Good SI, PI and EMC require this most of all...

07 Nov 2018

Good SI, PI and EMC require a proper, grown-up understanding of electricity...
Instead of what circuit designers are taught!

As electronic designers, we were taught the children’s version of electricity at school, college, university, etc. This is the version that pretends (just as the SPICE simulator does) that electricity flows as little packets of charge totally inside conductors.

And we were taught that only the signals’ send paths mattered. If we even thought about return paths at all, we just assumed they sort of happened somehow, and they couldn’t be very important because our circuit design text books barely mentioned them. Even now, our circuit diagrams (schematics) only show send paths!

If we took a course on Maxwell’s Equations, we learned lots of exotic maths, but nothing about how it related to actually designing electrical or electronic products, equipment, systems, etc.

And if we took a course on RF design, we learned all about designing with matched-impedance transmission lines – S parameters, Smith charts, and all that – but learned nothing about the RF behaviour of send / return conductors that were not matched transmission lines even though these are the majority, in most applications.

If we took a course on RF Antenna design, we learned all about designing antennas for specific radio, TV, radar, etc. applications and services – but learned nothing about the antenna-mode behaviour of all other conductors (which are the vast majority, in most applications).

We now need a proper understanding of electricity!

2.5GHz wireless datacomm’s are ubiquitous and becoming inadequate; even the smallest cheapest microprocessors use clocks and signals with bandwidths of several GHz whether we need such high frequencies or not, and 5G cellular communications will use the spectrum between 18GHz and 100GHz to handle near-future datacomm’s needs. The useful radio spectrum is now considered to extend up to 3THz (i.e. 3,000 GHz), which we will be using in the fullness of time.

Pretending that electricity flows as little packets of charge totally inside conductors does not allow us to design modern electronic technologies so that they even function correctly (SI and PI issues), never mind have good EMC.

So: here’s a very brief overview of a proper understanding of electricity
I will expand on each of these issues, with practical examples, in future blogs

  1. All electrical signals, data, power, wireless datacomm’s and broadcasting, radar, etc., are really electromagnetic (EM) waves that propagate at the ‘speed of light’.
  2. The ‘the speed of light’ is lower in all materials other than air or vacuum. This means that – at any given frequency – the EM waves propagating in such materials have shorter wavelengths than they would have if they were in air or vacuum.
  3. We generate and receive EM waves (which we call signals, data, control, power, radio, TV, radar, etc.) using two conductors, usually identified on a circuit diagram as send and return.
  4. Stray (leaked, parasitic, sneak, coupled, etc.) waves and currents always occur, but their send and return current paths may not be obvious. Despite the fact that they cause many EMI problems, we never show them on our circuit diagrams.
  5. When we are more concerned with the patterns that EM waves make in 3-D space, than in their propagation, we call the patterns ‘EM fields’.
  6. All currents – whether signals, data, control, power, etc. – always flow in closed loops. This includes all stray (leaked, parasitic, sneak, coupled, etc.) currents, too.
  7. All currents preferentially flow in the loops with the lowest overall impedance. These current loops are not limited to the conductors we draw in our circuit diagrams – they can include displacement currents (i.e. electric flux coupling) and/or magnetic flux coupling through the air and other insulators
  8. All currents flowing in conductors flow closer to their outer surfaces as the frequency increases. This is called the Skin Effect, and it is the reason why we can use conductors as EM shields. However, conductors don’t care what names we call them, so all conductors – whatever their function in our circuits, even signal or power wires and PCB traces – have skin effect whether we want them to or not.  
  9. There is a perfect correlation between a conductor’s surface currents and the EM fields in the insulating media around the conductor (usually PVC, or air). We can say that the surface currents create the near-fields, or that the near-fields create the surface currents – either/both statements are equally true. So, we can we can choose to visualise/work with either the surface currents, or the near fields – whichever is most appropriate for the issue we are dealing with.
  10. All conductors have series impedances for AC currents, even superconductors with absolutely zero series resistance. So, whenever two or more circuits share any conductor – for example those we might call earth, ground, chassis, 0V, DC power rail, live, phase, neutral, etc. – these series impedances cause noise to couple between them. Calling a conductor ‘earth’, ‘ground’, ‘chassis’, etc. does not endow it with magical properties! When an AC current flows in it, an AC noise voltage inevitably arises. This is called common impedance coupling, and it always happens because everything has impedance.
  11. The EM waves propagating along their intended conductors have both electrical and magnetic field components (its where the word ‘electromagnetic’ comes from).  
    They spread around in the space around the send/return conductors, depending on the physical arrangement of those conductors (closer send/return conductor spacing generally creates more compact fields). Other conductors exposed to these electric and magnetic fields pick up a proportion of them, and suffer noise as a result, Crosstalk is a typical example of such noise coupling.
  12. EM waves in the air or other insulators have ‘wave impedance’: the ratio of their electric to magnetic fields.
  13. Send and return conductors have ‘characteristic impedance’: the square root of the ratio of their mutual inductance to their capacitance, per-unit-length.
  14. Changes in wave or characteristic impedances reflect some of the propagating EM waves in the air or in conductors respectively, causing signal integrity (SI) and power integrity (PI) problems in circuits, and causing resonances in conductor structures. These reflections make every pair of send/return conductors (including those carrying stray currents), into ‘accidental transmitting/receiving radio antennas’. (Note that send/return conductors that are carefully designed as ‘matched transmission lines’ along their entire length don’t have significant impedance discontinuities, so help maintain good SI and PI, and are relatively ineffective as ‘accidental antennas’.)
  15. Resonating send/return conductors can have impedances between very low indeed (µW) and very high indeed (hundreds of kW or more). This means that even very low-resistance conductor pairs (such as thick bars of copper) can behave almost as open-circuits at certain frequencies, depending on their dimensions.
  16. Conductive structures such as planes and enclosures can experience ‘accidental’ structural resonances, which makes them behave as various types of ‘accidental’ antennas. The magnetrons in microwave cookers are an example of intentionally designing a metal enclosure (a cavity) as a resonant antenna, to help couple energy into food (in this application).
  17. All 2-dimensional and 3-dimensional conductive structures behave as accidental antennas due to the shapes and dimensions of their structures, and at frequencies where their shapes and dimensions are comparable with wavelengths – they can resonate. The lower the resistances of the conductive parts forming the structure, the higher the ‘Q’ at such resonances. So, at these structural resonances, sturdy metal structures with very low resistances in their parts can suffer overall impedances that can be very low indeed (even µW), or very high indeed (even hundreds of kW or more). This is true whether we call these structures ‘ground’, ‘earth’, ‘chassis’, shield, frame, 0V, etc.

« Back to Blog