaining how they do this. It would seem to be basically the CT around the earth lead that trips the 'contactor' in the earth line.


(3) Zappi 2 provides energy mangement. Part of this relies on up to 3 external Current Transfomer inputs (note these can be direct wired or more conveniently the current value can be sent by RF link).
The RF link is "NOT wifi or Bluetooth" for reliability reasons. There is also a CAT5 socket. Unsure if the RF link still used then for CT, probably.
     
There is a danger you can blow the main fuse if other high current equipment is also running. Zappi 2 monitors whole system via CTs [+RF link] etc to ensure no main fuse overload. For instance
    CT 1 on main supply to house  //Main fuse eg 100A
    CT 2 on cable to garage       //MCB rating to garage eg 40A "sub-main line conductor"
Then program Zappi 2 to respect those limits    //Is this via a page on an HTML server inside Zappi 2?
   
* Question: Assuming it is a real pain to replace a blown supplier's fuse, why is there not usually a high-current MCB immediately downstream from the supplier's fuse, thus protecting the supplier's fuse? Eg Maybe something just under the main fuse rating?   
   

(4) When contact weld seen, the PCB mounted DPST relay just up from the LNE input connector disconnects L & E.     


(5) The narrators say Zappi 2 incorporates a class A RCD into its PCB circuitry and a CT. Further, the same CT (with L, N through it) also monitors for DC current over 6mA.


(6) There is also a CT around Live, presumably they measure current draw this way.
   

(7) We know front right is a PCB mounted relay. Not sure what the 2 large black boxes at rear are. Rear centre looks like a transformer. Rear right looks like a relay.


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf