EMC issue 1

[autophysics]

Hello everyone,

with have a set of coils for inductive power and data transmission. These are arranged as two sets of concentric primary and secondary coils in front of each other. The coils differ in diameter and strand arrangement. The data signals operate in the MHz, the power signal in the kHz range.

In a initial design it is found that:

- if the receiving power coil is not wound as a single layer, the receiving data coils capture only a weak signal.
- all data coils must have a certain distance amongst them, as to capture a decent signal at the receiving data coils.

Aim is, to get low loss for the power transmission, as well as low loss/noise for the data transmission.

Any pointers ?

P.S. So far a LC-circuit atached to the power coils has been considered to dampen-out the influence of the power coils on the (high frequency range) data coils.

 

[VEBill]

Signals in the "MHz" (3-30 MHz) band can travel *thousands of miles*. So if you're having difficulty making them jump a few feet, you must be doing something wrong.

 

[autophysics]

VE1BLL,

ok, there is a factor x1000 between the kHz and MHz signal wave length. Still the field produced by the sending power coil penetrates the receiving data coils ? Could you be a bit less enigmatic, please.

 

[VEBill]

Is the problem that the kHz power signal is simply swamping out the MHz data receiver? Can you 'scope out the signal being presented to the MHz data receiver? Is the poor-old MHz data receiver being fed a huge kHz squarewave that is making the input go non-linear and creating broadband noise?

If so, then you either need to figure out a clever way to remove the kHz noise (nulling and filtering) from the data signal, or change your data transmission approach completely.

A different approach might be to (for example) simply stand-off with the radio link (as opposed to putting it into the exact middle of the worst possible location).

MHz is radio. Radio can span large distances. That's the point I was hinting at before.

Modern unlicensed radio links such as (for example) Bluetooth could make a 10m link without any problem whatsoever (through the air, staying away from the power coils).

Another approach might be to use an IR LED instead of MHz coils.

Filtering is obvious, but you have to reduce the signal enough to keep the amplitude in the linear region in subsequent active circuits otherwise it'll generate broadband noise.

Nulling might be the best first firewall. Figure out some arrangement of the coils so that the first-order coupling is reduced. Maybe orient them at 90° to each other.

Maybe use a differential coil-pair scheme so that the power signal cancels out while the data signal sums. But this is getting close to invention-on-demand via the Internet.

 

[autophysics]

VE1BLL,

thank you for the instructive overview on alternatives. I were looking only at the proposed design (concentric coils along a center line). So that brought fresh air already.

"...have to reduce the signal enough to keep the amplitude in the linear region in subsequent active circuits otherwise it'll generate broadband noise...":

if you had any readable info specific to this aspect, like a helpful link or pdf... Filter syntesis and design isn't my speciality. So directions to get me on course are appreciated.

 

[VEBill]

As soon as 'things' (*) become non-linear, then 'all h_ll' breaks loose in the frequency domain. At that point, it's too late for filtering because the unwanted signal (your kHz power) has generated harmonics and intermodulation products that overlap your data (MHz) spectrum.

* Most commonly active circuits, but even connectors in extreme conditions [google: Passive Intermodulation (PIM)].

That's why (given your proposal) an inherent nulling technique would be the ideal FIRST step. This could be as simple as using a figure-8 shaped, or dual, differential coil arrangement so that the power signal mostly cancels out on the data lines. Then you heavily filter the residual kHz noise, hopefully leaving a reasonably clean signal.

But an alternate approach might be better. By the time you got the 'concentric transformers' trick working, people would buy you a beer for accomplishing such a difficult method, and then they'd ask you why you didn't just use an easier approach.

 

[autophysics]

VE1BLL,

the signals are likely single frequency (1x spectral peak) and the circuit linear. If it isn't, for a given input amplitude, one may adjust the RLC values, not ?
If so - with no other disturbances present - the signal would produce neighter harmonics nor intermodulations.

Nulling, differential coils - inherently seem to hold a circle game promise, with a couple of transmit/receive coils at different frequencies, M's..

Best guess seems a shielded separation of the signals w/o filters, not ?! This wouldn't get me the beer. But (if a valid option) probably peace of mind earlier, with greater ease ;-)