High 3rd harmonic on the neutral

Sounds like nothing to worry about, although the information isn't very thorough about the circuit rating, fusing, neutral size, and proportion of time when the 300% value is found for the neutral current.  A bit unusual to get very well cancelled fundamentals, but student places are good at having lots of small IT equipment and very little else. If it were 30 normal domestic flats instead of 300 student flats, then kettles, showers etc would help bring this 300% value down at some times by giving some unbalanced fundamental current in the neutral.


"... to illustrate a SMPS which gives a harmonic content of 50%"


Waveforms needn't be square to have high harmonic content. When a waveform tends towards being made of very short pulses (even if their edges change smoothly), the closer the spectrum becomes to having harmonics that are as big as the fundamental. 


Some examples of IT-related loads are on this page it_power, which shows the waveform, harmonics and power for different operation states. See how big the third harmonics are compared to the fundamentals ('1st harmonic'), particularly in the monitor. But the bigger server-like computers have much lower third harmonics. Compare the waveforms with the laptop-charger waveforms below, to help estimate the harmonics for the chargers.


Mike has already given a good summary of rules that modern devices should follow. It's the large numbers of small, cheap electronic items that contribute the highest harmonic content. When one gets into bigger electronic loads such as heat-pumps, the waveforms are much closer to sinusoidal. Plenty of modern devices that students would be likely to have can have very peaky waveforms like the monitor in the above link, as they'd typically be only some tens of watts. A good general rule for cheap, small electronic goods is that unless you force them to be manufactured to have more than a diode bridge and capacitor at the input, they won't be. 

I did some measurements on lots of types of modern power electronic loads back in the spring, mainly interested in their reactive power production and in how broad a band of input voltage they draw near-constant power over (answer: a remarkable range, even for heat-pumps and induction hobs). Below are examples of waveforms from a couple of laptop chargers. See how the 65W one is peaky anyway, and the 90W is peaky too when on low load when the laptop is just sitting idle and fully charged, but it's better behaved when running nearer its rating. USB chargers tend also to be very peaky. So these sorts of loads can easily have third harmonics over 50% of the fundamental (or even above 50% of the total rms current) - compare to the waveform+harmonic plots in the above link.






"Maybe we are drifting towards domestic being as troublesome as datacentres."

I was interested by a comment a few months ago about datacentres with very oversized neutrals; it made me wonder how much of the oversizing is precaution, and how much is still a necessity. Modern server power supplies where I've seen measurements have far lower harmonic content than small domestic stuff, although this perhaps wasn't the case in days when design rules were established. 

I'm confident that the domestic power electronic loads are worse for harmonic content than servers. But the saving grace is that domestic dimensioning is based on other much bigger loads that don't have much harmonic content, so it's very unlikely that harmonics will be big enough in absolute terms to be a problem for overloading.  

"without harmonics, the neutral current cannot be larger than the largest single-phase current"

To be strictly true, you should require similar types of load on all the phases: i.e. the same power-factor and power direction.

For example, if the three phases had a lagging, leading and resistive load, the three phasors wouldn't be spread at equal angles, and their phasor sum could be more than any single one; but this is very unlikely to be a practical problem.  More practically important is a single-phase PV inverter on one phase, and resistive loads on the others. In this case, which equal current magnitudes in the phases and with no harmonics, the neutral would have twice the phase current magnitude. Utilities don't like customers to have PV installations anywhere near the maximum permitted load unless they're 3-phase ... though that's often for voltage-control reasons. I gather that 5 kW is a common single-phase limit in the UK, for installations with 80 A or 100 A main fuse.


With harmonics, note that the summing of third harmonics is on the assumption that the loads in all phases are similar. A load with a waveform that differs from a sinusoidal by having an increased peak, and another load with a waveform that instead has a flattened peak, can well have third-harmonics in opposite phase with each other, so they cancel in the neutral regardless of whether the loads are on the same or different phases. So third-harmonics don't inherently add in a neutral. 


Yes, power electronic inputs are taking over with more and more types of load. They change several features of load behaviour. For example, utilities have been noticing declining reactive power consumption, steadily, over the past 1 or 2 decades, part of which I consider to be due to the typically capacitive behaviour of power electronic inputs. My home typically generates between 300 and 900 var.  It's the induction hob and heat-pumps that push it up to the higher values, and various computers, input filters, lights and chargers that provide the base part. 

