Author: ABBdriveX

Home  ›  Posted by ABBdriveX  ›  Page 12

AQ: The basis of rating a NGR in electrical system

NGR stands for Neutral Grounding Resistor. When an earth fault current occurs on a plant, assuming that there is no external device presented to limit the earth fault current, the magnitude of the earth fault current is limited only by the earth impedance presented between the point of fault (to earth) and the return path (typically a star point of a transformer). If the earth impedance is low (type of soil being one of the reason amongst others), the fault current magnitude can be significantly high, and if left unchecked could damage the primary equipment. It is therefore mandatory that the earth fault current be limited to a suitable value, which is typically the rated value of the plant as a thumb rule. Why use the rated value? Because the plant has been designed to carry the rated current continuously.

Let’s take an example: say you have a transformer 60MVA, 132/33kV Star-Delta transformer. It is required to calculate the value of NGR to be connected to the zig-zag transformer on the 33kV Delta. the value of the resistor required to limit the earth fault current to the transformer’s LV rated value is (33 x 33) / 60 = 18.15 Ohms.

(Earth Fault current limited to rated value = (60 x 1000) / (1.732 x 33) = 1050A) When you go to a supplier you might find he supplies only 20 ohms resistor (as you might not get the exact value that you have calculated theoretically). No problem, use the 20 ohms and calculate what your new value of earth fault current would be (33 x 1000 / (1.732 x 20) = 952.6A, which is less than the transformer’s rated LV current. So you’re safe. This is how I would go about. In fact I would go a step further and introduce a safety factor of 20% i.e. I’ll bump up the value of the resistor from 20 ohms by an extra 20% and buy a resistor/ NGR of 1.2 x 20 = 24 ohms. So I am 100% sure that the earth fault current is way below the rated value and my transformer will be safe, even if the fault current goes undetected for any unforeseen reason say my earth fault protection has failed to pick up.

Make sure however that the earth fault setting that you choose is sensitive enough to pick up for the earth fault current calculated. I would generally put two relays a 64 or REF designed to pick up and operate instantly backed by a 51N with a sensitive setting but with a delay of a couple of seconds to pick up in case the 64 has failed to pick up.

So that’s it. I have described how I would go about calculating the earth fault current, selection of NGR value and how I would protect it.
Protection and related devices aiding protection don’t come cheap. Also I assume by your comment “this method is the most expensive option available since the cost of the transformer shall be astronomical”, you are referring to the Zig-Zag transformer and not the actual 132/33kV Star-Delta power transformer, under question.

I have taken a very generic example and tried to focus on how to arrive at a suitable value of an NGR, assuming an Star HV and Delta LV. My aim being to calculate how I could limit the fault current on the Delta LV. Being a Delta winding, I have to use a Zig-Zag transformer, for providing a low zero sequence path for the flow of earth fault current. It is really the Zig-Zag trafo. that bumps up the cost.

Note: If the above transformer is one of a kind, i.e. this is the only transformer in an isolated network, then I simply disregard the Zig-Zag transformer + NGR method and use the 3 PT broken delta method for 3Vo detection to drive a 59N. My cost here would be very low.

If the transformer is a Star-Star type with HV start solidly grounded, and LV star impedance (NGR) grounded, then I don’t need a Zig-Zag trafo. on the LV side. My cost is purely for the NGR alone.(Of course this transformer will have a Delta tertiary which may need it’s own protection depending on the whether one plans to load the tertiary or not. We could di

AQ: System configuration of grounding

There are different types of system configuration for grounding like TT,IT,TN-C etc. How do we decide which configuration is suitable for the particular inverter (string or central). What are the factors that help us to decide the configurations?

One of the main concerns in a system is to avoid large low impedance ground loops.
These are created by the return signal path connected to the chassis (metal work) at multiple points. The large current loop allows noise currents to radiate H fields and hence couple into other electronics. The antenna effect will be proportional to loop area.

Single point grounding of the return path to chassis prevents this. However single point grounding conflicts with good RF practice where you want to ground to chassis at the sending and receiving ends of a signal path. There is therefore no universal best practice.

In my field, spacecraft, the standard practice has all primary power electronics galvanically isolated from the spacecraft chassis. Individual modules must maintain the isolation with transformer coupled DC/DC converters. The centre tap of each PSU secondary output is then single point grounded to the module metalwork. We talk about primary side and secondary side electronics where only secondary side is grounded to the metalwork.

Anything powered directly from the primary bus must be isolated with a maximum capacitance to chassis of 50nF to avoid excess HF loop currents forming.

In general it depends on country specific law and standards required by Power Supply Operators. From design point of view it all depends on which point of grid you are going to connect and what type of inverter is used.

AQ: Snubber circuit for IGBT Inverter in high frequency applications

Q:
First i had carried out experiments with a single IGBT (IRGPS40B120UP) (TO-247 package) 40A rating without snubber and connected a resistive load, load current was 25A , 400V DC and kept it on continuously for 20min . Then i switched the same IGBT with 10KHz without snubber and the IGBT failed within 1min. Then i connected an RC snubber across the IGBT (same model )and switched at 10KHZ. The load current was gradually increased and kept at 10A. This time the IGBT didn’t fail . So snubber circuits are essential when we go for higher switching frequency.

