Core Size Selection
After the proper ferrite material is chosen to meet the extended temperature and DC bias conditions needed for this application, choose several size profiles from the vendors’ catalogs so that there will be a variety of choices to achieve all parameters and fit solutions.
Choosing the correct size core to fit into the inner volume of a package or connector case depends upon the wire and buffer material build on the core sets as well as the number of core sets in the package.
It is a good idea to allow a minimum margin of approximately 0.025 inch between the inside wall of the case cavity and the final buffered coil. Assuming that the buffering is about 0.020 inch, we can allow approximately 0.050 inch of margin from the cavity wall to the wire around the core. This can provide a close estimate to choose the closest outside diameter and height of a standard core from a vendor’s catalog.
Once this core has been chosen, the turns and wire size can then be calculated from chapter 4. When these have been established, then the wire build on the core is calculated.
The formulas for these calculations are as follows:
The following formula is used to approximate the wire diameter needed to accommodate a bifilar winding on the core.
The nearest HPN AWG size that will fit into the core ID is:
(Round off this to lowest value)Chapter 6 references formulas used to determine HPN, TPN and QPN wire diameters.
To estimate the coil outside diameter (OD) for HPN wire use the following relationship:
Where:
Now estimate the coil height (H) for HPN wire with:
Choosing either TPN or QPN wire to improve the hipot withstanding voltage, will probably produce a coil of larger dimensions. The additional polyurethane coating on the wire will force this issue. If this is the case, perhaps it is best to use the next smaller diameter wire, if that is within the winding reliability guidelines. Smaller wire can become brittle and break with soldering heat and assembly movements.
Coil orientation in package
After the coils are wound, it is very important to keep the wire lengths between the transformer and the common mode choke as short as possible to minimize EMI radiation and keep the leakage inductance as low as possible for rise time and return loss considerations.
At times, it is difficult to achieve the proper orientation because of the cavity constraints. This may have to be negotiated with the mechanical engineering group designing the package early in the design stage. A compromise may have to be met.
Crosstalk considerations
To provide the best crosstalk in the connectors, it is best to separate the coil sets as far away from each other as possible. This can be a problem most of the time because of the inside cavity constraints and the number of coils needed for the ports.
The winding start and finish window of the common mode choke coil should face away from the common mode choke of the next channel so that there is no linkage of the leakage flux between the channels.
Dress the leads of one channel as far away from the adjacent channel leads as possible to avoid coupling. Any added capacitance from close leads will only add to crosstalk problems.
Using the crosstalk estimating spreadsheet on the CD will provide a guide as to the effect of varying distances between coils or wires.
Common Mode Rejection considerations
To achieve the best common mode rejection, it is best to focus on the wire dress. Keep the wire lengths short but still allow proper stress relief in the leads to prevent wire breakage due to movement. Fine wire can be very brittle, especially after undergoing a soldering operation. The heat tends to destroy the ductility of the wire by changing the copper hardness.
Maintain the common mode choke winding start and finish as far away from each other as possible. A coil cover of 300 degrees or less is an acceptable maximum
The transformer coil should be perpendicular to the common mode choke for the best common mode rejection ratio. Again, keep the wire length between the transformer and the common mode choke as short as possible.
Buffering Materials
It is very important that the transformer coils be coated with a buffering material to prevent any chance of magnetostriction caused by epoxy or the package. It is important, however, to prevent too much buildup of the material. It may be better to thin the buffering material with a proper solvent and use multiple coats on the cores. The addition of buffering material increases the distributed and the interwinding capacitance on the transformer proportional to the dielectric constant of the buffer.
When the common mode choke is coated with buffering material, then the distributed capacitance of the coil is increased by the material’s dielectric constant. Thus, the resonant frequency decreases. This in turn lowers the peak frequency of the impedance curve and may definitely change the EMI attenuation response of the circuit. This is one reason why the choke peak should be adjusted by an appropriate amount above the desired frequency of attenuation.
There are many buffering materials to be found in the marketplace, but the most useful are the silicones. Their expansion rates can be less and the dielectric constants are acceptable.
Isolation Voltage considerations
The main concern with the coil set hipot is found between the primary and secondary windings of the transformer. There is no hipot problem with the transformer secondary and the common mode choke. They are at the same potential. This can be controlled by the choice of the wire insulation thickness. Usually the QPN or TPN wire is thick enough to provide a good margin for the 1500 VRMS required by IEEE.
There should be care given with the wire termination so that the transformer primary wires should not be near the output from the common mode choke. This includes any component or termination that may be close to either one.
Care should be exercised to keep the high potential wires away from any ground connections or capacitors that may be near ground potential. The layout of the PC board traces should also consider this danger.