Unscreened Output Cables

[Cable Man]

Hi all,

 

Can any one comment on this claim from New Zealand about being able to use unscreen cables between the drive and motor and still be EMC compliant ?

 

BR: Cable Man

[marke]

Hello Cable Man

 

Welcome to the forum.

 

This is a simple question, but has a complex answer.

 

Firstly, All VFDs generate noise and the level of noise generated is dependent on a number of factors.

 

The noise is generated by the switching action of the output transistors that are used to develop the sinusoidal PWM waveforms.

These transistor switches effectively switch the output phases between the positive and negative DC Bus rails.

Every time a switch operates, a burst of noise is released. The energy and frequency spectrum of that noise packet is a function of the time taken for the switch to operate.

The sudden change in voltage on the output terminals of the drive is directly connected to the motor terminals, so the motor and the cables between the VFD and the motor become an important part of the overall noise emissions. The switching voltages are connected to the frame of the motor via the capacitance within the motor, so the frame of the motor becomes "live" with noise voltages.

 

There are two types of noise emissions to consider, there is radiated noise and conducted noise. Radiated noise tends to be higher in frequency than conducted noise and is "radiated" from the cables (acting like antenna).

Screened cable reduces the amount of radiated noise by the action of the screen providing a Faraday shield to the high frequency noise. The level of attenuation is very cable dependent with some cable providing much better results than other cables.

The level of HF noise is dependent on the VFD design, particularly on the switching speed of the output transistors, physical layout or the output stage and internal filters. If the drive generates lots of noise, it is harder to screen than a drive that has a low noise output stage.

The conducted noise is largely determined by the return earth path between the VFD output and the motor. The noise generated by the drive is carried out to the motor and must return back to the drive output stage. The noise at the motor will return to the drive via all possible paths. If the impedance of the earth return is high, the noise return via alternative paths will be much higher than the return directly to the drive. This increases the conducted noise emissions.

 

At high frequencies, the impedance of a conductor is related to the surface area of the conductor rather than the cross sectional area of the conductor. To ensure a low impedance earth return, the earth conductor must have a high surface area. This can be achieved by using a flexible woven braid rather than a standard circular cable.

A screened cable offers a Farady shield to reduce radiated emissions, and a woven braid to increase the surface area of the noise return path. This is effective in reducing the radiated and the conducted emissions.

 

A screened cable will provide increased reduction in noise emissions relative to a non screened cable.

 

If the output noise from a VFD was low enough, then it would be possible to achieve acceptable results without using screened cables, but if the output is non sinusoidal, the noise will be further reduced by the use of a screened cable.

 

Best regards,

Mark.

[marke]

Hello Cable Man

 

On further investigation of these "claims", it has been suggested that the design of this product differs from most modern VFDs in that they employ soft switching of the output IGBTs and also employ dv/dt filtering on the output of the drives to reduce the dv/dt to less than 800V/uS.

 

Soft switching of IGBTs and MOSFETs is employed to reduce the rate of change of current (di/dt) and under severe fault conditions, this reduces the overshoot on the DC bus when switching off a high fault current.

The overshoot voltage is a function of the di/dt and the inductance of the DC bus. V = L di/dt.

 

Under normal operation, soft switching results in very high switching losses and is generally avoided in order to minimise the switching losses.

If we consider a single phase output stage, there are two switches in series across the DC bus with the AC output taken from the center point of these switches. In operation, the switches are turned ON one at a time. There are essentially three states, switch 1 ON and 2 OFF, 2 ON and 1 OFF, and both switches OFF. (Both switches ON is a fatal condition shorting out the DC Bus)

If we consider the first state with Sw 1 ON, current will flow through this switch to the negative bus from the motor and there will be a low voltage across the switch. When we turn this switch OFF, current will continue to flow as the field in the output circuit (and motor) collapses. As soon as the current through the IGBT begins to reduce, the voltage on the IGBT will jump to the positive rail with the balance of the current flowing through the reverse diode across the top IGBT. As the current through the bottom IGBT reduces, the current through the diode increases. For the duration of the switching period, the voltage across the bottom IGBT is equal to the full DC Bus voltage. The net result is that the switching losses in the IGBT are approximately 300 times the ON state losses for the duration of the switching time.

If the IGBT switches OFF in 1uS due to soft switching, and hard switching yields a switch OFF time of 100nS, then the switching time is ten times longer and the switching losses are ten times higher. - this is more heat to get rid of and reduces the drive efficiency.

 

Soft switching will reduce the switching noise on the DC bus due to the Ldi/dt reduction, but the output voltage dv/dt will not be significantly altered as the voltage will be driven by the inductive load.

 

The addition of a dv/dt filter on the output is designed to reduce the rate of rise of voltage on the output. It is possible that the soft switching is used to enhance the operation of the dv/dt filter.

 

A reduction in dv/dt will reduce capacitive charging currents and this will reduce EDM (bearing damage) and also reduce the bandwidth of noise that is generated. To gain a significant improvement, the dv/dt should be in the order of volts per uS. At around 700 V/uS, the noise generated in the output circuit could still be significant.

 

Best regards,

Mark.