In the July-August-September issue of The Product Safety Newsletter, I discussed “working voltage,” and its relevance to the safety of the equipment. Included in that discussion was a discussion of transient overvoltages.
I said that primary circuits normally have transient overvoltages, but that “Equipment secondary dc circuits are examples of “…circuits that have virtually no transient overvoltages.” I further stated, “It is an abnormal condition that secondary dc circuits have transient overvoltages.”
One reader said that such statements are only true for secondary circuits separated from primary by an earthed metal screen, or for secondary circuits having one pole connected to earth. He then said that a floating secondary circuit is subject to full mains transients.
In this issue, I’ll examine how transformers behave when subjected to transient overvoltage, both magnetically and capacitively. I’ll also examine the functions of rectification and capacitive smoothing when subjected to transient overvoltage. And, I’ll describe some transient overvoltage tests on a 40-watt transformer, a 20-watt unregulated dc power supply, and on two 50-watt switching-mode power supplies.
Let’s first examine how transformers behave, magnetically, in the presence of transient overvoltages.
Power transformers are designed to transform power at 50 or 60 Hertz or both. Transformers operate as transformers when the flux is within the limits of the design. Flux increases with input voltage. Flux increases with frequency. If the flux increases too much, the core will saturate, and there will be no magnetic coupling from primary to secondary.
Transient overvoltages are comprised of high voltages and high frequencies. The transient voltages often are greater than the rated transformer input voltage. The frequencies contained in the transient are much greater than the rated transformer frequency, usually in the hundreds of kilohertz to low megahertz range.
Because the core is optimized for the power line frequency, it is very lossy at higher frequencies. In fact, the core is so lossy in the presence of transient frequencies, the device ceases to be a transformer.
Due to the high frequencies, the core saturates at relatively low transient voltages. (In fact, at high frequencies, the core will saturate at voltages much lower than the rated input voltage of the transformer.) When the core saturates, there is no magnetic coupling primary-to-secondary. With no magnetic coupling, there is no coupling of the transient voltage to the secondary winding.
Likewise, due to the high voltage of the transient, the core will saturate. When the core saturates, there is no coupling of the transient voltage to the secondary winding.
So, transformers simply do not operate as transformers in the presence of transient overvoltages. There is no output of the transient from the transformer due to the operation of the device as a transformer.
Next, let’s examine the effect of capacitance between input and output windings of a transformer when the transformer is subjected to transient overvoltages.
Two conductors separated by an insulator constitute a capacitor. In an isolating transformer, the input and output windings are separated by an insulator, and therefore constitute a capacitor.
Similarly, there is capacitance from each winding to the core. Recall that transformers are wound with insulated wire. Each turn of a winding is fully insulated from adjacent turns.
At any particular point between two windings, a single turn is adjacent to one or more single turns in another winding. Each set of two single turns in different windings constitute the plates of one of the many capacitances that exist between the two windings.
In between the individual turns constituting one plate of a capacitor are a number of other turns of the winding. These turns constitute inductances and do not contribute to the interwinding capacitance. So, the distributed capacitors are connected by means of inductors.
At high frequencies, the various capacitances are distributed throughout each individual winding. The various incremental capacitances are connected by means of the turns of the winding. The turns constitute inductances. So, the distributed capacitors are connected by means of inductors as well as by the capacitance between layers of a single winding.
Series inductors attenuate high frequencies. Shunt capacitors attenuate high frequencies. Series capacitors couple high frequency to the adjacent winding. This results in a very complex high-frequency circuit.
Since the series capacitors between windings are just one part of the complex high-frequency circuit, very little transient overvoltage energy is transmitted to the secondary windings. Most of the energy is dissipated within the primary winding, or from the primary winding to ground.
At the high frequencies found in a transient overvoltage situation, the interwinding capacitance cannot be represented as a single capacitor from primary to secondary. Likewise, the capacitance from a winding to the core or to the earth cannot be represented as a single capacitor.
So, while there is capacitance from primary to secondary, there are also inductors in series with those capacitances, and there are capacitances from primary to core and primary to ground, all of which serve to highly attenuate the energy coupled to the secondary windings.
Now, let’s examine rectification and capacitive smoothing of the secondary output voltage when subjected to transient overvoltages.
