A problem with AC Power you may not have considered.


My posting is not about a stereo system but it is related to AC Power, from which all stereos draw power. Read on, I am sure you will find this interesting. I certainly did and it caused me to rethink and replan AC Power to my stereo.

At my real job as an electrical engineer, I manage a cross-disciplinary engineering team for a large energy company.   We make large, residential green energy management systems, a size that borders between most large homes and utility companies. A few months back, we released a new product to the 230VAC single-phase market (Australia, Europe, etc.) and recently introduced the same product to the 240VAC split phase market (USA, Canada, etc.).   In addition to a slew of UL, IEC, IEEE, CSA, TUV, and other safety codes, we also had to meet FCC Class B emissions (which all your digital audio equipment must also meet) and also meet FCC Susceptibility requirements (which digital audio does not have to meet, unfortunately).  

Since the two products are almost identical, I thought we could leverage what we learned for the 230VAC unit onto the 240VAC unit.   Well, this is where the impact of grid power to our stereos comes into our interest.  

The emissions requirement is of two parts, of which you may be familiar. One is radiated emission, which is the noise the product broadcasts into the air. The second part is conducted emissions, which is the noise the product injects onto the power lines and runs throughout your house and probably into your neighbors as well.  

The 230VAC unit passed emissions, which I expected as we did a lot of design work to make it pass.   The conducted part was a concern, since that injected noise is from the equipment our vendor produces, not something we designed in house. Well, when the certified testing house tested conducted emissions, it failed.   A couple of weeks of debug later, at 2K$ per day, the problem was solved when I suggested they test with the grid connection running through 8 feet of steel conduit, since all installations have at least 8 feet of conduit.

Fast-forward six months to the 240VAC testing, which took place here in the USA. Surprisingly, the unit failed conducted emissions, even though we used the same 8 feet of steel conduit.   Another week of debug, again at 2K$ per day, we stopped testing since it was clear a new design is needed to fix it. I designed a 50 Ampere Balanced LEMP Filter that had over 50-dBm isolation in the affected frequency range.   Problem solved.   So, why did 8 feet of conduit fix the problem one time and not the next? A good question.  

I took the same 8 AWG THHN wire we used to connect the unit to the grid, ran it through the same 8 feet of 1 ½ inch steel conduit, and rented some high frequency test equipment. In the conduit we had two 8 AWG wires for Line 1 and Line 2, one 8 AWG wire for Neutral, and another 8 AWG for Earth ground.   I ran a bandwidth test from Line 1 to Neutral and tied the conduit and Earth wire to earth, while the other Line wire floated. The test started at 60 Hz, which I referenced as 0 dBm and I ran the test all the way to 30 MHz.   The generator produced 10Vrms, the level I checked at each step, and fed a 50-Ohm load.   To my great surprise, I had a 2-dBm rise at 10 MHz where it began to roll off and was only 2 dBm down at 30 MHz, the limit of the test generator.   In other words, that length of pipe and THHN wire had a bandwidth of +/- 1 dBm from 60 Hz to 30 MHz!   Whoa! We are allowing a ton of injected noise into our systems!

To prove that, I grabbed the power supply from an analog stereo amplifier and fed the test signal through the cord, fuse, transformer, and measured the bandwidth on the secondary.   In spite of a UL/CSA approved transformer, it was surprisingly transparent to the test signal.   Throughout the test spectrum, it was never more than 6 dBm down and it peaked in a couple of areas, too.  

Our homes usually don’t have grounded conduit, what most homes have is Romex wire.   That stuff is transparent to radiated emissions and we live in a world of radiated emissions. Think cell phones, FM and AM radio, TV broadcasts, all the communication frequencies, plus who knows what we have for the dirty noise injected by electric motors. Think your fridge, your AC unit, your furnace, ceiling fans, light dimmers, electric vehicles (that is the reason they don’t usually come with an AM radio these days!), the list can go on for a long time.

For my stereo system here at the house, I built a smaller version of the LEMP filter, added additional suppression, along with 20,000 Amps of surge protection. I am also installing a dedicated earth ground as well.   However, you don’t have to home brew – you can purchase equipment that meets the local safety codes and is LAB certified to meet multiple suppression standards. These units have strong filters in them to clean up line power. There are replacement AC line cords on the market that contain RF suppression.   I don’t suggest you get a new mortgage just to buy AC noise suppression equipment or new line cords, but I do suggest you do something to kill those RF demons.  

Look for equipment that has at least 30 dB of suppression from 100 KHz to 15 or 20 MHz. Thirty to forty dB is the range where most emission problems fade away, so that is a good starting point.   Some equipment has lightening suppression as well; look for an IEEE spec stated in joules of energy, the more the better with a test pulse of 8/20 microseconds.   Don’t be afraid to stack some of the equipment in series.  

