Tuesday, September 3, 2024

Measuring Acoustic Offset via Interferometry

This post is about measuring the very small difference in distance between a listening location and the speaker drivers. 


Why?

All speaker driver's have what's called an acoustic center, a place from which the sound emanates.  This place is located towards the listener or away from the listener based on the speaker type, size and general geometry.  Consider a 2-way speaker with a dome tweeter and 10" woofer. 

The acoustic center of the tweeter may be in front of the baffle, while the woofer is probably several inches behind the baffle.  This means that for frequencies where they overlap the woofer's signal will lag the tweeter significantly, causing a phase misalignment, which then will cause additive (constructive) and subtractive (destructive) interference.  We'll show more on this below. 

A crossover designer needs to know this physical difference in order to accurately match the phase of the high and low pass filters.  Getting this right often requires inverting a driver's polarity or adding or removing a pole in one of the filters.   This is a major reason why we turn to crossover simulators like like VituixCAD or XSim but for them to work correctly we must know the relative distances between two drivers. 

Here we present a simple method of determining the relative distance.  For the example below we'll take actual data from a speaker I built, but we'll remove the crossover entirely.  This design uses an AMT tweeter with 7" mid-woofer:

https://www.diyaudio.com/community/attachments/1724678348818-png.1349184/ 

 

Safety Tip:

It's always good to keep in mind the delicacy of most tweeters, so you want to either use signals which have no bass or put a large (47uF is a good value) capacitor in series with the tweeter before testing.  If you do, include it in your simulation before proceeding.  Obviously, this testing does not require the measurements to be very loud, just loud enough so that at 1m you have a clean signal that's above the noise floor.  Usually you can do this well below 2.83V (1W at 8 Ohms). 

Step by Step

The first part of the process is simple:  Measure the individual drivers.  Save the results as FRD files.  In the chart below these are the blue and yellow lines (offset for clarity).

Second, measure the combined output of both drivers.  Save that and load it as a reference line. See the green line, below.  It looks rather ragged right?  This is what happens with two drivers and poor phase matching.  The simulator's idea of the output (in black) and the reality (in green) are not matching up. 

At this point you should have 3 FRD files.  One for each driver and a third for the combined output.  For each driver, load the appropriate individual FRD, as done in the schematic, above though you can't tell from the picture. The combined input you'll load to the FR plot as a reference line. 

I'll now show you what this all looks like.

In the chart below I show the XSim simulation with no woofer delay in black and the actual in green. The green one is what I measured driving both drivers at the same time without a XO. We start the simulation starts assuming there is no delay but the reality is that there is an unknown amount of delay and therefore the simulation and reality won't yet match.

The last part is to incrementally add delay to the woofer in the simulation, and as you do so the black line will become the green line:

 

Doing this is like setting the focus on an old-school camera.  You turn the focus dial until the image just passes perfect and then you turn it back a smidge and you have achieved ideal focus.  In this case you increase the simulated woofer distance until the simulation just passes the ideal, then turn it back and you now know the effective distance from the woofer to the tweeter. 

I hope this helps demonstrate how these measurements add up.  Worth pointing out that the sonic signature in the overlap area (~2-7kHz) isn't really part of either driver but it's own unique result. Also worth noting that the ugliness of the green line is completely eliminated once the actual crossover is put into place.  You don't have to worry about this being something you need to fix, demonstrated by the final result with the crossovers in place is below: 

As you can see, there's nothing of the  green line or black line left once we are done with the crossover design.


Monday, July 22, 2024

Surge Protection for Home Networks

There are a lot of blogs recommending products you should buy but this time I'm going to start by recommend what you should not buy. 

If you are reading this blog chances are good that you have an Internet service provider, and most of us are using copper based Internet services.  Meaning there's a coaxial cable that comes in from the outside, hooks up to a modem / router and then may go to high performance devices via a copper Ethernet cable.  That copper continuity is the complete surge circuit.   If your provider has an optical fiber to your home you may be in better shape, but long internal Ethernet runs can also pick up a surge. 

