Welcome to Euphoria Speaker Design.
The crossover is the heart of a speaker. It is our belief the sinusoidal waveform (music) must be preserved. The crossover should get out of the way. It’s only function is to separate the signal for each driver. Our crossover is not based on my theory. There is a mathematician in Asia. He proved mathematically an LCLCLC shunt circuit is the ideal way to separate frequencies. I simply put his theories into practice, and balanced it.
A caveat to start: Our testing and development system is fully balanced (this means all our amplification uses differential circuits). Our speakers take advantage of the balanced stereo chain. If there are RCA (unbalanced, single ended) cables in your amplification chain, some of what we describe in these pages may not be fully realized. The same for unbalanced components.
Technically speaking, we’ve developed a quasi second order 4-way series crossover, and balanced it. This is not something new: Bud Fried was almost there in the 70’s, Ray Kimber’s diAural circuit is a close cousin, and Jeff Bagby built a 2-way quasi second order series crossover that is apparently fabulous.
We call ours: The Fully Balanced Dual Cascade Parallel Series 4-way crossover.
The following is a guideline. Our speakers are custom ordered.
QAIDAM: $18,900.00 Cdn. Scanspeak D2608 tweeter, Morel EM1308 midrange, Acoustic Elegance TD6M mid woofer, Acoustic Elegance TD12H woofer. Mundorf EVO, Supreme EVO, and AXON capacitors. Jantzen coils. Lead time 5-6 months.
KUNLUN: Starting at $31,600.00 Cdn. Accuton ceramic cell concept 1″ tweeter, Accuton 2″ cell concept ceramic midrange (arguably the best midrange made today, with the exception of the 2″ Diamond midrange). Acoustic Elegance mid-woofer and woofer. Top of the line Mundorf crossover components. Lead time 5-6 months.
Our reference, the PINNACLE, starts at $68,000.00 Cdn. Lead time 8-10months (the laminated Baltic Birch ply needs two months to cure and be re-sealed).
A quick thought. If you’re like us, you’ve never been satisfied. You are continuously adjusting Azimuth, Side Rake Angle, Cartridge down force, etc.. Maybe you’ve purchased dozens of interconnects, power cords, line conditioners, sound absorbing panels, sound diffusing panels?. How many different components have you owned??? Nothing gets you there.
We’ll suggest we figured it out. It’s the speakers crossover. From this moment forward, we are unlikely to implement a parallel crossover ever again. Your ears will be the judge.
The waveform coming out of the dual differential (balanced) amplifier is exactly what the speaker driver should see.
Let’s look at what happens with a normal parallel crossover, with a single capacitor or inductor in the signal path. This would be a 1st order crossover.
The signal coming out of your amplifier is the blue trace. It is simply a perfect sine wave. Your dual differential amplifier is doing it’s job.
Once your amplifiers perfect wave goes through your ‘normal’ crossover, the purple trace is what your speaker driver sees. This is unacceptable. This is a type of distortion. There is pre ringing, there is post ringing. This is what we’re all trying to get rid of. This is why we are all on the upgrade merry-go-round.
Some of us love full range planars and/or full range drivers. There is no denying their cohesiveness. However, we know their limitations. Euphoria Speaker Designs goal was to develop a dynamic 4-way speaker which preserves all that is perfect with the crossoverless full range speakers, and makes fabulous their minor deficiencies.
This next picture is the beginning of my crossover explanation. It is absolutely imperative we understand the balanced amplifier/speaker relationship. Forget the crossover for a moment. Keep it simple. The positive terminal of your amplifier pulls the waveform UP. This action causes your driver to move outwards. That’s it. That’s all.
This next picture is obvious. The negative binding post of your amplifier pulls the waveform DOWN. This action causes your driver to move inwards. Do this 1000 times per second, and we have a 1000Hz tone. This is how a dual differential amplifier works. Unbalanced amplifiers do not pull the driver all the way back in. I believe ‘damping factor’ is the term used for pulling the driver half way back in. This last statement may have to be edited. Time to study some more.
Now we get to the beginning of our ‘Balanced’ explanation.
If a SINGLE capacitor is added to our crossover, and we assume this capacitor is on the positive terminal of our driver. The positive portion of the waveform ‘lags’ 90 degrees. To be more specific; voltage lags current by 90 degrees when going through a capacitor.
There is a moment of distortion when the negative portion of the waveform has started pulling down. This happens before the positive portion has reached it’s peak (we lose 3dB of TRANSIENT output. We lose a little bit of focus.).
The purple trace of the first picture shows a smooth line.