The lighting in your listening room can also matter a great deal. Stick with plain, old school incandescent bulbs; avoid the CFL’s, LED’s, neon’s, light dimmers, and other lights that require power supplies to run.   Incandescent bulbs are very quiet, which is why they appear regularly in emission anechoic chambers.   Although digital equipment is less sensitive than analog equipment, it is not immune to susceptibility.   Vacuum tube equipment usually has an edge over solid state, too.  

I hope what I wrote is of help to you in your quest for improved sound.  

Robert
128x128spatialking

Showing 15 responses by spatialking

I am glad the posting is useful.   One thing about lightening, if it hits close enough, it won't matter if the system is plugged in or not, the EMF pulse will most likely destroy all the solid state junctions.   Vacuum tube equipment in this case might actually survive the hit.
@mrmb, @soix - Here are some ideas for noise suppression. There are a number of quality noise suppression devices out there. Furman is a good brand, so is PS Audio and Tripplite, but do check out the specs. Here are the specs for a Furman M-8X2, a basic power distribution and noise suppression, and priced at $80.

· Standard Level AC Surge Protection: Merit Class (sacrificial)

  • Spike Protection Mode: Line to neutral
  • Energy Dissipation: 150 joules
  • Peak Impulse Current: 12,000 Amps
  • Let Through Voltage (@125 amps, 8/20μS waveform): 400 volts

· Noise Attenuation: Transverse mode: >23dB, 200 kHz to 10MHz

So let’s explain what this means. Sacrificial means when a lightening pulse hits the device will short the input power forcing a fuse or circuit breaker to blow and opening the circuit.

Essentially, the unit will have to be replaced; the philosophy is it’s better to replace something inexpensive than expensive stuff downstream. This is how I designed the LEMP filter, I had all the noise filtering up front and following by sacrificial electronics.   It is good of Furman to specify this, since you know what you are getting.   The other philosophy is non-sacrificial, that is a design that has aggressive filtering and stores the energy in a reactive space, thus dissipating the energy more slowly.   A good example of this is here: https://zerosurge.com/wp-content/uploads/2018/04/2R-Series-0418.pdf and here: https://zerosurge.com/

There is no reason you can’t combine the two philosophies. J Put the heavy filtering up front, followed by spike suppression.

Spike Protection Mode means the protection is from Line to Neutral and there is no protection from neutral to ground.   This is a reasonable concept since neutral is tied to ground at the main panel. There are other arguments against this, stating neutral should be spike protected to ground. However, the more protection one adds, the more the cost goes up.   At some point, has you have to evaluate bang for buck.   Personally, I put in neutral to ground protection, both as spike and as filtering but then I wasn’t trying to hit a price point.

Joules is the unit to measure energy, equals the work done when a current of one ampere passes through a resistance of one ohm for one second, which is one Watt. Joule units define the energy dissipated by the device; the more the better.   150 Joules here is not much, 3000 Joules is getting somewhere, 6000 Joules or more is healthy. Expect to pay a lot more as these numbers go up.  

Peak Impulse Current is the maximum spike current that the surge protection circuit can handle to protect downstream electronics.   Again, 12,000 Amps is a starting point, I put in 44,000 Amperes and there are bigger devices out there than what I used. That LEMP filter I designed for the energy system had 300,000 Amps of Peak Current.

The 8/20 microsecond waveform is a standard IEEE test waveform for lightening spikes where the maximum voltage is reached in 8 microseconds and decreases to 50% in 20 microseconds.   Therefore, the higher the voltage pulse the narrower the pulse becomes.   There are other IEEE standard pulses but this is the most commonly used one.  

Let Through Voltage is an important spec. It tells you the amount of voltage that the protection circuit will allow to pass through before the circuit begins to attenuate the spike. Here it is say it will allow a 400V pulse at 125 Amps.  Understand there is always a Let Through Voltage, keeping it tame is a tough call and an engineering compromise.   Reduce the Let Through Voltage and the pulse circuit works a lot more, heats up, and is in a weakened state when the big spike comes along. Relax the design and the pulse circuit is more able to handle a big pulse but allows more energy through when it does happen. The only good solution here is a good design.

Everything I wrote about so far is about spike suppression, which is lightening and spike noise from motors, air conditioners, etc.   This has nothing to do with EMI and RF noise, which is more problematic for sound quality.   The spec 200 KHz to 10 MHz is the bandwidth where the EMI filter works and produces at least 23 dB of suppression.   In my first posting, I stated look for something that has at least 30 dB of suppression from 100K to 20 MHz, so this unit is a bit on the shy side of what I suggested.   Forty dB is much better and it is unlikely you will find anything above 60 dB.

