ProSoundWeb Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 [2] 3   Go Down

Author Topic: Non Power Factor Corrected Pro Audio Gear  (Read 15638 times)

Keith Broughton

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 3667
  • Toronto
Re: Non Power Factor Corrected Pro Audio Gear
« Reply #10 on: October 04, 2014, 09:34:57 PM »

Logged
I don't care enough to be apathetic

Guy Holt

  • Full Member
  • ***
  • Offline Offline
  • Posts: 130
Re: Non Power Factor Corrected Pro Audio Gear
« Reply #11 on: October 05, 2014, 03:49:38 PM »

I'm trying to understand exactly what this means. Did the manufacturer suggest to use half the generators published capability as the "real world" capability for those kinds of loads? In other words, if the genny was 100 amp, you should consider that it is only 50 amps when used with those loads?

One determining factor for how much you need to de-rate a generator is its’ subtransient reactance value.  For instance, I ran the PF of 4’-8 Tube Fluorescent Fixtures by the generator alternator manufacturer Marathon Electric because we were trying to buy a 175kW studio generator capable of powering  Kino Flo Image 80s by the hundreds for effects productions.  Here is the reasoning behind Marathon’s recommendation to de-rate a standard studio unit by 50%.

In the case of a load with a PF of .63, they recommended limiting  the sub-transient reactance to 6% or less in order to keep the voltage waveform distortion to within levels that non-linear loads could tolerate.  The standard alternator used in studio blimped generators is the Model 431PSL6208 with an X"d = 0.158 pu.  To cut the  X"d  in half, one option was to use an alternator model with a lower subtransient reactance. 
 The problem we encountered was that the one 175kW alternator that met that criteria, Model 433PSL6220 with a X"d = 0.062 pu, does not fit in the standard studio blimped housing, so oversizing the alternator was not an option.  So de-rating the standard alternator by 50% was the only alternative.

If the load requirements were smaller, a third alternative would be to use an inverter type generator.  If say we only required 100 Amps, then two paralleled Honda EU6500s, or two of the new EU7000s, would be better than a 100A conventional generator when it comes to powering non-linear loads like the Kino Flo Image 85s that use Switch Mode Power Supplies (SMPSs.) The reason lies in Ohms Law.

According to Ohm’s Law, harmonic currents react with impedance to cause voltage drop. Therefore the magnitude of the voltage waveform distortion caused by SMPSs is a function of the source impedance. Which means that the  generator with the lowest internal reactance to an instantaneous current change at a given load will typically have the lowest value of total harmonic distortion under nonlinear load conditions. This is one of the great benefits to using inverter generators over conventional generators: inverter generators have much lower internal reactance and so they are less prone to voltage waveform distortion caused by the harmonic currents drawn by SMPSs.



Left: A non-linear Load powered by Grid Power. Center: A non-linear Load powered by Conventional AVR Power. Right: A non-linear Load powered by Inverter Power.

As can be seen in the oscilloscope shots above, inverter generators are less prone to voltage waveform distortion because the inverters completely processes the raw power generated by its alternator (converting it to DC before converting it back to AC) –making the AC power it generates completely independent of the engine. In fact, its’ microprocessor controller can vary the engine speed without affecting the voltage or frequency of the power the inverter module puts out. Now that the internal reactance of the engine is separated from the power output, harmonic currents encounter very little impedance; and, as is evident in the oscilloscope shot above right, there is considerably less voltage distortion at the load bus of inverter generators than there are conventional generators. The net benefit is that non-linear loads, like the SMPSs of most electronic equipment, do not adversely affect the power of inverter generators as they do the power of conventional generators. Which means that electronic equipment powered by SMPSs will operate more reliably on inverter generators - making paralleled Honda EU6500s roughly equivalent to a 200A conventional generators when it comes to powering predominantly non-linear loads.

For a more detailed explanation of what causes voltage waveform distortion and how to mitigate it, see my white paper on the use of portable generators in motion picture production that is available at http://www.screenlightandgrip.com/html/emailnewsletter_generators.html.

Guy Holt, Gaffer
ScreenLight & Grip
www.screenlightandgrip.com
Logged

Guy Holt

  • Full Member
  • ***
  • Offline Offline
  • Posts: 130
Re: Non Power Factor Corrected Pro Audio Gear
« Reply #12 on: October 22, 2014, 05:43:38 PM »

I have moved these posts from "Re: Long 12 Gauge Runs" so as not to hijack that thread and this seems to be the more appropriate thread.

