Solving Harmonics ?

Status
Not open for further replies.

brian john

Senior Member
Location
Leesburg, VA
I have looked at literally 100's of projects for harmonic issues and .

Now it may be the area I do most of my work in but the average service and downstream transformers are typically 30%-40% loaded so neutral issues is not a problem.

The few cases where an overload neutral was an issue was on the secondary of a 208/120 Wye transformer that was at 90%-100% load.

We did have issues in the early days of UPS where there were mechanical governors, and no input filters on the UPS.

Additionally when VFD's were replacing motor starters years ago on large projects that had single transformers dedicated to mechanical loads, there were issues with the utility transformers and early generation electronic relaying.


But most harmonic issues I see are not harmonics but typically hardware issues, grounding issues, such as neutrals grounded downstream of the transformer compounded by other wiring issues.
 

Jraef

Moderator, OTD
Staff member
Location
San Francisco Bay Area, CA, USA
Occupation
Electrical Engineer
I teach a class on harmonics as it relates to VFDs. Here's the bottom line I bring up early on.

Harmonics are always a concern now with modern equipment because virtually everything electronic now uses Switch Mode Power Supplies of one sort or another, but harmonics are NOT always a problem.

Quick and dirty rule-of-thumb; if your non-linear load is more that 30% of the total load, you MIGHT have an issue that warrants further investigation. With the advent of lighting no longer being linear (because incandescent lighting is going away), the only linear loads left are AC motors running across-the-line, or resistance heating. Virtually everything else now is non-linear.

But in a single phase system, even order harmonics cancel each other out, and in a 3 phase system, so do 3rd order harmonics. So how much "BAD" harmonics remains can often end up insignificant. Getting down to exactly what the problems are and the solutions for it is not something you deal with by slapping some little line reactor on at the service entrance, in fact that really does little to address it completely, it just lessens it a bit. The real solutions involve discovery, investigation, planning for future expansion (if any) and either point source mitigation or active neutralization at the Point of Common Coupling.

So bottom line Holly, going to work selling this kind of snake oil will not help your career because you will only learn how to bilk guillable people out of their money, not provide a valued service. That skill set makes you a scam artist, not a competent salesperson.
 
Last edited:
OK, I'll be the dumb one again.

I don't understand how 3d order harmonics cancel each other out. In a 3-phase system, doesn't a 180 Hertz component reach a peak simultaneously on each conductor? Isn't that what overloads neutrals, when that happens?

Did you mean they cancel out in terms of not supplying any power to the loads?

(Kudos to Holly for researching a product before trying to sell it).
 

GoldDigger

Moderator
Staff member
Location
Placerville, CA, USA
Occupation
Retired PV System Designer
In a three phase delta system it is possible for third harmonic load currents to be handled as circulating currents within the delta windings and so have no effect on the current or voltage at the primary (upstream) side.
Those harmonics will not affect other users.
In a three phase wye the same harmonic currents will add up in the neutral conductor on the secondary side and may cause severe problems. And with a wye-wye transformer those same harmonic currents can be seen upstream too.
 

Besoeker

Senior Member
Location
UK
I teach a class on harmonics as it relates to VFDs. Here's the bottom line I bring up early on.

Harmonics are always a concern now with modern equipment because virtually everything electronic now uses Switch Mode Power Supplies of one sort or another, but harmonics are NOT always a problem.
One of the best papers I have read on harmonics was by a Professor Reed and it was written in the 1930s. It mainly referenced mercury arc rectifiers but the figures are relevant to solid state equipment too. So it isn't just a modern concern.

My background, as I'm sure you know, is power electronics and mostly related to variable speed drives, both AC and DC and primarily in process industries. Anything from 6-pulse to 24-pulse and up to quite a few MW. Those need serious consideration at the design stage or even at the bid stage. Get it wrong and fixes are neither easy nor cheap.

The company I retired from last fall* has a lighting division the main product being dimmers. Single phase of course. I got dragged in to look at overheating neutrals in a very prestigious hotel in the middle east. There was a total of 3MVA of these dimmers distributed line to neutral across the three phases. Single phase phase controlled anything produces triple-n harmonics. These are additive in the neutral of a three phase system. Hence the hot neutral.

*picking up on my dear wife's Americanisms.........:)
 

Jraef

Moderator, OTD
Staff member
Location
San Francisco Bay Area, CA, USA
Occupation
Electrical Engineer
One of the best papers I have read on harmonics was by a Professor Reed and it was written in the 1930s. It mainly referenced mercury arc rectifiers but the figures are relevant to solid state equipment too. So it isn't just a modern concern.

My background, as I'm sure you know, is power electronics and mostly related to variable speed drives, both AC and DC and primarily in process industries. Anything from 6-pulse to 24-pulse and up to quite a few MW. Those need serious consideration at the design stage or even at the bid stage. Get it wrong and fixes are neither easy nor cheap.