What are the general guide lines for snubber circuit design in You are not looking close enough at the whole system. My first observation is that you are using the slowest speed silicon available from IR. Even though 10KHz is not fast, have you calculated/measured your switching losses. The second and bigger observation I have, is that you think your circuit is resistive. If your circuit was only resistive, any snubber would have no effect. The whole purpose of a snubber is to deal with the energy stored in the parasitic inductive elements of your circuit. Without understanding how much inductcance your circuit has, you can’t begin designing a cost and size effective snubber.

To echo one of the thoughts of Felipe, you need to know the exact purpose of the snubber. Is it to slow down the dV/dt on turn off or is it to limit the peak voltage? Depending on which you are trying to minimize and your final switching frequency, will dictate which snubber topology will work best for you. The reason that so many snubber configurations exist, is that different applications will require different solutions. I have used snubbers in various configurations up to 100KHz.

AQ: Neutral current is less than phase current?

In a balanced 3-phase system with pure sine waves, the neutral current is zero, ideally.
If there is phase imbalance, it shows up in the neutral, so check for imbalance.

The other major cause of high neutral currents is full wave rectification, where the current of each phase is flowing only at its peak voltage. In this case, the neutral current can be as high as three times the phase currents, theoretically.

If you can see the frequency of the neutral current, line frequency currents indicate imbalance. Current due to full wave rectification is high in third harmonics, so it may show mostly 3 x line frequency, or be a ratty square wave at 3 x line frequency.

High neutral currents, and some resulting fires, are largely responsible for the adoption of power factor correction requirements. If your loads are balanced and pfc corrected, you should not have neutral currents.

The neutral current (In) is summation of the phase currents. And obviously, the three phases are decoupled now; and not loading Y makes Iy=0.
So In = Ir + Ib (vectorial sum). Now depending on the amount of loading, nature of loads and their respective power factors, a variety of possibilities (for neutral current magnitude and phase) arise; which may include the case of In being higher.
The statement “neutral current is usually less than phase currents” is naive and not universal.

Nonlinear loads (i.e. rectifiers as Ed mentioned above) draw significant harmonic current. In many cases the current Total Harmonic Distortion (THD) is >100%. In a 3-phase, 4-wire system, the triplen harmonic currents (3, 9, 15, 21…) sum in the neutral wire because they are all in-phase. This is why the neutral current can be much higher than the phase currents even on an otherwise balanced load application. If you can put a current probe on the neutral and look at the waveform – you can see how much fundamental vs. harmonic current there is.

AQ: Harmonic current

I hate to call them harmonic currents. The do submit to Fourier analysis, but you are probably dealing with AC to DC power supplies. If you look at the current pulses, you will see that each pulse is about 1-2 milliseconds in duration in alternating directions. If you sum these all in the neutral there is the appearance of what looks like 180 Hertz in the neutral. If you use different sized power supplies on each phase, you can see that it is just the addition of the three phases. So the neutral current when you have non power factor corrected power supplies is the sum of the three phases. Unless the current waveforms overlap, there is no cancellation of current in the neutral, hence the neutral current is the sum of the phase currents. The reasoning behind this is the rectifier diodes in the front of the power supply and the DC storage capacitors size relative to the DC load on the capacitor. The general rule of thumb is that the capacitor is about 800 to 1000 microfarads per amp of current in the capacitor.

Realize that the extra heating in the three phase delta-wye transformers is due to the extra circulating current in the primary delta causing excessive heating of the primary conductor. The world calls transformers designed to deal with this “K” factor transformers. Let the world of electrical engineers bury all this simple stuff behind the maze of Fourier analysis. Change the incoming voltage slightly and your Fourier analysis is garbage. The issue here is switches and storage caps— not some magical mathematical garbage.

By the way if someone wanted to use the wire sizing guidelines of the National Electrical Code in the US to size wire for 100% power supply load, the neutral wire would be 8 gauge sizes larger than the phase conductors. People need to start demanding PFC power supplies. Fix a switching problem with switches.

AQ: Designing Gate drivers for IGBT

Q:
When designing gate drivers for IGBT’s, how reliable are the gate driver IC’s ? Now there are a lot of gate driver IC’s available in the market. For example i am using the Hybrid IC M57962L for driving IGBT’s for 3 phase inverter application. The peak output current of this Hybrid IC is 5A and it’s written in data sheet that it can be used for driving IGBT’s up to 200A, 1200V and many features in it.

For an initial design and for lower power rating the configuration is working fine. But, before going for higher power rating, i want to make sure about the reliability of Gate driver IC’s in general.
Is it advisable to design gate drivers using commercially available IC’s or go for a design which includes a gate drive transformer . What are the issues that may arise when using driver IC’s.

