Cancellator: a magic active CM noise canceller to upgrade any brick or module SMPS

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
That would be possible.... in theory at least.

First, you would have to acquire the µA's CM current with a sufficient accuracy, against a much larger DM load current; OK, doable with modern tech (but not necessarily easy or straightforward)

Then you have to generate a counter-stimulus to cancel the signal. A problem is that you cannot use a Y coupling cap larger than ~4.7nF, to keep the leakage within safe limits in case the circuit fails.
But that implies driving voltages of the same order as the offending voltage: a good proportion of the mains voltage.

Again, not impossible: I can design an opamp capable of 800Vpp output, but that is not going to be possible with half a dozen of ordinary components.

An alternative solution would be a low voltage amplifier combined with a step-up transformer, but the bandwidth will be limited.
If it is just the 50Hz that worries you, that approach could be fine, but if your ambitions are greater, only the direct, HV, wideband amplifier will work
 
I have a CM clampmeter.it's just a coil around the N and L wire together. the DM currents cancel ,so only the CM flux adds. shorting the the coil with an I->V converting opamp creates the true current info. opamps like this arrangement for speed ,you can use a CFA for that. so you have to drive currents into a Y cap.suppose the interwinding capacitance is 50pf and 110V rms across it,you should be easily able to obtain the same current with a 4.7nF cap and some drive voltage..
 
I never said it was impossible: just challenging.

For example, suppose your interwinding capacitance is not 50pF, but 500pF to ease matters.
The impedance @50Hz is around 6.3 megohm.

If one full phase is applied to this cap, it will generate a current of 36µA.

If you want to compensate this current with some accuracy, you need a resolution of 1 ~ 10µA, preferably closer to 1µA than 10.

Does your clampmeter have such a resolution?

If it is the case, can you process the voltage and handle closed-loop issues properly?

As I said, nothing impossible, but somewhat challenging, especially if your aim is not 500pF but 50pF.

Anyway, I will think about it and I may come back with a half-realistic sim result -or not, if it is really too difficult-
 
OK, here is the half-realistic result.

A caveat: it is certainly far from optimal, and someone really intent on building it could significantly improve matters, both on the material aspects and the concept (that's less certain though).

I based the (virtual) design on a real, high-quality transformer core: a toroidal audio line Xformer.

attachment.php


I figured out that it would be possible to wind fifteen turns of figure-of-eight mains cable as a primary, and used the existing windings as a secondary.
Not optimal, but this core has a respectable Al of 113µH/t².

I used a zero-flux configuration, to eliminate most of the residuals, and to cast in bronze the compensation ratio, something essential.

With a regular opamp, the zero-flux means difficult tradeoffs to ensure freedom from DC problems whilst avoiding FR quirks.
Not an easy task.
The injection cap (4.7nF Y) also poses problems, as it changes the injector into a differentiator.


Here is the embryo:

attachment.php


L1/L2 is the 15t bifilar primary, I1 is the leakage current source, and B1 (crudely) symbolizes the driving transformer having a 1:10 ratio.
Why is a step-up transformer necessary?
With a 50pF to 4.7nF capacitance ratio, it looks unnecessary, but you have to remember that the driving circuit will need to be referenced somewhere, but you have no clue about where, and even if you had, a general circuit needs to be general, ie. cope with whatever condition is thrown at it (otherwise, use a manual compensation scheme).
In practice, the supply voltage would be derived from a supply similar to the ones described, meaning its average reference would be ~1/2 the mains voltage.

Since it would have to be able to correct anything between the Ph and Neutr potentials, the swing required is at least half the mains voltage (even when no correction is needed).

This means that the stepup is needed, and also the buffer to drive it.
Here, all is made with spice expedients, but the real thing would be more complicated, and have issues, like added phaseshift.

The compensation achieved (painfully) reaches 30dB from 50Hz to several kHz, which should be sufficient for purpose.

Is it worth the trouble and complication? I don't think so, but anyone can have his/her own ideas on the question, and make it into a real project (which this certainly isn't).