Unfortunately, for some of us, we hear this ‘distortion’ of the true waveform. This is why we are constantly ‘tweaking’ our system. We’re trying to get rid of it.
The next picture shows what happens if we install our capacitor on the negative side of a driver.
What happens if our driver has a capacitor on the positive side (90 degree lag before moving outwards), and one on the negative side (90 degree lag before moving inwards). Identical cap in both construction and value. Each half is shifted identically (one manipulated by your amps positive terminal, the other by the negative).
Until you hear a ‘balanced’ crossover speaker, driven by a fully balanced amplification chain, you’ll have no idea how incredible it sounds.
A quick note; we all want the performers to ‘appear’ in our listening room. This requires the brain to reconstruct the event in our heads. Unfortunately for a lot of people, something known as Auditory Spatial Acuity is needed. From our observations, the following is a guideline: if you can parallel park easily, then you have spatial acuity. If you can’t parallel park, our speakers will sound ordinary.
Let’s show you I’m not entirely stable.
I will first thank Elliot Sound Productions. I will then reprimand Elliot Sound Productions.
Read the article, it’s kinda mandatory.
There are two points that stuck with us.
1) A parallel crossover has a pronounced impedance swing. A series crossover a really mild dip.
This is one of the reasons we use series crossovers. To repeat, our crossover and our speakers are voiced to be used with tube differential amplifiers. This type of amp sounds best with stable impedance. There are a number of Class A Dual Differential Solid State amplifiers, which will also sound fabulous. They too benefit from stable impedance.
2) This next point is the point that floored us. This is what our crossover gets rid of. This is the ‘sound’ we’ve been forced to listen to for decades.
Read this over and over, and look at the picture carefully: Elliot Sound Productions input a pure square wave into the crossovers. The SERIES crossover outputs a pure square wave. Perfect. The PARALLEL crossover outputs …………?. Everything outside the square wave is wrong. The aqua (parallel crossover) changes the incoming waveform. The high end industry doesn’t mind. We all keep buying the next great thing because instinctively we know it must exist. The problem was, we never knew what ‘it’ was.
Did I mention we’ll never use a parallel crossover ever again?
After showing this graph, Elliot Sound Productions went on to conclude series and parallel crossovers are not that different. This was a great disservice to audiophile’s.
THE EVOLUTION OF WHAT I BELIEVE IS THE PERFECT CROSSOVER
This will not be easy. Not to explain, or to understand. I apologize this first web page is so long. You can ‘skim’ over this next point if you’re getting bored.
Please remember the positive terminal of your amplifier manipulates the positive portion of the waveform, and the negative terminal of your amplifier manipulates the negative portion of the waveform. A speaker driver needs both halves in order to make sound. It needs to move out, and then it needs to move in. Simple.
This first picture is a computer simulation of a standard 4-way series crossover. We will start here. This is obviously not our final topology.
- Let’s follow a frequency of 10,000Hz. This is intended for the tweeter, and the tweeter alone.
a) The positive portion of the 10KHz waveform comes out of the amplifier. It travels to point A where inductor L1 (820uH) blocks it. It proceeds to point B (330uH), where it is also blocked. It easily passes through capacitor C4 since capacitors allow high frequencies to pass. The tweeter sees the positive portion of the 10KHz waveform.
b) The negative portion of the 10KHz waveform comes out of the negative terminal of the amplifier. It travels to point D where it can easily pass through capacitor C1 and then C2. Let me explain a few other things. If the signal goes to E, it can pass through C3, but is then blocked by L3. It can’t go through the mid woofer. At F the signal is blocked by L4. Told you this wouldn’t be easy. Current travels at almost the speed of light, a few road blocks won’t change a thing.
The picture below shows the final path of the negative portion. It passes through both C1 and C2 easily. The tweeter moves.
We don’t like that the negative signal had to traverse two capacitors. This shifts phase an extra 90 degrees, but only for the negative portion of the wave form. Distortion is added, and 3dB of transient is gone. Remember this is the old way of doing it. Keep reading.
2. Our Mid Range plays 1KHz to 5KHz. Let’s use 3KHz as our example.
a) The positive portion comes out of the amp, gets blocked at A by L1, proceeds to B where L2 let’s it pass. Even though it also gets through C4, our tweeter has an impedance (6.8 Ohms). Our inductor has an impedance of 0.5 Ohms. The path of least resistance. The tweeter also sees very little of the negative portion (of 3KHz (capacitor C2 blocks lower frequencies)). Controlling the flow of current is how a Series crossover works. Simple.
b) The negative portion goes through C1, and onto the mid. Use you finger to trace other paths. Yes the negative portion tries to make its way to the Mid Woofer, but there is little of the positive portion, so less sound. Because the mid woofer does get a little of the 3KHz signal, it must be well behaved, read no cone breakup at higher frequencies. The tweeter doesn’t see the negative portion.
