Joe Rasmussen Usher S520 "Current Compatible" Crossover

Status
Not open for further replies.


NO LAWS OF THE UNIVERSE WAS BREACHED IN THE DEVELOPMENT OF THIS PROJECT.


Joe Rasmussen Usher S520 "Current Compatible" Crossover

Subtitled: Can be used with the JR "Trans-Amp" - 40 Watt Transconductance "Current Amplifier"


Link:
Trans-Amp on DiyAudio

S520-Xover-CC.gif


This has been some time in the making. The Usher S520 has been available for some time and extensively reviewed including by Stereophile and with John Atkinson measurements, which correlates nicely with mine.

Link: Usher Audio Technology S-520 loudspeaker | Stereophile.com

Rightly, they were described as solid engineering at a bargain basement price. So what is the weakness that can be addressed? I was tempted, the price of these little buggers is hard to resist and I have been asked by the Elsinore Project builders for something along these lines, for a small satellite speaker. But in that case it had to be able to fulfill the same trick, that it could be driven by amplifiers with any output impedance in the same that Elsinore MK5 onward (now Mk6) can. This would mean that a number of things would need to come together and happily did. Those who have heard it say they can hear the family resemblance to the Elsinores, so must be satisfied by that.

My little "Trans-Amp" has an output impedance of 270 Ohm. So we are effectively talking about zero "damping factor" and for a speaker to rely on zero "damping factor"

The speaker would of course end up measuring somewhat different and also because the primary goal is not the flattest frequency response that can be achieved, but rather, in order to reach a goal that is primary even over frequency response, some small but acceptable trade-off has been made. The Elsinores does likewise. The primary goal can be seen in the electrical and phase measurements rather than the crossover. The design goals have a different order.

So what is the primary goal - the one that precedes others in importance? Basically it comes down to 'noise' that many conventional crossovers generate to varying different degrees. Now I don't intend to get into say too much except that every designer must have at least adopted some kind of philosophy before he designs anything, and that is particularly true of loudspeakers. Note that Dave Wilson now uses and "anti-jitter crossover strategy" in his up to $200.000 USD loudspeakers. This is basically saying the same thing that we are addressing here - it is 'noise.'

Edgar Kramer interviewed Dave Wilson and Edgar's three part interview is on YouTube.

Link: Edgar interviews Wilson Part 1

Before that interview, Edgar and I had lunch in Haberfield NSW and I was able to inform him that we too were working on this too and that it had to do with suppressing noise in crossovers. As I knew an engineer who has done work on Wilson speakers (to remain nameless), we indeed know how it is done (we had already guessed it and it was confirmed) and I was able to tell Edgar. Indeed the technique does work but is rather wasteful when it comes to drawing power from the amplifier - but as you can see this from the impedance and electrical phase angle measurement done by Atkinson.

An externally hosted image should be here but it was not working when we last tested it.


Note the flattened out peaks and troughs below 100 Hertz and a knowlegable person can figure it out. That compression can also be seen as a flattened, but not flat, electrical phase angle. This is always to be expected.

Now note what our respective measurements looks like for the alternative Usher S520:

S520-Z-1.5uF.gif

[Actual measurement, not the modelling.]

Compare the above and it's fair to say that something more, ahem... sophisticated is going on. The difference is that this approach achieves something that commercial speakers with crossovers can not do: Can be driven by current and hence any amplifier of any output impedance. We will show examples later to prove what the affect is from driving from zero Ohm (which is voltage delivery) and near infinite output impedance (which is current delivery).

Additional Notes:

Some of you may have heard of Esa Merilainen's book on Current Driving Loudspeakers. I have a copy and I recommend it. But Esa and I have possibly a different approach, one that I have discussed with him, about how to get current drive from a voltage source. He certainly recognised that as desirable and made some suggestions and we both agreed that a 5:1 ratio is a good target as a minimum. So 4 Ohm should be driven from 20 OHM min and 8 Ohm should be driven by 40 Ohm min. This can of course be achieved with series resistors, but now it means we can use only a small amount of available power from the amplifier - or get speakers that have very low sensitivity. Yet what are we trying to achieve? Simply this, give the drivers a chance to track "clean current" and avoid contamination of the current which is essentially, as Dave Wilson and increasingly others, have perceived as a higher noise floor (so anti-jitter is really anti-noise).