The action of a full-wave rectifier is such that no matter the polarity of the input voltage, only one polarity is present at the output. This means that transient overvoltages (exceeding the peak-to-peak mains voltage), regardless of polarity or phase position with respect to the ac input voltage, will appear at the output of the rectifier (assuming the rectifier diode has sufficient frequency response to turn on during the transient).
The energy in the transient is then stored in the smoothing capacitor and should raise the capacitor voltage depending on the value of transient voltage, the duration it is above the voltage on the capacitor, and the available current.
Since the transient voltage is attenuated by the transformer, and since the time is very short, and since current is small due to the inductance of the transformer, the voltage increase on the smoothing capacitor is almost imperceptible.
The same action takes place in the off-line rectifier-capacitor circuits of switching-mode power supplies. The input EMI filter attenuates the transient overvoltage. The off-line rectifier usually is slow, and may not turn on during the transient overvoltage event. If it does turn on, or is late in the event, only a small amount of the energy is dumped into the capacitor and there is negligible voltage change.
Off-line rectifiers have lots of capacitance across the junction, but that capacitance is very small compared to the bulk capacitor. In a series circuit of two capacitors, the voltage division is inversely proportional to the value of capacitance. Therefore, there is a small proportion of the transient overvoltage across the large capacitor (the bulk capacitor) and a large proportion across the small capacitor (the rectifier).
Primary and secondary dc circuits have negligible transient overvoltages due to transient overvoltages on the power line.
To confirm these hypotheses, I tested a 40-watt transformer, a 20-watt transformer-rectifier with smoothing capacitor, and two 50-watt switching-mode power supplies.
The two 60-Hertz transformers were of triple-flange bobbin construction. The switching-mode power supplies use reinforced insulation between primary and secondary.
I injected 1.2 x 50 microsecond transient voltages onto the power line for each device. I monitored the output with a scope (through an isolating amplifier to eliminate any affects of connecting the circuit to ground through the scope).
I started with 500 volts, line-to-neutral, and worked up to 2.0 or 2.5 kilovolts. I also applied the pulse
line-to-ground.
With 1.5 kilovolts input to the transformer, the transient output was about 70 volts. The transient output voltage remained about 70 volts regardless whether the pulse was applied line-to-neutral or
line-to-ground, and regardless whether the output was grounded or floating.
With up to 2.5 kilovolts applied to both the trans-former-rectifier and the switching-mode power sup-plies, there was no more than 1 volt change for no more than 2 milliseconds in the nominal 30-volt and 25-volt outputs, respectively.
CONCLUSIONS
Transient overvoltages are not magnetically coupled to the output windings of 50-60 Hertz transformers.
Transient overvoltages are capacitively coupled from primary to other transformer windings. The magnitude of the transient across the output winding is a function of the capacitance between the windings, the capacitances of all windings to the core, and the capacitances of all windings to ground.
Since the transformer output transient overvoltage is capacitively coupled and not magnetically coupled, the value of the output transient voltage is dependent on transformer insulation construction, and independent of the output winding voltage.
At least for triple-flange bobbin construction, the magnitude of the transient across the output winding is largely independent of whether the output winding is grounded.
The magnitude of a capacitively-coupled transient across the output winding of a triple-flange bobbin-constructed transformer is largely independent of whether the transient is applied pole-to-pole or
pole-to-ground.
Rectification and capacitive smoothing of an ac waveform containing a transient overvoltage virtually eliminates the transient from appearing on the dc voltage.
For the evaluation of insulation (spacings) in primary and secondary dc circuits, the value of “working voltage” determines both clearance distances and creepage distances, whereas in primary ac circuits the value of transient overvoltage determines clearance distances and the value of working voltage determines creepage distances.
ACKNOWLEDGMENT
Thanks to Tinny Srinivasen, Western Transformers, Portland, Oregon, who provided me with the technical details of transformer operation in the presence transient overvoltages.
Thanks also to Daven Tester, Nick Manwell, and Kevin Cyrus, all of Hewlett-Packard, for help in testing the hardware.
Richard Nute is a product safety consultant engaged in safety design, safety manufacturing, safety certification, safety standards, and forensic investigations.