I know I had problems with the Crown XTI 1000 at 100v (long run from lousy genny) - ran hot and was clipping prematurely.  Sounded REALLY bad when it did...
But they did not shut down.

frank

Do you remember what generator that was? How long the run was? What gauge of wire it was? And, finally how much the total load on the generator was?

Guy Holt, Gaffer
ScreenLight & Grip
www.screenlightandgrip.com

Hi Guy,
I Vaguely recollect.  Couple years ago.  A Celtic festival, late June in DC along the Potomac - miserably Hot (95+) and very humid. Rain late in the day(s)
Genny was a noisy construction POS.
Our stage used it, it was near the beer tent, and near some vendor tents.  I can only assume they tapped some power.
Figure the genny was about 75 ' from the stage (as the crow flies), and another 75' or so of it back to FOH. A single 10"EON II (250w) was set up outside FOH.
Cabling was 12ga SOOW.  I think we were running off some type of home-made distro.  No clue as to whether we had two circuits or not.
Voltage was showing a pretty steady 101 at the distro (stage side), regardless of what was running.  I had a Furman 1214 that was kicking FOH voltage up to 120 for the desk and outboard gear (minimal)
Stage was acoustic, with two Crown XTI, each channel - oops nope.
One crown XTI 1000 with four 8-ohm speakers per channel (2ohms/channel)
Thermal light popping on and off.
The other Crown got snatched for another stage (on a Whisperwatt, dang it!) 
Now that I think of it, I think we were also running one of those large direct drive pedestal fans.


Missing is the gen size, and total load.  Probably not of much help.
frank

What you do remember confirms my suspicion that your problem was not only a lousy generator but also voltage “flat topping” caused by the harmonic currents drawn by the XTI 1000 encountering the high impedance of the long 12 gauge run and the high impedance of the construction generator. Always bear in mind that voltage drop tables and calculators are for the 60Hz primary only and do not take into account the effect of harmonic frequencies.


Let’s look at the voltage drop component of your problem first. As you can see in the power quality meter reading above of similar sized non-power factor corrected HMI light that uses a similar Switch Mode Power front end, SMPSs draw a distorted current waveform that is rich in harmonics. The higher harmonic frequencies create what is known as "skin effect" in the cable. Skin effect is a phenomenon where the higher frequencies cause the electrons to flow toward the outer sides of a conductor. Since the flow of the electrons is no longer evenly distributed across the cross sectional diameter of the conductor, more electrons are flowing through less copper and the resistance of the conductor increases. The increase in resistance reduces the ability of the conductor to carry current, resulting in greater voltage drop over shorter distances and overheating of the conductor.



The area of the cross sectional diameter of a conductor used by DC current (left), Low Frequency AC Current (center), High Frequency AC Currents (right).

The graphs below illustrate the difference of varying harmonic numbers on both Proximity Effect and Skin Effect for 12 AWG and 4/0 cable (the cable spacing used to obtain the Proximity Effect values is based on National Electric Code (NEC) insulation type THHN.) Comparing the graphs, it is immediately apparent that Skin Effect is more significant in the smaller cable, than the large cable. Within the range of the triplen harmonics alone (i.e. 3rd, 9th, 15th), Skin Effect increases by 60%. 



Which means that the increase in voltage drop due to harmonics is appreciably more significant in the jacketed multi-conductor cables (10/2 or 12/3) commonly used with small portable generators, than with the larger gauge feeder cables (single conductor #2, 2/O, & 4/O) used to distribute power from tow plants. The increase in resistance due to Skin Effect reduces the ability of stingers (12/3 cable) to carry current, resulting in overheating of the conductors and greater voltage drop over shorter distances than with larger feeder cables.


Increased voltage drop from harmonics was not the only cause of your problem. The other contributing factor was voltage “Flat Topping” caused by the high impedance of your genertor. Always remember, there are two components to the impedance of a distro: cable and generator. The skin effect caused by the harmonic currents drawn by the non-pfc SMPS of the Crown XTI increased the impedance of the cable. The second contributing factor was that you were operating on a conventional AVR generator rather than an inverter generator.


Left: Non-pfc SMPSs powered by grid power. Right: Same Non-pfc SMPSs powered by conventional AVR Generator (Honda EX5500) Note different effect that the same non-linear load harmonics have on grid power and power from conventional AVR generator.

