Tips for making a good Gate Drive Transformer... 1. Use a high permeability Mn-Zn toroidal core. High permeability keeps the magnetic flux lines confined to the core and ensures good coupling between the primary and secondary windings. The lower the permeability of the core, the more tempting it is for some of the flux lines to leave the core and travel through the surrounding air. (This makes the transformer radiate more interference and perhaps more seriously makes its secondary windings more susceptible to picking up external interference from nearby conductors!) 2. Choose a core material that is not too lossy at the frequencies in use. Pulse waveforms with steep rising and falling edges contain lots of harmonics. So the core material should have low loss from the switching frequency up to 10 times the switching frequency. Many ferrites are intentionally made lossy for interference suppresion applications! It is OK to use these material grades provided that they are not excessively lossy in the range of frequencies that we plan to use them. 3. No air gap is required in the magnetic path because this is a low power application. We will ensure that there is no DC current in the windings that could cause saturation. Air gaps are undesirable as they decrease the permeability and cause the flux to leak outside the core material. 4. Choose a toroidal core with a diameter large enough to fit the required number of turns using a decent thickness of wire through core. It is the inside diameter of the toroid that is critical to fit in the required number of turns neatly. Generally bigger cores are used for lower frequencies so that it is possible to achieve sufficient magnetising inductance without requiring too many turns. 5. Choose fat toroids, or ferrite ferrules rather than large open rings. The inside diameter of the toroid should not be excessive. Large, thin rings tend to give higher leakage inductance. They are also unnecessarily bulky for this application. 6. Use the minimum number of turns required to get sufficient magnetising inductance. Adding more turns reduces the magnetising current, but it increases the leakage inductance. Use only enough turns to get an acceptably low droop on the tops and bottoms of the pulses. If you have chosen a suitable core material, you should not need more than 18 turns. 8 turns is a typical value using +/- 15 volt drive. 7. Do not seperate primary and secondary coils around the toroidal ring! Spacing the primary and secondary coils around the toroid gives relatively poor magnetic coupling, and increases leakage inductance considerably! Instead the primary and secondary wires should be interleaved. 8. Use the thickest wire that will just allow you to fit the required number of turns in the winding space around the toroid. It is best to take up as much of the winding space as possible, rather than leaving space between the turns. 9. Twist wires together into a bundle before winding. This twisted pair (or twisted threesome) can then be wound through the toroid the desired number of times to make the transformer. The close physical proximity of the primary and secondary windings ensures very tight coupling. It is also easier to wind this bundle through the toroid than several loose conductors. 10.As an alternative to point 9 above: Small diameter screened cable can be used for the windings. The outer screen of the cable is used as the primary winding, and the inner conductor is used as the secondary winding. If two secondaries are required, then two lengths of screened cable can be used. The outer screens are connected in parallel to form a single primary winding. The inner conductors form the two secondary windings. The author has found this method very effective to achieve the absolute minimum leakage inductance for a particular core size. 11.Don't wind all the turns on a small part of the circumference of the toroid. Spread the combined windings equally around the circumference of the toroid. eg. If each winding has 12 turns, then first twist the wires together then wind the bundle through the toroid 12 times. These 12 turns should then be spaced evenly around the circumference of the toroid at 30 degree intervals. 12.Wind the wire tightly against the ferrite core. Any air gaps between the wire and the core encourage leakage of the magnetic flux from the core. This increases leakage inductance. For the same reasons do not use wire with un-necessarily thick insulation. 13.Keep lead lengths from the pulse-transformer as short as possible. Lead inductance in free space is around 25nH per inch for each wire! So minimising lead lengths helps to keep stray inductance down. It is pointless striving for minimum leakage inductance in the transformer if you have 6 inch leads coming from it! 14.If lead lengths need to be more than an inch, then twist together the leads from opposite ends of the same winding. This minimises the loop area between the "to" and "from" current paths and helps to keep stray inductance to a minimum. It also reduces the tendency for the winding to be susceptible to external fields. 15.Keep the turns ratio close to 1:1 It is hard to maintain good coupling with extreme turns ratios. If the turns ratio must be 2:1, then make 3 identical windings twisted together on the ferrite core. Then connect 2 windings in series to form the primary. This tri-filar approach maintains tight coupling for ratios like 2:1 or 3:1 etc, but it is generally better to stick to 1:1 if you can. 16.Step-down ratios like 4:1 are preferable to step-up ratios. This is because the stray inductance on the primary side is stepped down by the transformer ratio. eg. If you have a 4:1 transformer driving a MOSFET gate, you can afford 16 times more lead inductance on the primary side than on the side connected to the MOSFET. For this reason, such a transformer should be located right up against the MOSFET to minimise stray inductance on the secondary side. The inductance of any flying leads on the primary side is greatly reduced by the effect of the transformer! 