Tube Type or JFET Electronic Crossover, No Op-Amps, 180 or 360-Degree Phase

Here is a tube type active crossover, with schematics, suitable for 12ax7/12au7 type tubes. Not sure it does all you want, but should be adaptable, or even convertible to JFETs (I'm not that clever with electronics). You're right, not too much of this stuff about.

What does a 360 phase shift do?

This sounds like an audiophile type design.

Basically filtering components cause phase shifts in increments of 90 degrees. so, two 90-degree shifts brings you to 180 degrees. Then everything can be brought back in phase by reversing the leads on the tweeter which adds another 180 degrees of phase shift, 90+90+180=360, and brings everything back in phase. The Linkwitz-Riley filters do that by taking two 180-degree phase shifts in series, 180+180=360, bringing everything back in phase again. The L-R filters also have the sharpest demarcation at 24 dB/octave and are basically the only design found in pro audio electronic crossovers. It is desirable to hold the number of active circuit elements to a minimum for sound quality, so I would actually prefer a 180-degree phase shift active filter (versus 360-degree) and then just reverse the speaker leads where necessary.

Then you are looking at a second-order Linkwitz-Riley filter with 12 dB/octave.

Honestly, the designations of the various filter topologies is confusing me, to say the least. I saw the Linkwitz-Riley described as two Butterworths in series. I then saw some discussion about differences in the critical crossover area between the low-frequency filter and the high-frequency. It seems to me that you want the crossover to be as sonically seamless as possible, so why these different crossover slopes? We don't want attenuation in the crossover zone, and we don't want boost. So, which topology is best? I thought ideal was for the two to "cross" at 3 dB down? Maybe before I get too far into this I should get a frequency generator or pink noise generator?

You will find this site an excellent primer on the subject (and others).

I use 6L6GC pentode for the bass modules (15-inch active passive) and 6V6 for the midrange and highs. I am currently using a pro audio fully adjustable electronic crossover. For some reason that escapes me, even tho the midrange drivers are capable of getting way down in the bass range, it sounds best crossed over at 1 KHz. I did, as an experiment, deploy a passive three-way crossover and the sound quality was noticeably degraded (I know that could be from a number of factors). When I looked at building a passive crossover with quality components, the cost (especially for the inductors) really caused me to re-evaluate. The bi-amp was an experiment and (in my system anyway) made such a huge diff that I'm pretty much a believer. Plus, I can match the levels and, as I have no active equalization (kind of punks out the sound) the crossover allows me to boost the bass on poorly mastered material. I'm interested in discovering if getting away from op amps will make a diff. I built the 6V6 ultralinear amp for less than half of what a KT88 amp costs you. I haven't used KT88s. One of the things that motivated me was I wanted to engineer an audiophile system that is affordable, within reach of anyone who is fanatical about hi-fidelity. I also noticed that KT88 designs seemed a bit more "touchy" than most, not sure why that is. My comments on KT88 should not be construed as disparaging to you or KT88s, just covering my posterior :)

It sounds like you are on the right track.

Some observations:

First, the lower end crossover should likely be closer to 500 to 600Hz. It is very hard to find a 15" driver that works well up to 1 kHz without sacrificing directionality. Yes, there are some drivers that will such as the Acoustic Elegance TD15 series, but most start beaming well before that frequency.

You also need to consider cone breakup with 15" drivers. Again, they tend to do this by 1 kHz or earlier.

Remember, 50% of the total audio power in a high fidelity system lives below 500 Hz. That is another reason why selecting a crossover frequency around that point works well.

Obviously, room acoustics are the big gorilla in the room, but I would recommend exploring closely directionality when picking a crossover frequency, not just frequency response of the driver. Directionality has become my new campaign in my latest design. You should match the off-axis frequency responses for each driver as closely as possible.

My latest design uses a ribbon tweeter (Bayma TPL-150) coupled to the woofers in such a way that the total directivity is very wide through the whole spectrum. My previous system did not fare so well. Now I can sit anywhere on my sofa or even to the sides and still be in the listening sweat spot. This is a huge difference compared to the cone mids and tweeters I was using which has about an 18" circle where they came alive.

If your current midrange is not performing well enough below 1 kHz you may want to reconsider another driver.

Bi-aping a 3-way design is a great idea. The passive crossover has little ill effects at the higher frequencies, so using a passive crossover for the mid to tweeter will work well while reducing the complexity of the active crossover if you want to stick with tubes for that job.

When designing quality speaker systems, testing has always been the limiting factor. The room contributes so many reflections as to confound measurement. You can do this better outside in a quiet environment with the cabinets suspended high into the air.