Under your circumstances one could expect similar voltage “Flat Topping” to that in the example above. What causes flat-topped voltage? According to Ohm’s Law current reacts with impedance to cause voltage drop.  Since electronic ballasts consume current only at the peak of the voltage waveform, voltage drop due to system impedance occurs only at the peak of the voltage waveform. The zig-zag saw tooth pattern in the right oscilloscope above suggests the flat topping of the voltage waveform we see here is caused not only by the 60hz fundamental but also by the harmonic currents drawn by the non-pfc SMPSs at higher frequencies that also create voltage drops as they pass through the system impedance. For example, when encountering the high impedance of a conventional AVR generator, a 3rd harmonic current will produce a voltage drop at a 3rd harmonic voltage. Likewise a 5th harmonic current will produce a voltage drop at a 5th harmonic voltage, etc. Each harmonic current drawn by the non-pfc SMPS flows through the system impedance resulting in a voltage drop at that harmonic frequency. In other words, where a distorted current waveform is made up of the fundamental plus one or more harmonic currents, each of these currents flowing into an impedance will according to Ohm’s Law, result in a voltage drop resulting in voltage harmonics appearing at the load bus and distortion of the voltage waveform.

This pattern does not appear in the voltage waveform of the grid power on the left in the oscilloscope shots above because of its’ much lower impedance.  The impedance of a generator is not an easily known quantity. Depending on its’ size and design, the impedance of a generator will be 5 to 100 times that of a utility transformer and it will change as the load changes. But where you were using a conventional construction generator, the internal reactance of the generator would have been sufficient to cause appreciable voltage flat-topping as well as a voltage drop.
 
The first step in mitigating the problems caused by harmonic currents is to eliminate the currents. Using only power factor corrected (PFC) gear will go a long way in reducing the number of harmonic currents in the power stream. A PFC circuit realigns voltage and current and induces a smoother power waveform at the distribution bus. PFC circuits successfully increase the power factor to as much as .9, making SMPSs with it near linear loads. As a result, the SMPSs use power more efficiently with minimized return current and line noise and also reduced heat, thereby increasing their reliability.


Left: Conventional AVR Generator w/  1200W non-pfc SMPS. Right: Inverter Generator w/1200W pfc SMPS

A second step in mitigating the problems caused by harmonic currents on small putt-putt generators is to operate them on only inverter generators. For example, the power waveform above on the right, is the same size load but with power factor corrected SMPSs operating on our modified Honda EU6500is Inverter Generator. As you can see, the difference between the resulting waveforms is startling. Even though the load is the same, the fact that it is power factor corrected and the power is being generated by an inverter generator, results in virtually no power waveform distortion. For this reason, sensitive electronic equipment will operate reliably and without damage on the same power. And, the generator is capable of operating a larger total load

The extremely low line noise exhibited in the inverter generator power waveform above (right) creates a new math when it comes to calculating the load you can put on a portable generator. Where before you could not operate more than a couple of non-PFC SMPSs on a conventional putt-putt generator because of the consequent harmonic distortion, now you can load an inverter generator to capacity. And if the generator is one of our modified Honda EU6500is inverter generators, you will be able to run a continuous load of up to 7500W as long as your gear is Power Factor Corrected.

These power quality issues have vexed film electricians for years.  If you haven’t already, you might want to read an article I wrote for our company newsletter that explains the electrical engineering principles behind these issues and how to resolve them. The newsletter article is available at http://www.screenlightandgrip.com/html/emailnewsletter_generators.html.

Guy Holt, Gaffer,
ScreenLight & Grip,
www.screenlightandgrip.com
Logged

Mitch Miller

  • Newbie
  • *
  • Offline Offline
  • Posts: 12
Re: Non Power Factor Corrected Pro Audio Gear
« Reply #13 on: October 23, 2014, 12:19:38 AM »

So many big words, but I love the electric theory being presented and demonstration of how it applies to this situation!
Logged

Jonathan Johnson

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 3209
  • Southwest Washington (state, not DC)
Re: Non Power Factor Corrected Pro Audio Gear
« Reply #14 on: October 23, 2014, 01:48:33 AM »

How does skin effect affect heating of the cable? Can a high frequency load cause overheating of the insulation at lower currents than lower frequency loads?
Logged
Stop confusing the issue with facts and logic!