The company I retired from last fall* has a lighting division the main product being dimmers. Single phase of course. I got dragged in to look at overheating neutrals in a very prestigious hotel in the middle east. There was a total of 3MVA of these dimmers distributed line to neutral across the three phases. Single phase phase controlled anything produces triple-n harmonics. These are additive in the neutral of a three phase system. Hence the hot neutral.

*picking up on my dear wife's Americanisms.........:)
I spent a couple of years working for Boeing Computer Sciences in Seattle in the facilities dept., right at the time when they had a huge push to change everything over to IBM/Dassault CATIA Workstations for engineering, each one was essentially a minicomputer (more powerful than a PC). All of the buildings were old office complexes, so built under old codes with old equipment, meaning neutral conductors were one size smaller than line conductors, and nobody had even heard of a "K-Factor" transformer design. We had transformers melting down, sometimes with some pretty fireworks as a bonus. We would sometimes go into older buildings and be able to trace the routing of wiring inside the walls because the neutrals would get so hot that they would discolor the paint from inside. We were tasked with finding the cause of this and at a meeting in around 1992, I remember us all reading IEEE519 and slapping our collective foreheads in a "Eureka" moment. Up to that point, I was unaware that harmonics was anything outside of what a piano tuner would think of.
 

junkhound

Senior Member
Location
Renton, WA
Occupation
EE, power electronics specialty
Jreaf:

Hah! Wonder if YOU were the facilities eng I talked to back then ?? I forget the name of whoever I spoke to.


Building I was in at that time was Kent East, one of the twin 8 story buildings. Had heard tell the old Plant 2 mezzanine had some pretty bad problems also. (Plant 2 torn down now, all but the old HQ building and WT complex gone.

Did you ever work on the substation just outside the WT or the 40 MW motors there ? Our groups big 0.4F, 6 kV capacitor bank (for EMP and lightning simulation) was powered right off that substation, up until co. got rid of all those PCB caps.
 
Last edited:

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
Backyard experiment

Backyard experiment

160205-2156 EST

In the past I have measured ground voltage in my backyard. This was done with two screwdrivers pushed into the ground, and a high impedance voltmeter between the two probes. I believe in the past the probes were about 24 ft apart. Typical voltage was well below 1 volt with a Fluke 27 with no filtering. Our primary lines are essentially a floating delta, but possibly grounded at the substation. Thus, the only grounding for my supply is the transformer center tap at the pole, and water lines. I believe that ground currents and thus voltages are more distant related than my neighborhood.

I have been curious as to whether I could see any 50 Hz on this side of the world. My experiment did not show any 50 Hz.

The weather has been somewhat warm the past few days, all the snow has melted, and temperature was high 30s. The top soil was soft down to about 6 to 8 inches.

I placed two screwdrivers 24 ft apart on an east-west line. Put scope on foam, and I was on cardboard with a black sweater as my sun shield. AC power to the scope. Input filter to scope was low pass, 10 k and 1 ufd.

Attached is an FFT plot of one sample. The horizontal scale is about 12.5 Hz per major division, and vertical 10 db per major divison. Note the major spike at 60 Hz. There appear to be spikes at 40, 60, 80, 120 and 160 Hz. In other plots I also saw a spike at 20 Hz. Is the 20 Hz and its multiplies an instrumentation anomaly?

.
DS2_QP_63-B-Gray.jpg

.
Are any of you familiar with experiments like this?

In time plots I saw fast phase shift changes, and some substantal short period, 100 cycles or so, voltage (meaning current) variations.

In future experiments I will simultaneously measure east-west and north-south signals and if I can get Pico to work have greater voltage resolution. If winnie is reading this, then I can report that Pico works sometimes, but has an intermittent.

.
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
160205-2458 EST

Note that 10k and 1 ufd results in 3 db at 15 Hz. I did not bother to check this before the tests. 0.1 ufd and 3000 k might be a better choice with 3 db at 1 kHz.

.
 

GoldDigger

Moderator
Staff member
Location
Placerville, CA, USA
Occupation
Retired PV System Designer
Some electric railways use 20 Hz traction power at currents that can cause gradients far away. Not super likely to be what you are seeing, but there are other 20 Hz systems out there.
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
160206-1150 EST

GoldDigger:

Thanks. That is an interesting possibility.

Today I got the Pico ADC-216 to work for a while. This is 16 bit vs 8 bit resolution. Claims to have near 100 db range.

Too cold and hard to play outside so I decided to look at just my distorted AC line voltage.

For the FFT I used Gaussian shaping, and 2048 spectrum samples, with sampling set for full scale of 81 and 326 Hz.

In the 81 Hz plot there are side bands on either side of 60 at +/-10, and about +/-8.3, and something less pronounced here at +/-17 hz. Note: the +/-10 appear to be right on 50 and 70.

The results I see on the 326 plot referencing 60 Hz to 0 db are 60 Hz at 0 db, 180 Hz at -43 db, 300 Hz at -38 db.
Then there appears to be +/- 18 Hz side bands about the 180 and 300 peaks.