17.Paralleling multiple windings can give a further reduction in leakage inductance. This is because a bundle of interleaved primaries and secondaries facilitate better coupling than that achieved with just one primary and one secondary winding. (It is also possible to reduce the total leakage inductance by paralleling several pulse transformers, but beware, this reduces the total magnetising inductance too.) 18. If an E-I, or E-E core must be used, then the primary and secondary windings should be concentric and on the centre core leg. The primary should be closest to the centre, and the secondaries wound over the top of the primary. In cases where "top" and "bottom" switching devices in a bridge are being driven from one gate-drive transformer, the secondary driving the bottom MOSFET should be next to the primary winding, and that driving the top device should be furthest from the core. The secondary winding connected to the bottom MOSFET acts as an electrostatic shield between the high dv/dt present on the top MOSFET and the driven primary winding. Fast transients are capacitively coupled safely to the ground via the bottom device, rather than being coupled back to possibly sensitive drive circuitry. Always use a DC blocking capacitor in series with the primary winding. The high permeability of the core makes "flux walking" or saturation likely if there is any DC component to the voltage applied to the primary winding. DC saturation of the core causes the shape of the drive waveforms to become distorted and is highly undesirable. Fortunately, a DC blocking capacitor between the drive circuit and the primary of the pulse-transformer ensures that this does not take place. (I would recommend a 10uF electrolytic in parallel with a 100nF ceramic to get good current handling and high frequency performance.) The drive signals should also be kept fairly symmetrical and close to a duty ratio of 0.5 Deviating significantly from a duty ratio of 0.5 causes a DC shift in level to appear at the MOSFET gates. This is because DC voltage levels cannot be passed through the pulse-transformer, and are indeed blocked by our DC blocking capacitor! This can mess with the turn-on and turn-off instants if the voltage level shifts by more than a few volts. The leakage inductance of the pulse transformer forms a series resonant circuit with the gate capacitance of the MOSFET. This resonant circuit is shock excited by the steep rising and falling edges that are applied to the primary of the transformer, and the large current pulses that are required to charge and discharge the MOSFET gate capacitance. This resonant action is usually underdamped, and results in a voltage overshoot followed by ringing at the resonant frequency. A small amount of overshoot and ringing is acceptable, however excessive voltage overshoot can exceed the maximum gate-source voltage specification of the MOSFET, and cause damage to the gate-oxide insulation. Likewise excessive ringing is undesirable, since it can cause the MOSFET to re-enter the linear region if the gate voltage rings low enough to start the device turning off. This is very bad! This overshoot and ringing should be damped by the addition of a low value resistor in series with the resonant circuit. This resistor can be placed in series with the primary of the gate drive transformer, or can alternatively be placed in series with each of the secondary windings. There are benefits to both approaches, but the damping action of the resistor is the same regardless of the location. The value of the damping resistor should ideally be sufficient that there is no voltage overshoot when measured at the gate-source connection to the MOSFET being driven. But it should be no larger. Increasing the resistor value beyond this value results in an over- damped situation and seriously degrades the rise and fall times of the gate drive waveform. If the value of the transformer's leakage inductance is known, then the MOSFETs effective gate capacitance can be estimated from Qtot, and the damping resistor can be calculated from the following formula. R = sqrt ( L / C) This equation gives a good starting value for the designer to try. However, the final choice should only be made when the PCB is laid out and the actual MOSFET gate voltage waveform has been checked with an oscilloscope. Connecting the damping resistor in series with the primary winding has a number of advantages. Firstly, it can provide damping for the resonant circuits formed by several secondary windings driving several MOSFETs. The damping resistor also absorbs some of the energy stored in the transformer's magnetising inductance, making life easier for the gate drive circuit. Finally, placing the damping resistor on the primary side, ensures that all MOSFETs gates are essentially very closely coupled via the secondary windings on the transformer. This ensures that they receive exactly the same gate drive waveforms. It prevents shoot- through where top and bottom MOSFETs are driven from the same transformer, and it aids dynamic current sharing if many parallel devices are driven from their own secondary windings. This work is unfinished.