So, is this bad? It costs $80 and for that price, the protection is quite fair. If you have the budget, definitely look for more EMI suppression. If you live in areas where high winds are common, earthquakes, or massive storms, having a lot more spike suppression is a good idea.   It is not just lightening that produces massive electrical spikes.   A storm knocking out a power pole can do a real number on your system.


BTW, I am using three extension cords on the right channel 200W Conrad Johnson amp.  It is really a kludge setup but far better than using that other outlet.  If I use the wall receptacle next to it I get a lot hum for reasons I have yet to determine.  The fix this weekend is to replace the extension cords with a single 8 AWG cord 25 feet long. 
I'll take a look at the Equitech, Furman, and PS Audio.  About that 100 foot power cord, just use a good thick one, say a 3 wire, 12 AWG, and see how it works.  
I don't want to get into a situation where I am reviewing equipment but I will try to answer all your questions. 

Here is the PS Audio device: https://www.psaudio.com/dectet-power-center/  There are no specifications given so I have no idea how much attenuation it provides, what frequency range it attenuates, or anything like that.   The parts in the photo do look appropriate, though.
The Furman IT-REF 15l and 20l do have excellent specifications throughout the bandwidth that needs filtering.  If you have the budget, these look very promising.  Certainly I would feel confident plugging one into my system given the published specifications and the fact it was tested by an independent testing lab.  Amazon has the 15l for $2500 and $4000 for the 20l.  As I said before, if you want to get serious about EMI it also gets expensive.  

The Equitech also looks very good: https://www.equitech.com/productsold/son-of-q-shelf-or-rack-mounted-chassis-systems-2/ All torridal transformers provide good high frequency isolation since it doesn't have interwinding parasitic capacitance that are inherent with the typical E style transformers.  I found it online at $2640.   The Equitech didn't say much about filtering and surge protection, it has it but I don't know what it has or how effective it is.
The thing to understand here is there is no one perfect product, if one addresses the problem, most EMI problems will fade.  I doubt you would go wrong with any of these but do understand you get what you pay for.  So, don't expect a $200 power strip to equal these EMI dreadnoughts!  Given what I have seen here, I'd say buy one of these and forget about EMI issues.  

Case in point - a number of years back I was working for a very large, well known test equipment company, of which I am sure everyone knows the name.  We were pushing the bandwidth limits a standard FR4 controlled impedance PCBA could manage, so we made a test board with exactly a 50 Ohm line, a line with 10% error, a line with 30% error, a line with 50% error, and a line with no controlled impedance whatsoever.   What we found was even the 50% error line had reasonable performance while the one with no control was not usable.  If you address the problem, even with error, you are far ahead of the game than ignoring it.

One of the nice benefits from these units is they monitor the line and shut down during long term brownouts or over voltage conditions.  The utility companies have specifications on this sort of thing, call "ride through" where the user equipment can ride through the brownout or surge without self destructing.   Back when I worked in audio, I knew a few audio companies tested their products for this but I doubt the majority do.  The idea is products have a given ride through while the utility companies work to prevent problems that exceed the general ride through spec.

I didn't look at the regenerative products such as the big ones from PS Audio.   These are an entirely different animal since your system is now powered from the power supply / amplifier / inverter contained within.  The grid becomes secondary and not really in the picture.   Unless you have horrible line regulation, you probably don't need to go to this extreme.  

