berkeman said:
Thanks folks for the replies. I need to be careful and circmspect for a bit about the patent and application of the technology for now. My employers would own it. I'll post the full deal when it clears one way or the other. Pretty cool applicaion of Physics in EE, actually.
The patent finally issued, so I can mention what we were doing:
US patent 7,969,270
http://patft.uspto.gov/netacgi/nph-...rke+AND+AN/echelon&RS=IN/berke+AND+AN/echelon
We ended up buying Intusoft's Magnetics Designer software, but honestly, it wasn't all that easy to use or useful in the end. Building up prototypes with standard cores/bobbins and then soft-tooled cores and bobbins was more useful than the simulation software in this case.
The challenge was to build a communications transformer that was inexpensive, but had good rejection of external magnetic field noise. Our customers build our communication circuits (transceivers and transformers) into their devices, and often there are sources of B-field noise nearby. Either the device's own DC-DC converter power supply magnetics, or external motors, etc., nearby.
We had solved this problem in the past by building toroidal comm transformers with both the primary and secondary wound as full-circumferential windings, with a dielectric spacer between the two windings. When an external B-field passes through the toroidal core, the voltage induced in one half of the windings is opposed by the voltage induced in the other half of the coil's windings, and you get no net receive voltage induced by the B-field. In practice, the windings are not totally uniform, so there will be some small level of receive noise voltage induced by the interfering B-field.
But winding lots of turns on a small toroidal core is an expensive process, and for the newest generation of comm circuits, we wanted to cut the cost of the transformer in half or better. To do this, we needed to use a more standard construction technique, and we ended up with a more standard plastic bobbin on a UU-core construction. The trick was how we distributed and connected the windings to achieve good rejection of interfering B-fields.
We did this (the main part of the patent) by splitting the primary and secondary coils into two halves each, and interleaving them on the core in a special pattern. The patent describes in minute detail how the placement and connection of the "split double bobbin" arrangement of the coils helps to cancel out the receive noise voltage from external B-fields that flow through the core in different directions, but I'll leave most of that extra reading to those who are interested (see the patent write-up). Here are a couple of figures that will help you to see how the split double-bobbin arrangement and connection scheme provide the noise voltage pickup cancellation.
Was a fun little project! (it was a small part of a much bigger project involving a new mixed-signal ASIC transceiver)