A big problem is that the sim only shows the open-loop aspect of things: in the real world, R2 is not going to be negligible, and there will be additional, uncontrolled interactions, unless the driving transformer is changed for an ideal current source, which mean more complications
 

Attachments

  • RCC0.png
    RCC0.png
    163.8 KB · Views: 312
  • RCC4.jpg
    RCC4.jpg
    234 KB · Views: 308
  • RCC2.asc
    4.6 KB · Views: 28
in real life the compensation trick is used, in PV panels that have a large parasitic capacitance to earth. actually in my system I have many thin film panels and on a rainy day the differential trips. the trick is to bridge the differential with a capacitor that generates the same leakage current as the panels do so the differential sees only true error currents. there is an SMA application note about this..
I bought a ground leakage (UNI-T UT251A clamp leaker) meter, just a current clamp across the 2 conductors (L, N) together to verify the ground leakage.
 
meanwhile I see the reference problem. the transformer can be made wide band by shorting it's output and converting the current into a voltage as pure current information.
this trick has also been used in audio to increase the bandwidth of audio line level transformers. I think studer did it and meyer sound.
 
the trick is to bridge the differential with a capacitor that generates the same leakage current as the panels do so the differential sees only true error currents. there is an SMA application note about this..
Yes, I mentioned the possibility of a manual compensation, and that is probably the most effective solution: all you need is a small variac, or even a 47K/3W potentiometer and a Y capacitor of a sufficient capacitance

the transformer can be made wide band by shorting it's output and converting the current into a voltage as pure current information.
this trick has also been used in audio to increase the bandwidth of audio line level transformers. I think studer did it and meyer sound.
The zero-flux configuration used here does exactly that. I has been slightly degraded by a series damping resistor to allow DC stability, but it could be bypassed with a small cap if necessary.

With a discrete frontend, or a suitable type of CFA, the transformer could be tied directly to the input.

This wouldn't solve the reference or variable load impedance issues though, and the stepup transformer cannot work in zero-flux mode, meaning imperfections and complications.
The manual compensation is the way to go if you need it
 
Hi
How would this hold up on a P.M.E. system where there is no "real ground"?
Since we were rewired from the traditional way to p.m.e. we have had a multitude of problems to amateur radio equipment, and awful problems with static electricity on all socket and switch mounting screws. Now, virtually all the smpsus we have generate loads of hash.
regards john
 
I had to search the for PME acronym, I am not familiar with this particular naming, but it is in fact the "normal", modern way to supply the mains, in Europe anyway, and is also known as TN-C-S.

The symptoms you describe (static electricity, ...) should not be related to that particular regime.

Something that can happen is that with some class II devices (=having no earth), you can "feel" the leakage current on metallic parts, depending on the way the mains plug is inserted, but it is of no consequence.

Anyway, the cancellator was designed with class II devices in mind, meaning the presence or nature of the PE is irrelevant.
Would it improve matters? Maybe, in some cases, but given your complaints, I doubt it could solve every issue.
 

PRR

Member
Joined 2003
Paid Member
> PME acronym, I am not familiar with this particular naming

I think some of these fine distinctions, you have to go in the box or up the pole/down the manhole to appreciation the differences.
Earthing system - Wikipedia

"PME" seems to be UK jargon. Depending exactly where the splices go, it seems to be "similar" to how we try to do it in Maine (in the whole USA). There's a dirt-rod on the power poles, there's a dirt-rod at the house(?), and part of the feeder is combined N/G (PE) function.

Oh, I suppose the US way is more like TN-C: N and G are not split until the main fusebox, "Consumer". (And most homes are '2-phase' not 3-phase.)

The main difference seems to be: in some systems the consumer gets Dirt-Rod (Ground, PE) ONLY from the company, in others there is a dirt-rod at the customer. In almost all cases the company is full of dirt-rods. In a system with Multiple Earths, the ground currents can be unpredictable.

This may matter to amateur radio which can really depend on ground currents to launch longer wavelengths. It shouldn't have a large effect on a domestic hi-fi.
 

Attachments

  • TN-C-S-PME.gif
    TN-C-S-PME.gif
    28.8 KB · Views: 110
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.