Lyle Williams

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1558
Re: Non Power Factor Corrected Pro Audio Gear
« Reply #15 on: October 23, 2014, 02:52:06 AM »

The amplitude of the harmonics is going to fall off significantly, maybe linearly with harmonic order.   The power in those N-th harmonics is related to the square of the amplitude.  There is very little power in high order harmonics.  The frequency of these high order harmonics is still low enough that the skin effect is a few mm thick.  A negligible amount of power will encounter a negligible effect.
Logged

Lyle Williams

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1558
Re: Non Power Factor Corrected Pro Audio Gear
« Reply #16 on: October 23, 2014, 02:53:55 AM »

None of the above is meant to suggest that a screwed up waveform is good and wholesome.  Just that skin effect isn't going to have a huge impact.
Logged

Guy Holt

  • Full Member
  • ***
  • Offline Offline
  • Posts: 130
Re: Non Power Factor Corrected Pro Audio Gear
« Reply #17 on: October 31, 2014, 01:04:45 PM »

How does skin effect affect heating of the cable? Can a high frequency load cause overheating of the insulation at lower currents than lower frequency loads?

The amplitude of the harmonics is going to fall off significantly, maybe linearly with harmonic order.   The power in those N-th harmonics is related to the square of the amplitude.  There is very little power in high order harmonics.  The frequency of these high order harmonics is still low enough that the skin effect is a few mm thick.  A negligible amount of power will encounter a negligible effect.

Although losses in cables are related to I^2R, when high frequency harmonic currents are present, they create eddy currents which reduce the effective cross sectional area of the cable by what is known as skin effect.  The equation below shows the relationship of eddy currents to harmonics.  What is significant about the relationship of heat loss as a result of harmonic currents is that the harmonic current (Ih) and harmonic number (h) are squared which means that instead of increasing in a linear fashion they increase exponentially. Put another way, the heat generated by harmonic currents just doesn’t increase gradually at higher harmonic frequencies, but it jumps drastically.


(Where: PEC = Total eddy current losses, PEC-1 = Eddy current losses at full load based on linear loading only.
 
Ih = rms current (per unit) at harmonic h , and h = harmonic # )

In our industry, we are fundamentally concerned with the Triplen Harmonics (principally the 3rd), but as indicated in the equation above, higher order harmonics generate as much heat as the 3rd, but at much lower amplitudes. For instance, according to the equation above 1 Amp of the 5th harmonic will generate as much heat as 1.7 Amps of the 3rd, and 1 Amp of the 9th harmonic will generate as much heat as 9 Amps of the 3rd. For this reason, the higher order harmonic currents (5th, 7th, 9th, etc.) drawn by a non-linear load also contribute to overheating of cable and electrical components. For instance, it doesn’t take much 21st harmonic current to generate heat when, according to the equation above, the harmonic frequency (h) is squared for a multiplier of 441x (212) compared to the multiplier of 9x (32) for the 3rd harmonic. For this reason the heat generated by the harmonics drawn by a non-linear loads should not be taken lightly.

Guy Holt, Gaffer,
ScreenLight & Grip,
www.screenlightandgrip.com
Logged

Guy Holt

  • Full Member
  • ***
  • Offline Offline
  • Posts: 130
Re: Non Power Factor Corrected Pro Audio Gear
« Reply #18 on: October 31, 2014, 01:23:29 PM »

None of the above is meant to suggest that a screwed up waveform is good and wholesome.  Just that skin effect isn't going to have a huge impact.

To give you an idea of how much additional voltage drop can be caused by the harmonics drawn by non-linear loads I’d like to share a demonstration I do in a workshop I offer to members of IATSE Local 481 on “Advanced Power Quality.”


For the demonstration we run 400’ of #2 AWG banded 5-wire to a couple of Strand CD80 6x12kw packs. To look at only the voltage drop due to harmonics rather than voltage flat topping,  we power the packs off of the low impedance of the  stage power rather than the high impedance of a generator.


We use the Strand packs to compare the voltage drop caused by equal linear and non-linear loads. By simply dimming the packs we can convert a linear incandescent load to a non-linear load. As you can see in the above slide from the workshop power point, the feeder wire was run out of the stage dimmer room and back so that the packs could be in the same dimmer room from where the power is being drawn. The no load voltage of the power supply at the end of the 400ft run was 122.8V.


For the first part of the demonstration, each phase leg powers a non-dimmed incandescent load of 17kw (1 – 12kw & 1 – 5kw per leg.) As you can see above, the non-dimmed 17kw load drawing 132.8A over 400’ of #2 AWG cable resulted in the voltage dropping to 113.8V (a 9V drop.) Also note that at 100% the packs are a linear load because they draw a nice sinusoidal current and the voltage waveform is undistorted. We then checked the potential difference at the end of the cable run (before the packs) and a receptacle in the dimmer room that was on the same phase leg of the same service - in other words, the voltage at a distance of 50’ from the service head with no load and at a distance of 450’ with a 17kw linear load. As one would expect, the difference was 9V (the same as the voltage drop over the 400’ #2 AWG run.)