 

TL;DR

Don't buy Ethernet surge protectors that connect to ground.  At all.  That includes separate Ethernet surge protectors, rack mounted units and those included with surge strips.  Yes, that includes Furman. 


 

Why not? 

At least a couple of engineering industry newsletters I've seen say the same thing: The presence of grounding Ethernet surge protectors is associated with higher, not lower, chances of surge damage.  The technical argument they make has me convinced me they actually make things worse.  

To make a long story short, there are two types of surges.  Common mode and differential.  Differential is more dangerous as it can cross the transformer boundary at much lower voltages than common mode. These types of surge "protectors" can convert common mode surges into differential.  In addition, common mode surges are (pun not intended) more common or frequent.  So by using these devices you turn a potentially non-issue into a device zapping event. You can read about the research and science in this link.

In a common-mode surge both wires have the same voltage.  In a differential-mode surge the voltage between the two wires is not equal, and this difference in voltage is what can slip through the  transformer.

At the end of each Ethernet wire pair is a tiny transformer.  That transformer has at least 2 kV of  common mode isolation built in.  This means if a 2kV surge is common mode the output of the transformer is still 0 V, or at least severely attenuated below 2kV.  However, a differential surge of practically any voltage can cross the transformer barrier, sending potentially hundreds of volts through your gear until it finds a path to ground.  A hundred or two hundred volts on the equipment side of the transformer will happily jump through micrometer sized solid state elements burning up everything in it's path.  Remember, most surge damage isn't even visible.

 

Research Has Been Slow

Knowledge and research on Ethernet surge protection has been slow in coming in large part because Ethernet surge damage is pretty rare, and users of Ethernet surge protectors even more rare.  Also, the damage per incident is low vs. say a house fire or traffic accident.  Damage to your TV is just not going to be reported or investigated the way a house fire would be.  Even significant damage in a data center doesn't usually make the news.  As a result of this a lot of surge strips with Ethernet protection are 20 years behind the latest research, or more.

Further complications arise in that there are at least three major failure/surge patterns which are discussed.  The one thing that IS pretty clear is to avoid these types of surge protectors. 

Having said this, I've had multiple surge events enter my home from outside coaxial sources, and I have a lot of gear I can't afford to replace hooked up to Ethernet cables so for me network surge protection is not optional.  I remind my readers that damaging surges are rarely visible or dramatic.  Those who think electrical surges only happen when there is smoke coming from your carbonized roof will never understand the fully justified paranoia of being a videophile and IT worker in storm prone areas. 


What are my alternatives?

External Coaxial Protection 

Outside my home at the grounding block where the cable provider's cable ends and my internal coaxial begins I use a surge arrestor with a 90V gas discharge tube.  In my experience the units are not perfect but they lower the surge voltage so that it is very unlikely to jump through multiple devices.  Also, note that while this is a grounding arrestor it won't cause a common to differential mode conversion unlike a grounding Ethernet surge protector.

If you have any external antennas you should absolutely use one of these for each of them as close to if not ON the required external grounding block.  

In the Boston area we had ferocious wind storms which took out our DirecTV receiver multiple times until I put one of these units in.  It seems those antennas are susceptible to developing a static charge due to the wind rubbing off electrons on the surface and then zap!! Yes, the cable was already grounded.  Didn't help as much as one of these units.  Also, there was no visible damage nor was there any smell.  All we knew is that the receiver no longer received and this would happen suddenly after a long gust of wind.  That should be a good indicator of how small an event can destroy sensitive equipment.

From a variety of sources I've been discouraged from using coax protectors that come with surge strips as the quality is variable and often degrades the signal and a major cause of service calls.