Sounds like nothing to worry about, although the information isn't very thorough about the circuit rating, fusing, neutral size, and proportion of time when the 300% value is found for the neutral current.  A bit unusual to get very well cancelled fundamentals, but student places are good at having lots of small IT equipment and very little else. If it were 30 normal domestic flats instead of 300 student flats, then kettles, showers etc would help bring this 300% value down at some times by giving some unbalanced fundamental current in the neutral.


"... to illustrate a SMPS which gives a harmonic content of 50%"


Waveforms needn't be square to have high harmonic content. When a waveform tends towards being made of very short pulses (even if their edges change smoothly), the closer the spectrum becomes to having harmonics that are as big as the fundamental. 


Some examples of IT-related loads are on this page it_power, which shows the waveform, harmonics and power for different operation states. See how big the third harmonics are compared to the fundamentals ('1st harmonic'), particularly in the monitor. But the bigger server-like computers have much lower third harmonics. Compare the waveforms with the laptop-charger waveforms below, to help estimate the harmonics for the chargers.


Mike has already given a good summary of rules that modern devices should follow. It's the large numbers of small, cheap electronic items that contribute the highest harmonic content. When one gets into bigger electronic loads such as heat-pumps, the waveforms are much closer to sinusoidal. Plenty of modern devices that students would be likely to have can have very peaky waveforms like the monitor in the above link, as they'd typically be only some tens of watts. A good general rule for cheap, small electronic goods is that unless you force them to be manufactured to have more than a diode bridge and capacitor at the input, they won't be. 

I did some measurements on lots of types of modern power electronic loads back in the spring, mainly interested in their reactive power production and in how broad a band of input voltage they draw near-constant power over (answer: a remarkable range, even for heat-pumps and induction hobs). Below are examples of waveforms from a couple of laptop chargers. See how the 65W one is peaky anyway, and the 90W is peaky too when on low load when the laptop is just sitting idle and fully charged, but it's better behaved when running nearer its rating. USB chargers tend also to be very peaky. So these sorts of loads can easily have third harmonics over 50% of the fundamental (or even above 50% of the total rms current) - compare to the waveform+harmonic plots in the above link.






"Maybe we are drifting towards domestic being as troublesome as datacentres."

I was interested by a comment a few months ago about datacentres with very oversized neutrals; it made me wonder how much of the oversizing is precaution, and how much is still a necessity. Modern server power supplies where I've seen measurements have far lower harmonic content than small domestic stuff, although this perhaps wasn't the case in days when design rules were established. 

I'm confident that the domestic power electronic loads are worse for harmonic content than servers. But the saving grace is that domestic dimensioning is based on other much bigger loads that don't have much harmonic content, so it's very unlikely that harmonics will be big enough in absolute terms to be a problem for overloading.  

"without harmonics, the neutral current cannot be larger than the largest single-phase current"

To be strictly true, you should require similar types of load on all the phases: i.e. the same power-factor and power direction.

For example, if the three phases had a lagging, leading and resistive load, the three phasors wouldn't be spread at equal angles, and their phasor sum could be more than any single one; but this is very unlikely to be a practical problem.  More practically important is a single-phase PV inverter on one phase, and resistive loads on the others. In this case, which equal current magnitudes in the phases and with no harmonics, the neutral would have twice the phase current magnitude. Utilities don't like customers to have PV installations anywhere near the maximum permitted load unless they're 3-phase ... though that's often for voltage-control reasons. I gather that 5 kW is a common single-phase limit in the UK, for installations with 80 A or 100 A main fuse.


With harmonics, note that the summing of third harmonics is on the assumption that the loads in all phases are similar. A load with a waveform that differs from a sinusoidal by having an increased peak, and another load with a waveform that instead has a flattened peak, can well have third-harmonics in opposite phase with each other, so they cancel in the neutral regardless of whether the loads are on the same or different phases. So third-harmonics don't inherently add in a neutral. 


Yes, power electronic inputs are taking over with more and more types of load. They change several features of load behaviour. For example, utilities have been noticing declining reactive power consumption, steadily, over the past 1 or 2 decades, part of which I consider to be due to the typically capacitive behaviour of power electronic inputs. My home typically generates between 300 and 900 var.  It's the induction hob and heat-pumps that push it up to the higher values, and various computers, input filters, lights and chargers that provide the base part.