One thing is for sure, I would definitely spend $400 or $500 on EMI filtering and surge protection before I would spend $400 or $500 on spiffy line cords.  Ignoring the awesome cosmetic improvements they provide, this equipment is very real and very effective.   On the other hand, if you want really spiffy and sexy looking power distribution after you buy one of these ..... :-)
Did you have the wiring in your house reviewed by an electrician after the lightening strike?  Did the left channel get repaired as well?  It is also possible the cabinet has developed a leak or a joint has come loose.  Keep looking, you will find it.
Update on Equi-Tech:  I found a more detailed description online about how it works.  Yes, the entire balanced line concept is a well known and well defined power distribution concept.   Envision how balanced interconnects work except this is for your grid power. 
Do be aware if you spring for this - currently there are no NEMA receptacles or plugs designed for balanced 120VAC power.   Balanced 240VAC, yes, there are options but not for 120VAC.  This means you can't get a building permit to install it and the NEC and UL/CSA/IEEE governing bodies don't recognize it. 
Here in CA, if you install this without a permit and your house burns, even though it is no fault of your own, your homeowners insurance isn't required to pay for the damages.   There are other states that have similar laws as well.   It pays to investigate.  If you can get a permit, you are home free (not to make a pun).
Currently, Equi-Tech is working with Intertek (the same safety company we use for our high energy equipment at work) and the NEC to get this concept approved.  I have no idea when that will occur, safety regulation changes take years to happen and I don't know how far along they are.   NEMA will also need to get involved since it will require a new plug receptacle format.  If you are considering buying one of these, call them and find out where they are in the approval process.   Get all the documentation from them and any suggestions on getting a permit to have it installed.  Then call your local building permit folks, have a meeting with them, and see if you can get a permit.  A permit is required since it doesn't plug into your outlet, rather it ties into your main distribution panel, so a permit is definitely required.
It is a great concept for audiophile power, it's a bummer about NEC.  The closest thing in the NEC handbook is called a separately derived system with an isolated ground.   That is a good concept and you can get a permit for that but it is not a balanced power distribution network like Equi-Tech.
I was at Home Depot last night and I picked up this unit for my home workstation.  For $40 it is a very good unit and well worth the cost.  https://www.homedepot.com/p/Commercial-Electric-12-Outlet-USB-RJ45-Coax-Surge-Protector-with-6-ft-Co... The webpage doesn't give all the specs, it has RF filters providing 43dB of attenuation from 150KHz to 100MHz.  It also meets UL 1449, more on that below. It also has both TVS and TSR protection.   All in all, a good unit for the money.  I did note they specified 4500 Joules of max energy dissipated as the sum of L-N, L-G, and N-G, but they did clarify this on the back of the box.  3300 Joules from L-N is pretty good; 6600 was the most but that unit was more than twice the money, too.  My continual gripe with all these units is the plug layout.  To save space (cost) they squish them together so if you have large plugs you can't use all the outlets.  Likewise if you have the plug in transformers or wall-wart units.   
About UL1449 - UL got a lot more restrictive with leakage currents in this latest revision. In so doing, the "let through voltage" is higher now than it was in the previous version.  This unit has a 500V let through voltage, which is the consequence of UL1449.   If your equipment has any internal protection, it should handle this easily since it isn't present for long.  However, I'd rather have something in my house meeting UL1449 than something that is not.   It might self destruct if I get a big enough surge but I am not worried about it bursting into flames. 
Either way, I don't see any AHJ issuing a permit for residential use of this equipment.   Yes, NEC 647.7(4) allows for use in restricted situations but as you pointed out a residential home does not qualify.  Gosh, I wonder if the audiophile owner would even qualify as "qualified personnel" under NEC definitions?
I'd like to see an application where an AHJ signed a permit for the orange hospital ground NEMA 5-15R or NEMA 5-20R receptacle for a balanced 60VAC line.   I don't see that happening, especially in the stronger cities such as New York or Los Angeles.  (Stronger here meaning a number of AHJ's in the city sit on the UL or IEEE code boards)   Neutral has to be tied to earth, not to the end of 60VAC transformer leg.    Granted, maybe most equipment can deal with 60VAC on the neutral leg but I'd never want to bet my money all equipment is okay with it. 

An AHJ might buy off on NEC 647.7(4) for a 60 VAC single ended 60VAC line using the NEMA 5-15R or -20R receptacle in a restricted area but I sure don't see it happening for a 60VAC transformer leg on the neutral line.  However, that being said, they might indeed sign off on it if the equipment was hardwired, though.  That is surely a thought to ponder.
If you know of a installation that had a permit signed off for a Balanced 60VAC line using a Hospital Grade NEMA 5-15R or -20R receptacle, I sure like to hear about it.
That is good to know!  I wouldn't expect LA to approve it, given how picky they are.
Yes, the Furman is very similar to the Equi-Tech unit and no permit is required to install it, a most definite plus.   I am sure it is a solid unit but like all balanced 120VAC units, there is no safety code approval for this.   The datasheet did state a NRTL tested it but it did not state to what UL, IEEE, CSA, or ASTM code. 

If your equipment can be run from 240VAC split phase, such as our USA power, then you can have a UL recognized balanced power source along with NEMA connectors.  Our 240VAC is a balanced power line and twice the power available at 20A.  One would still need to dig up filtering and surge suppression though.
A reputable company like Furman is probably fine without a UL certification on the back, as opposed to a much smaller unknown company.
I am not sure what you mean by shunt mode filters - you mean like TVS diodes and MOV's?   If so, yes, they are sacrificial and will blow if they are overpowered and aren't designed properly.   UL1449 approval means they won't burst into flames and should die harmlessly.  The problem is, they will die and you may not realize it!
I don't believe the torridal transformer in the Furman unit is high impedance. I've designed torridal isolation units before and they supply surge power quite readily.   It is higher impedance, though, than plain copper wire.
I agree with everything else you said, though.   Picking a cutoff frequency for the low pass RF filter as well as the design topology is as much philosophy and art form as it is science!   I probably wouldn't pick a Fc lower than 3KHz - remember you have a Power Factor and harmonics on the grid to deal with.    However, the ones I built are usually around 8KHz to 10 KHz, so that isn't too far from 3.
That Furman is a good looking unit, I'd buy it if I had $3100 to spend on line isolation and conditioning.   At least it has very good specs!