I probably have $1,000 in testing equipment and I always find getting quality measurements the biggest challenge, particularly with ported cabinets holding 15" drivers and weighing close to 100 pounds each.

Testing is the best way to really understand what the system is doing and explain why certain crossover frequencies perform better than others. I suspect that some of your midrange issues with a lower crossover point may be due to directivity and how that interacts with your room.

As far as engineering an audio system that is affordable, tubes are definitely the hard way to go. Regardless of the amplifier technology you choose, the biggest bang for the audio buck is in the speaker system and the room acoustics as they are the weakest links in the audio chain.

There are two main schools of though when it comes to power amp design. Mullard is the popular design that is pretty friendly with components. The Williamson (what I used as my seed design) is much more demanding on the output transformer selection. My system uses Edcore output transformers which seem to yield excellent results and performance. I got lucky, I think.

You might want to lookup TubeLab at the DIY Audio forum. That guy has a huge amount of experience with tubes amplifier design.

Glad you mentioned that. The first thing I built was the 15-inch active-passive modules. VERY happy with their performance. They compare very favorably to the SDA1C's which Julian Hirsch of Stereo Review said had the lowest bass distortion of any loudspeaker they had ever measured. I used the Polks as a benchmark of sorts as I was tweaking. Anyway, I was in a house with a decent sized living room and no coffee table or anything like that in the way. The floor was carpeted. I tried different midrange drivers, any one of which should have been great below 1 KHz. It just didn't seem to matter what changes I made,the optimum crossover was about 1 KHz, maybe slightly less. I'm still very puzzled by that because it flies in the face of conventional wisdom. It's a plastic cone 15-inch, BTW. I get the idea of room tuning, but haven't studied it per se; I really need to. The real problem regarding directionality is from the planar tweeters. Even tho they distort more than domes, they sound great, but vertical dispersion is lacking. you gotta kind of aim them at your listening level. horizontal dispersion is great. I was contemplating building a Polk-style system with dual small bass drivers and a passive radiator and then planar or dome for the highs, and then high quality passive xover or bi-amp. That would simplify things dramatically. Thanks for your help. I think you are one or two light-years ahead of me in knowledge on this topic.

I have been seriously contemplating going to an audio engineer school. Maybe I'll do that and hold off on any major system changes as there are obviously big gaps in my knowledge base. Plus I lack test gear. What testing equipment do you use and for what purpose?

I have CLIO, but you can cobble something together with individual pieces and something like Holmimpulse or Speaker Workshop for the software.

The advantage of systems like CLIO is the analyzing software does a great job with directionality plots as well as other measurements and the measuremenst are calibrated. The only other system that is in the real of affordability is the LinearX LX-500 with LMS software. That is like $2400, way more than CLIO (mine is an older system).

For speaker and crossover design I use LinearX LEAP5, which is powerful, but not cheap either.

I use the CLIO system to determine if what I have built matches what I designed (verification/validation).

Jerry, lets step back for a minute and answer a few more fundamental questions. Why are tubes even being considered? The point of a speaker/crossover/amp is to ACCURATELY REPRODUCE the audio that is presented to it's input WITH NO DISTORTION. Musicians use tube amps because they like the way they distort the sound when overloaded, but that is on the sound PRODUCTION side of the equation where the musician is not interested in perfect reproduction but instead is interested in how it sounds at the end of the day without really caring how it gets to that endpoint, not the REPRODUCTION side. If you want the signal clean, tubes would be the absolute LAST thing you'd want to use because not only are they hard to come by, waste MASSIVE amounts of energy in the form of heat, and require very large high voltage high current power supplies (which are very prone to generating noise of their own) but they are also microphonic, meaning the vibrations from the sound being produced by the speakers causes the filaments to vibrate and thereby reintroduce the signal back into earlier stages forming a inadvertent feedback loop.

You have been listening to "audiophiles" who don't know their arse from a hole in the ground and who's engineering knowledge wouldn't fill up a thimble. ANY amplifier, no matter what it's contruction, if it's input and output impedances are matched with the input and load respectively and operated within it's save operating range will produce an undistorted signal (as long as no microphonics are involved causing inadvertent feedback loops.). Anyone who claims to be able to do a truly blind comparison test between a solid state and a tube amp operated with matched impedences and be able to tell the difference is lying, not just to you but themselves as well. full stop. if modern test equipment can't measure it, it simply is not there because it can see things that the human ear simply cannot hear.