Now the question might be is there 18 Hz or thereabouts modulation of the primary system? These same +/- sidebands may be present on the 60 Hz fundamental, but buried in the skirts. See above comments on the 81 Hz plot. I don't have a precise value on the 18 Hz. But the spike near 200 Hz was just a little below 200. I have not found that Pico has an adjustable cursor to spot these frequencies.

There was a time or two where some unexpected frequencies popped up between 180 and 300 Hz. Each spectrum analysis takes a number of seconds. At 81 Hz a large number of seconds, possibly a minute. The .PSD file created by Pico doesn't import to other programs, so I switched to a .JEPG.

Note something at 25 Hz, but it is not exactly 25 Hz, just slightly higher.


mivey:

Are there some system factors that can account for these low frequency modulation components?

.
2016_02_06_005.jpg

2016_02_06_006.jpg
.

.
 

GoldDigger

Moderator
Staff member
Location
Placerville, CA, USA
Occupation
Retired PV System Designer
160206-1150 EST

There was a time or two where some unexpected frequencies popped up between 180 and 300 Hz. Each spectrum analysis takes a number of seconds. At 81 Hz a large number of seconds, possibly a minute. The .PSD file created by Pico doesn't import to other programs, so I switched to a .JEPG.

Note something at 25 Hz, but it is not exactly 25 Hz, just slightly higher.

My guess about the unexpected frequencies popping up is that there was some kind of impulse transient present during the analysis period which contained a broad spectrum of frequencies and it averaged down to a low overall level. Rather than a true long lived contribution at that frequency.
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
160206-1715 EST

GoldDigger:

Yes some sort of transient load or change in the sysytem. I would like to have caught the occurance I saw.

I have been thinking about my observed sidebands. Looks like amplitude modulation. The 10 Hz I might link back to a once per revolution effect from alternators run at 600 RPM or twice per revolution at 300 RPM. What is the RPM of a large power station alternator? Because of the large size the RPM won't be high.

The 8.3 Hz component probably does not relate to mechanical rotation. However, in a set of roller bearings I have found a frequency component that related to the precession rate of the roller cage to the inner race that was not once per revolution of the inner race. This same frequency could be seen in the vertical stack height of the bearing assembly.

The observed 8.3 and 10 Hz signals were basically constant signals along with the fundamental and its harmonics.

.
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
160206-2355 EST

FionaZuppa:

Have you run any of your own tests?

The YouTube presentation is very poor.

The time domain waveforms were not clear and may not have been a good representation of the actual current of that particular bulb. Only one LED was shown. Some LEDs have high harmonic content and some don't.

Further the LED was shown connected to a phase shift dimmer. If you connect an incandescent to to a phase shift dimmer, then you also produce a current with high harmonic content.

Tonight I looked at three different LED bulbs with full sine wave applied:

One unknown, fairly good sine wave current.

One a new Cree, good sine wave current. This is dimmable with either phase shift or Variac.

A bargain basement low cost Feit, very bad current waveform. Virtually rectangular wave pulse at the voltage peak. On time 3 mS, and off time 5.33 mS. This is not a dimmable bulb. My guess is that this will produce substantial radio noise. To be determined.

Current to Feit at 120 V 60 Hz with sine wave input voltage measured with a Hall current sensor.

2016_02_07_001.jpg

.
FFT from above current waveform. Note: the slow decline of the harmonics as expected, and only starting at about 10 db down from the fundamental.
2016_02_07_002.jpg
.
.
 

FionaZuppa

Senior Member
Location
AZ
Occupation
Part Time Electrician (semi retired, old) - EE retired.
my ~8yr old cheapy scope does not have the FFT function, but i still have use for my scope so i have not yet upgraded. i think the video was just showing what it is vs comparing different bulbs, etc.
 

Besoeker

Senior Member
Location
UK
my ~8yr old cheapy scope does not have the FFT function, but i still have use for my scope so i have not yet upgraded. i think the video was just showing what it is vs comparing different bulbs, etc.
Can you download the data to a PC?
 

gar

Senior Member
Location
Ann Arbor, Michigan
Occupation
EE
160207-1230 EST

My just above current and FFT plots were done with a Pico ADC-216 which has band limiting probably around 100 kHz.

Today I was going to use a resistive shunt, 10 ohms, to get a cleaner current signal, and not have the Hall sensor added noise. To quickly look at the signal I used the Rigol scope and was totally surprised by the signal.

With the Feit I could hardly see the basic rectangular pulse. Rather I saw a large high frequency signal superimposed in the middle of the current pulse. This was many times the magnitude of the basic current pulse. Pictures later. The measured frequency in one portion was about 760 kHz. This is directly on top of our 50,000 W station WJR. Got my all band receiver out and the Feit is producing a wide band of interference from about 700 to 1000 kHz. This did not wipe out WJR, but did interfer, and wiped out the weaker stations in this frequency range. The Feit signal obviously has a 120 Hz signal modulating it as would be expected.

I would think the FCC, at least, would put a stop to these bulbs.

.
 
Status
Not open for further replies.
Top