Current sense transformers on ETD type bobbins?

[treez]

Still incomplete drawings. Where's the return path?

The return path is as I explained in post #32. This is a CST that will be used in the drain connection of the power fet of a boost pfc converter.

Please find UCC28070 datasheet with CST pictured on first page..
https://www.ti.com/lit/ds/symlink/ucc28070.pdf

 

[FvM]

I don't ask for the function of the current transformer, I'm asking for actual geometry.

 

[treez]

I don't ask for the function of the current transformer, I'm asking for actual geometry.

I would have to send the PCB layout file for people to see the full primary current loop. In any case, it is the same in each case of the example pictures, so effectively cancels the requirement to see it exactly.

I don't see what's the point of distinguishing transformer leakage and wiring inductance. Electrically they are indistinguishable.

The thing is that we need to reduce that inductance, as we do not want this inductance to ring with the mosfet drain capacitance and produce overvoltage at switch off of the fet. We will reduce the pcb trace inductance of the mosfet loop by tight layout...we also wish to reduce that part of the loop inductance that comes from the leakage inductance of the CST. We wish to do this by suitable winding arrangement of the primary and secondary turns on the torroid core.

 

[FvM]

I would have to send the PCB layout file for people to see the full primary current loop. In any case, it is the same in each case of the example pictures, so effectively cancels the requirement to see it exactly.

More a question of principle arrangement than exact layout files. But anyway, this sight "effectively cancels" my ability to continue the discussion.

 

[treez]

As a result, you'll get the lowest leakage with the secondary winding concentrated under the primary loop.

Well I thank yourself for your opinion stated a few posts back. I am trying to find corroborative literature to this, but cannot. I tend to entirely agree with yourself here though, it just kind of makes good sense.

The other point is about the "twisting" brought forward by Andre Luis, and depicted in my digram of post #39.
I am wondering if the twisting can possibly reduce the leakage inductance seen from the primary of the CST.

I think all would agree that we are into the deep secret territotory of transformer manufacturrs here, which we have to get into ,
1...because at this high current level, there is nothing in stock, off-the-shelf, for our high current level.
2...the transformer manufacturers will not wind to a spec for just a few prototypes, (at least not any time soon) and we need to go through the prototype stage first.

 

[andre_luis]

which of the following two CST's has more leakage inductance, as seen from the primary?...the one with the primary and secondary bunched together, or the one where the secondary is spread evenly around the torroid core?

View attachment 120790

A priori, this should not matter in terms of magnetic coupling if the transformer as a whole ( core, windings ) is operating under acceptable conditions, not so subjected to saturation or skin effects. However, in general, to decide the suitable wiring arrangements, we must take into account all devices around it, as well the predominant orientation of the routed tracks nearby. If you have not yet such information, focusing on the original question, I would adopt the second schema, due this option tends to reduce the maximum scattered level of electromagnetic field which would otherwise be condensed at a single region.

By the way, as certainly is your case, you must be careful regarding to the fact of the primary wire is rigid, being risky to crush the secondary windings right below, breaching its varnish insulation.

 

[FvM]

As previously stated, an accurate prediction of transformer leakage inductance can be only achieved by a 3D field solver or empirical measurements.

In an early design stage, a rough estimation can be sufficient. The key is to understand the properties of basic conductor geometries, e.g. two-wire line, stripline, wire above a plane, circular and rectangular wire loops. Knowing their specific inductance numbers gives you a first idea of expectable design peformance.

Twisting a two wires doesn't reduce the wiring inductance as such, but it can assure a minimal distance and thus loop area.

 

[andre_luis]

I was sure of that...

I am wondering if the twisting can possibly reduce the leakage inductance seen from the primary of the CST

But after this statement, I fear that have made a wrong assumption...

Twisting a two wires doesn't reduce the wiring inductance as such

Anyway, the following analysis is why I still believe on that.

Considering that the expected inductance of the bifilar parallel wires would be something like that:

• L = u.A/l ; were A is the loop area between both wires

Looking at the Inductor formula :

• Φ = L.I ; were Φ is the magnetic flow generated by the current flowing across the loop area

Twisting the wires, the I component inside each spin'ed loop tend to produce an infinitesimal magnectic field dΦ flowing at an opposite direction of the neighbor’s loop, like that:

Therefore, the final inductance as a sum of each field, should be reduced.
That’s not right ?

 

[FvM]

Therefore, the final inductance as a sum of each field, should be reduced.
That’s not right ?

It's right. But, in a typical two-wire the wire distance is too low to get a considerable inductance reduction. You get about the same nH/cm number for a twisted or untwisted pair.

 

[mtwieg]

To add to the discussion on twisted pair, the twisting would only decrease self inductance if the turns per unit length were so high that the distance between twists were lower than the spacing between the two wires (like in the above figure), which is impractical. Twisting the wires is much more effective at reducing mutual inductance between the pair and other circuits at moderate distances away (farther than the twist spacing).

 

[andre_luis]

Twisting the wires is much more effective at reducing mutual inductance between the pair and other circuits at moderate distances away (farther than the twist spacing).