For the second part of the demonstration, each phase leg powers a load of 31kw (2 – 12kw, 1 – 5kw, & 1 – 2kw), but now dimmed to 43% to create a non-linear load of roughly the same amplitude  (132.6A) as in the first part of the demonstration. Again, we used that load size and dimming percentage in order to approximate as closely as possible the amperage on the conductors in both scenarios. As you can see below, the 31kw load dimmed to 43% drawing 132.6A over 400’ of #2 AWG cable resulted in the voltage not dropping, but increasing slightly to 124.5V (a 1.7V increase.)


To understand why our Fluke 34B Power Quality Meter reads a voltage increase rather than decrease, it is necessary to note that the Strand packs dimmed to 43% are now a non-linear load drawing pulsed current that is rich in harmonics (see Current waveform in the Power Point Slide above and current FFT on the left of the Power Point Slide below).


Also note that these harmonic currents encountering the impedance of the 400’ cable run results in appreciable Voltage Waveform Distortion (see Voltage waveform and FFT in the Power Point slides above).

To demonstrate that the voltage increase is in fact a reading error caused by the distorted voltage waveform, we again check the potential difference at the end of the cable run (before the packs) and a receptacle in the dimmer room that was on the same phase leg of the same service but unloaded. Even though our meter reading at the end of the 400’ run shows a slight increase, there is in fact now an 18.8V difference.

What accounts for this over 100% increase (9.8V) in voltage drop simply because the load is rich in harmonic currents? The fact that the higher frequency components of the current drawn by the now non-linear load (dimmed Strand CD80s) cause electrons to flow toward the outer skin of the 2 Awg feeder cable rather than evenly throughout it. Since the flow of the electrons is no longer evenly distributed across the cross sectional diameter of the cable, more electrons are flowing through less copper and the resistance of the cable increases. The increase in resistance reduces the ability of the cable to carry current, resulting in greater voltage drop over shorter distances and overheating of the cable.

The reason why a true RMS meter like the Fluke 34B does not read this voltage drop, is because there is roughly the same energy in the distorted voltage waveform above as in the original sine wave (less that which goes into the generation of heat.) Loads cannot use the harmonic component of the distorted waveform because it goes into the generation of heat. And since, as we saw in the graph above, the increase in voltage drop due to harmonics is appreciably more significant in smaller gauge cables (10 AWG or 12 AWG), than with the larger gauge feeder cables (single conductor #2, 2/O, & 4/O) partially accounts for why Frank’s Crown XTI 1000 ran hot and was clipping prematurely on his long 12/3 run (the voltage flat topping caused by the harmonic currents encountering the high impedance of the generator accounts for the rest.) 

These same power quality issues have vexed film electricians for years, which is why most movie lights are now power factor corrected.  If you haven’t already, you might want to read an article I wrote for our company newsletter that explains the electrical engineering principles behind these issues and how to resolve them. The newsletter article is available at http://www.screenlightandgrip.com/html/emailnewsletter_generators.html.

Guy Holt, Gaffer,
ScreenLight & Grip,
www.screenlightandgrip.com





Logged

Frank Koenig

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 1318
  • Palo Alto, CA USA
Re: Non Power Factor Corrected Pro Audio Gear
« Reply #19 on: October 31, 2014, 07:56:49 PM »

Guy, thanks for this clear tutorial. Nice work. For anyone wishing to dig deeper, the Wikipedia article on skin effect is remarkably in depth, so to speak.

http://en.wikipedia.org/wiki/Skin_effect

Two little take homes: Even at 60Hz (the fundamental in North America) the skin depth (at which the current density is 1/e) in copper is only ~1/3 inch. Also, aluminum wins. Since the skin depth varies as the square root of the resistivity, an aluminum conductor has significantly lower AC resistance than a copper conductor of the same DC resistance.

Now maybe we should all just use Litz wire for our power cables :)

--Frank
Logged
"Nature abhors a vacuum tube." -- John Pierce, Bell Labs

ProSoundWeb Community

Re: Non Power Factor Corrected Pro Audio Gear
« Reply #19 on: October 31, 2014, 07:56:49 PM »


Pages: 1 [2] 3   Go Up
 



Site Hosted By Ashdown Technologies, Inc.

Page created in 0.054 seconds with 25 queries.