Create an Air Gap 

If you have incoming coax the next thing you'll need is to make sure your modem and router are separate devices in order to put a solution in between them.  Any Ethernet cable plugged in to the cable provider's modem is a potential surge path so we have to find a way of isolating the cable modem from your computers and streamers.  Once that is done we have a couple of different solutions.

Ethernet to Fiber Converters

Air is a fantastic isolator, providing about 30kV / cm of isolation.  WiFi connected devices are  naturally immune to network cable surges but not as high performing as Ethernet.  We need a solution that is completely immune to network cable surges but as high performing as the cable modem.  The answer is to use fiber based network cable in between our modem and router.  Fiber uses light over glass or plastic conductor to send the network signal.  It has no metal componet at all and so is immune to electrical surges.  Optical cables can carry in excess of 10 Gbps, well in excess of most home Internet bandwidths and definitely overkill for anything short of a hotel.  More bandwidth does not make your audio streams sound better I promise. Just match what your modem provides, which is now usually 1 Gbps and you'll be fine.

Use a pair of Ethernet to Fiber converters with fiber in between them to block all incoming surges.  Should a surge enter from outside your modem and maybe even the first converter may fry but anything downstream of the fiber cable will be safe.  Make sure to have a high quality surge protector or UPS that your modem connects to.  This approach is the most complete protection you can get, but it comes at the cost of additional wall warts and cables.  The good news is that this method is 100% plug and play.  There's nothing to configure, no software to load or web page to sign into.  Plug everything together and it will work.

Make sure you physically separate your modem and router as well! Having one sitting on top of the other can provide a chassis to chassis surge path and defeat the entire purpose of this exercise.

Carefully check the cable requirements for your converters as there are multiple types/sizes of fiber which can be a little maddening.


 

Use a Medical Grade Isolator

A simpler alternative to the fiber converters is to use an isolator.   If you are an audiophile you may already be considering using a boutique Ethernet ground isolator.  I'm going to advise you to instead use a medical grade unit like this one.  They are sometimes cheaper than an audiophile marketed unit but these are thoroughly tested to add about 4kV of common mode isolation.  Some manufacturers call these "medical grade" others "hospital grade" because the standard is meant to apply in patient care situations.  What is important not the name but that they are tested to the IEC 60601-1 standard. Boutique units can be more expensive and lack surge testing.

Note that unfortunately in the rare case when there is a differential surge these units won't be as effective as the fiber approach, but they completely avoid the mode conversion problem, and it's super easy to install. 

Which approach should I use? 

Personally I use all three approaches.  

  • Outside: Gas discharge coax protector at the grounding block
  • Between modem and router: Ethernet to fiber conversion with a two meter physical gap between them
  • At the end of long (20' or more) Ethernet runs inside the house:  Medical-grade Ethernet isolators

The potential for the largest surge generation is a strike that strikes and travels through an outside cable or piece of equipment.  That's where I have the highest and redundant surge protection.  

The next danger area is for an outside surge to come in through magnetic coupling.  This is why I use isolators at the end of long runs.

Of course if your equipment is expensive, you like having perfect ground isolation, or say you are running the network to another building running fiber instead of Ethernet may be a good choice.


What if I use Fiber or 5G Internet? 

These types of connections are inherently immune to surges from the Internet provider but long Ethernet runs inside can act like antennas.  The longer the run the more likely they can induce a surge.  For these cases putting a medical grade isolator at the farthest end from the entrance is probably your best form of protection.


Should I really not trust my surge strip?

If you follow my blog you know how big of a fan I am of Furman power conditioners with SMP and LiFT for surge protection as well as noise filtering.  They are still the very best you can get, and still I think you need to avoid any built-in Ethernet protectors.  The surge technology used for Ethernet is not the same as for power and based on outdated understanding of Ethernet surges.  Honestly researchers are still working on developing better understanding  of how Ethernet surges happen and only in the 21st century has this moved forward.  Furman's Ethernet protection doesn't take this research into account.  The AC power conditioning and surge protection is still state-of-the-art however.