...as well reducing noise susceptibility, as mentioned before. In general current transformers are placed near to the switching modules at the power section of the cabinet, which is a region exhibiting high level of electromagnetic fields.

 

[treez]

In relation to this, if we look at the datasheet for the 81xx-RC series of common mode chokes, (by Bourns) we can see that in actual fact, having primary and secondary totally separated from each other, on either side of a torroid, still gives amazingly good primary to secondary coupling.
Take for example the 8120-RC common mode choke which has a single coil inductance of 2.4mH , and a leakage inductance of just 9.6uH. –That’s a coupling factor of 0.998…that’s very tight coupling. The situation of the single_turn_primary current sense transformer still worries me though, since the single primary turn seems completely “adrift” from the torroid. It just doesn’t seem like a turn in the normal sense, and I fear a deterioration in coupling in reality will occur?

Bourns 8120-RC common mode choke datasheet:
https://www.farnell.com/datasheets/1910174.pdf

 

[D.A.(Tony)Stewart]

Thing is, if the loop area is large, you are probably stuffed whatever CT you use, high power switcher layout is ALL about minimizing current loop areas.

I recommend picking (or specifying) a part that meets the required spec and not worrying too much about how the part manufacturer gets there as long as they do meet spec.
The nice thing about this approach is that you as the design engineer do not need to be a magnetics design and magnetics DFM guru (Transformer companies have those guys), you can treat the details of core behavior and winding methods the same way you treat the silicon lattice constants when you pick a chip (Interesting, but fundamentally somebody elses problem).

Regards, Dan.

I agree with Dan, trying to reinvent the wheel costs more in getting a product to market and then you find out the ferrite has microphonic resonances that cause issues and becomes a major recall for quality.

You can always repurpose the design with a well placed gap to create leakage or large loop magnetic fields to treat Malaria.
https://www.washington.edu/news/200...key-to-malaria-treatment-uw-researchers-find/

 

[treez]

Thanks,
Thing is, there is little available off the shelf at this current level, and winders won’t rush to do this kind of low volume work for a couple prototypes.
Which of the attached diagrams of current sense transformers do you think has the least leakage inductance as seen from the primary?

 

[FvM]

Please review my comment in post #47. Wire diameter definitely matters for loop inductance, you can be pretty sure that the thick wire version has lower leakage.

You still seem to ignore that current flows in a loop, respectively the drawing is incomplete up to meaningless if you don't include the current return path in some way, e.g. as a ground plane or strip.

 

[mtwieg]

In relation to this, if we look at the datasheet for the 81xx-RC series of common mode chokes, (by Bourns) we can see that in actual fact, having primary and secondary totally separated from each other, on either side of a torroid, still gives amazingly good primary to secondary coupling.

Likely because it's a common mode choke, and uses a very high permeability core with no gap. The same can't be applied to an inductor used for energy storage.

This is likely why current transformers are often implemented with their own separate core. That way its core can be optimized to give excellent coupling in the CT, which isn't plausible when using the core of the inductor.

Has anyone here actually suggested determining what level of leakage inductance or coupling is acceptable? If this is just for line current shaping or current mode control, then I doubt you need very tight coupling. For transition mode controllers a large amount of leakage can be tolerated.

 

[treez]

ok thanks, as you say, this is for a current sense transformer which often use high permeability torroids with no gap.

 

[Easy peasy]

In relation to this, if we look at the datasheet for the 81xx-RC series of common mode chokes, (by Bourns) we can see that in actual fact, having primary and secondary totally separated from each other, on either side of a torroid, still gives amazingly good primary to secondary coupling.

I'm afraid the above might be misleading to newbies, amazingly good? what is that? coupling of 0.999? more likely the leakage is 15-20% of the inductance in one coil, this fact is exploited by designers of mains filters to give differential impedance...

For CT's poor coupling can be acceptable, as long as its repeatable and the freq of interest doesn't change, then it is a linear system with output proportional to input, it is the magnetising inductance of the CT that gives most of the error...

 

[treez]

I'm afraid the above might be misleading to newbies, amazingly good? what is that? coupling of 0.999? more likely the leakage is 15-20% of the inductance in one coil, this fact is exploited by designers of mains filters to give differential impedance...

The coil inductance is 2.4mH and the leakage is 9.4uH. That's a leakage of 0.998 coupling.

 

[FvM]

Whatever the specific leakage and coupling numbers of the Bourns common mode choke are, I doubt that they give much insights for the current transformer leakage discussion.

A coupling > 0.99 can still involve an inconvenient amount of leakage inductance. Using a high µr core increases the magnetising inductance but keeps the leakage inductance almost constant. So it can give a nice coupling number.

Large magnetising inductance is an important performance parameter of current transformers, because it reduces the burden and winding resistance induced error. It's also wanted in a common mode choke to improve the low frequency attenuation. As a side effect, these components achieve impressive coupling numbers with a considerable amount of leakage.

If you are primarly interested in leakage inductance, you should "assume away" the core in a first step. Because the leakage inductance is usually almost identical to that of the respective air core transformer.