A:
I’ve seen and developed designs using these hybrid gate drives quite successfully with long term field reliability in applications requiring from 800 V to 1.25 kV voltage isolation in power conversion products for the semiconductor capital equipment market. Powerex offers various different isolated drivers like the M57962L – my personal favorite is the VLA-502 which also contains the isolated DC/DC converter used to power the isolated gate drive electronics.

There are only two problems that I remember in the last 10 years with these types of commercial drivers – and both problems, if I remember correctly were with the stand alone DC/DC converter intended to be used with the stand alone isolated driver. One problem was a voltage isolation issue from primary to secondary inside the DC/DC switcher. Powerex acknowledged the problem, and upgraded the design. I simply do not recall the part numbers involved. The second problem was with regard to how the isolated VEE rail was established – the monopolar output of the DC/DC converter was offset negative, and ground referenced with a zener diode – and when the IGBT gate would become active at high frequency (25 kHz for that particular application), the gate charge was high enough to sag the negative supply rail against the zener shunt.

Bottom line: Use a good isolated DC/DC converter, with solid VCC and VEE regulated outputs. The isolated drivers themselves are solid in my experience – a nice, simple solution with typically better rise and fall times than gate drive transformers. They also have the added benefit of being capable of holding positive or negative DC bias if the application requires it.

AQ: Are variable speed drives harmful to motor?

Variable speed drive switches very fast which brings high dv/dt on motor. How often do we face with problems coming with VSD? How harmful is the common mode currents in windings and other parts of motor due to high dv/dt. Do we see winding isolation failure? How much does the life of motor reduce? Also, is the filtering of voltage at the output of inverter common or applicable practice in the field?

The waveforms for the INVERTER are not good to the motor…. Makes the motor run hot and less efficient….. and all the above….
In-line filters to reduce harmonics is a must in many cases…
Depending on power levels you can have in line reactors for CM and DM or balanced bridge methods for CM… There is methods of harmonic canceling with reactors called harmonic blockers, where you arrange the 3 phase windings in such a manner to cancel certain harmonics….not all harmonics will be blocked, usually in grouping intervals…you need to be aware of what harmonics are your worst offenders…

Mostly in medium and high voltage motor drives the very fast change of the voltage can induce high capacitive currents inside the motor with harmful results.
A way to reduce this negative effect is to increase the number of voltage steps (levels) such that the dV/dT will decrease proportionally (dT=turn on switching time, dV=one voltage step). The most popular method used is SVPWM (space vector sine PWM) NPC (neutral point clamped) multilevel frequency converter. Line L-C filters are also used for EMC.

The first step in any filter analysis is knowing what harmonic vectors your dealing with.
Mathcad is a great tool for modeling the PWM modulation with the sub carrier and generating the harmonic matrices..vectors…I usually go above the 100th harmonic in some analysis, then doing this over the operating ranges of the motor….you then pick your Worse Case operating point and now you have a matrices to work with…. Summing the harmonic magnitudes will give you an idea of how much garbage your feeding your motor windings.

They could be harmful for high frequency current and voltages which are not economical to be eliminated.
But this weakness is so neglect able to the benefits providing. These benefits are very comprehensive. The harmful harmonics are controlled by the standards, so in order to improve harmonic characteristics, we need an improved standard.

AQ: Add filters to frequency inverter to eliminate harmful

The high frequency edges of switched waveforms can cause capacitively coupled currents to flow from windings to frame, returning through the bearings, and these can accelerate corrosion in the bearings, causing early failure. Small filters on the motor leads allowing these currents to return locally to ground will avoid this.

The best way, though, is to use filters which can eliminate sharp transitions and leave only (like +/-10% ripple) fundamental frequency (motor’s RPM at given point) of the motor drives. However if somebody can handle 40 – 50kHz of the switching frequency the filter’s size shrinks dramatically and it is not too expensive anymore. Again, the problem is in ability to handle 100 (or so) kWs and 50kHz together.

AQ: Motor connection

Many years ago I had an experience of 4nos 37kW fin-fan motors wrongly connected at site to a star. After running for almost 1 year, the operators reported these motors were very warm and felt unusual. We removed one of them to the workshop and opened for inspection. All windings were OK but the rotor lamination surface had turned to light blue colour which showed a sign of abnormal heating.

I asked different experts in the industries for advices. From the advices, we suspected the motor could be designed for a delta connection even though the nameplate indicated a Star connection for 415V. We contacted the motor manufacturer by quoting the motor serial no. The manufacturer confirmed that the motors were designed for delta connection at 415V. The manufacturer apologized for the error in nameplate and gave us a free spare motor.

One clear sign that could lead us to believe that the motor was in a wrong star connection instead of delta was, for a 2 or 4-pole motor the no load running current should be more or less around 30% of FLC. When we tested run the motor in the workshop, the no load current was less than 15%xFLC.

After the rectification of all the 4 motors to delta connection, we had no complaint anymore. It was a good lesson out of this solved problem.

AQ: Transformer Saturation