3) You’re getting the hang of it. Trace it if you like. 600Hz Mid Woofer path.
4) The woofer. 100Hz.
This schematic was a good starting point. It helped us visualize the path of frequency dependent currents. It has a few values which are correct. Some are way off. Unfortunately its not balanced. The standard way of calculating a 2-way series crossover is also not applicable to a 4-way.
The evolution of the ‘Dual Cascade’.
I look at it this way;
We are trying to build the ultimate speaker at any given price point. There is an option of using the Accuton BD20 3/4″ Diamond tweeter in our KUNLUN and PINNACLE speakers. This is one of the most accurate, lowest distortion tweeters on the planet. For this tweeter we want to use the best capacitors available. It’s not enough to use the best. It must also cause no distortion. If we look at our computer schematic, C4 is the only capacitor the positive portion sees. This is good. What about the negative portion. This signal needs to go through C1 and C2, bad. We therefore need a more direct path to the tweeters.
Inductors; We want copper foil air core to feed the mid range and the mid woofer. For our woofer we absolutely want a low Rdc ‘cored’ vacuum impregnated inductor. Looking at the simulation schematic, our woofers positive signal goes through L1 (air core) on its way to L3, and then onto the woofer. This too is bad.
This is our very first drawing of a single cascade. Each inductor feeds one driver. No sharing. Good start. (November 29, 2015) It was pointed out to us to look at the DiAural circuit, patent #6,115,475. Everything Ray Kimber said about his crossover is also true for ours. I hope he doesn’t mind us using his name.
This is the beginning of the ‘dual cascade’. Each capacitor feeds one driver. We’re not yet ‘balanced’, but proceeding nicely.
This is how we will draw future Dual Cascade schematics. This makes assembling and soldering easier to visualize.
Don’t look at the values. This was drawn when we still followed the ‘normal’ series crossover guideline. 1/2 an equivalent parallel crossovers inductor, and twice the cap value. For a 4-way this thinking is wrong.
We now need to add the ‘balancing’ components. We experimented with a few ‘placements’ for the balancing components. I won’t bore you with all the measurements. The picture below is the only orientation which negated ‘free air resonance’ of our drivers. Distortion measurements plummeted with this configuration. There are no L-C-R networks needed. No Zobels are needed. If you choose your drivers properly, no resistors are needed.
This next part is important. You will notice C1 is the same value as C1B. If testing changes this value, both components need to be changed. This applies to all components. Everything is matched and therefore balanced. The driver needs to move ‘out’, exactly the way it’s going to move ‘in’.
The following is the final schematic. Once we fine tune the values, we will update the picture.
Our crossover starts out as a simple 1st order series crossover. The ‘balancing’ components are there to restore 3dB of transient information as well as focus’ the sound (wait till you hear female/male singers, guitar picks, and drum solos (Oh my, the drum solos, not in your wildest dreams)). Some of the ‘balancing’ components appear to make it a 2nd order filter. This is simply a byproduct. Our slopes emulate a second order 12dB/octave low pass, and a third order 18dB/octave high pass. We now know this is called a Quasi-Second-Order.
If you try to think of it in conventional crossover terms, you will be lost.
A quick re-cap, some conclusions, and more ramblings. I can’t believe you made it this far.
Lets show a previous picture.
Lets visualize the whack of the drum stick on the side of a snare drum. If you’ve ever been close to a drummer, this whack is dramatic. You jump.
The blue trace is what your amplifier is playing. The stick hits the drum, the transient is massive, and then its gone. It stops on a dime.
The purple trace is a normal crossover. The purple trace goes HALF as high as the blue trace. Half the height means HALF the volume (of the transient whack).
There is an overhang with a parallel crossover. The impulse response is still playing. Your snare drum whack is no longer true. Its a facsimile. Your listening room is still energized with the same volume of sound over the same period of time (think Kepler’s law of planetary motion). Just not what it was suppose to be.
We believe our speakers play the whack perfectly. Every other sound is also perfect. A 4-way guarantees each driver plays well within its comfort zone.
Problem: our base has no pre-ringing or post ringing. There is no overhang at lower frequencies. Our speakers will not overload a room with base frequencies. Since this is something most of use have become accustomed to, it will take time to get used to clear concise controlled base.
Thank you for making it this far.