JFETS are better, but they dissipate a large amount of heat because the on resistance is still quite high because the signal must pass through the junction constriction. Ideally you would use power MOSFETS with very low Rds numbers (typically fractions of an ohm) for your switching elements. they will dissipate virtually no heat which means more power handling capability and smaller heat sinks. a simple L/C filter will generate a 6db rolloff all by itself with no active components. This is typically more than enough for a speaker crossover network. two of them in series will give you 12db. again without any active components.

so again, why are you wanting to do what you are proposing?

http://www.sanderssoundsystems.com/technical-white-papers/172-tubes-vs-transistors

I hear this same debate over and over again and the white paper you linked makes the same fatal claim about distortion being the difference we hear in sound.

While this may be true for an amplifier used for a musical instrument it is hopefully not the case for any amplifier used for audio reproduction! No audiophile worth their salt would operate their equipment at the levels of or beyond the levels of clipping, so the distortion argument is mute.

However, you did hit upon the dirty secret of what we do actually hear between the two systems and that has to do with impedance. You just did not go far enough with the discussion.

The real difference has to do with how solid state and tube amplifiers behave with a changing impedance load, which is reactive because it is an inductive load.

First, a speaker's impedance response is not linear over its operational spectrum. Generally, it has one, two, or more (depending on the type of enclosure employed) high impedance bumps at the bass region, drops to its nominal impedance value in the midrange, then slowly rises in the treble region. All speakers do this to one degree or another, even with Zobel networks.

Now, this is the interesting part. A solid state amplifier is a voltage drive amplifier and a tube amp is a current drive amplifier.

With a voltage drive amplifier, power is inversely proportional to speaker impedance, so power falls off as the speaker impedance rises.

With a current drive amplifier the power is directly proportional to speaker impedance and power increases as the speaker Impedance rises.

This is why tube amplifiers sound richer and have more sparkle than a solid state amp. It is all about the impedance over the frequency spectrum. If a speaker was truly a resistive load of 8Ω across the whole spectrum, you would never hear the difference between the two different amps if they were run below their clipping levels.

In the line stages of the audio path tubes do not contribute any audible color to what we hear, but the output stage does. So, your question to Jerry has merit in that the crossover network does not really benefit from tubes. However...

In my case, the preamp I have only uses tubes and I did it (along with the tube power amp I designed) just because I can. It is a little like driving a Jaguar XKE or a Porsche Boxster. Both perform admirably, but the XKE just has more class.

AH, you are incorrect. BIPOLAR transistors are voltage devices, MOSFETS and JFETS are not.

I will however state once more with feeling, that an amplifier used to reproduce sound cannot add anything to the sound itself or it is no longer reproducing the sound the musician worked hard to record, but is instead producing it's own sound which is not true to the sound originally recorded. Now if you just happen to LIKE listening to distorted sound, great, more power to you, but don't call it "clean" because it is not and can never be "clean" if the amplifier (or speaker, or crossover) is introducing it's own sound into the system.

Calling a tube amp "classy" is really stretching it IMHO. YMMV.

That was what I said. Solid state amps are voltage-drive amplifiers.

I understand that MOSFETs and JFETs are not BiPolar transistors, but you also need to consider the amount of negative feedback in the system.

If the system is constructed as a voltage-drive amplifier (regardless of what components are used) you will get lower amounts of power at the output when impedance rises.

I totally agree that the amp should not add any coloration to the sound. Ideally, it should be a straight wire with gain.

Under proper operational conditions no amplifier should be introducing distortion components into the audio path, regardless of typology (tubes, MOSFET, BiPolar, etc.).

So, once again, distortion is not the reason for selecting tubes over solid state for audio reproduction. That is a myth. For a guitar amplifier that is another matter.

You wrote, "Calling a tube amp "classy" is really stretching it IMHO. YMMV."

I respect your opinion, but the definition of classy is highly subjective, not objective.

I would suspect that if two amps side-by-side provided the same quality of sound reproduction (tubes versus solid state) that most people would find the tube amp more ascetically pleasing for its looks and that when correctly implemented a 100 year old technology can still cut the muster. You personally may feel different and that is your right.

Excuse me, I was incorrect, I meant Bipolars are current devices, and fets are voltage devices. I don't know where my head was when I wrote that.

And the subjectivity was my point BTW. it becomes an aesthetics issue, not a technological one.

I do not think I am confused with the terminology here, but maybe I am.

Regardless of terminology, any amplifier that produces less power at higher output impedances will demonstrate the issue I pointed out, which is lower audio sound levels at those higher impedances.

Compared to an amp that has increased power out with impedance there will will be an audible difference that is proportional to the amount of impedance variation.