Dynamic drivers are current devices, motors are current devices, simple as that. Yet we insist on driving them from a voltage source. Then we make it even worse by inserting a crossover? It now becomes clear that the crossover needs to be looked at very closely.

Now I want to make a statement that I believe is an essential issue. The voltage source can produce current - no doubt about that - the problem is that it can also produce current of any phase angle. But a genuine current source can only produce a phase angle of zero since it sees it own output impedance as a series resistor of (ideally) infinite value. The voltage produced across the final load will always track current and voltage together without contamination.

What the above crossover does is comes to terms with those problems, and it produces the same (or extremely similar) results no matter what the source impedance is and also equalises the current. Some interesting things happens as not only will the bass alignment track well down to 30 Hertz, but also the crossover is also immune.

Please keep in mind, we are not really aiming to produce a loudspeaker that can only be current driven, and you can if you build a Trans-Amp of some sort, but the realisation that if it performs the same at both extremes and that current is the ideal way to go, that similar performance will intuitively tell you that we must be getting some of the benefits even when voltage driven.

Now for some pictures:

Frequency Response 15 degrees Off Axis:

Frq15-SE.gif



Crossing Over:


Freq-Xover-SE.gif


The hump circa 1KHz was also indicated in the Atkinson measurements, it seems a reflection due to the mechanical layout of the driver.

The above are modeled.

Actual In-Room RTA Response:

RTA-2CH-at-2M.gif



Additional Stuffing:

Damping-Dim.jpg


Box_1000W.jpg



Remove Crossover & Links and Wire Directly External Crossover:

Rear-Terms.jpg


Xover-Box.jpg



Later:

The above pictures speak largely for themselves and the rest should be easier to work out. I will also prepare some info on the modeling and how the modeling shows how the alignment fares when changing the source impedance of the amplifier. Typically voltage amplifiers have output impedances below 1 Ohm, many tube amps, especially those that eschew feedback, have typical impedances similar to speakers and 4-10 Ohm. Finally, current amplifiers should have an output impedance 5x greater at the very minimum.

Enjoy the fruits of my labour. :)

Cheers, Joe R.


Pair-S520.jpg


 
Last edited:
Part Two:

Two extremes, Red for voltage drive and Green for current driving (270 Ohm source impedance as per 40 Watt Trans-Amp):

S520-270R-Z.gif



Now let's look at the current phase angle, same Red and Green:

S520-0R-270R-E-Phase.gif



Next the Frequency Response Family, 0, 15 and 30 degrees of axis:

Frq-Family.gif



Now comes the really interesting part, how does the crossover perform from a 270 Ohm source impedance (current drive):

S520-270R-XO.gif



Now let us overlay that with crossover driven from zero Ohm (voltage drive):

S520-XO-Overlay.gif


Note that the crossover performs almost as identically as can be.


Finally, we shall look at the minimum phase:

Min-Phase-SE.gif


Please note this is with 4 Watt input and indeed the minimum phase goes well above 10KHz when one Watt.

What the above shows is a remarkable immunity to any source impedance, the largest deviation is at 25 Hertz, low enough for a small speaker system as this (in the Elsinores it even better controlled and below 20 Hertz).


 
What are these things?

That part is a bit clearer. If you want to use a current source amp, the crossovers have to change. And as is the case with crossovers for voltage source amps, they have to restrict out of band signal and control polar pattern to hit the desired design target.

The "noise" thing is mysterious. In the linked video, Wilson prattles about "jitter," but that of course is total FUD marketing to the ignorant.
 
I'm with you on this, Joe. Impedance matters for an amp friendly load.

I don't know if you followed Cousin Billy's interesting series network? Here for a Bud Fried-style 3 way.

515946d1448203830-4-way-series-crossover-accuton-acoustic-elegance-drivers-cousin-billy-series-filter-simple-3-way-fr.png


515945d1448203830-4-way-series-crossover-accuton-acoustic-elegance-drivers-cousin-billy-series-filter-simple-3-way.png


You can do this with a parallel network too. The mid filter becomes your impedance correcting LCR in effect.
http://www.diyaudio.com/forums/multi-way/279278-4-way-series-crossover-accuton-acoustic-elegance-drivers-12.html#post4525459

Two-way is harder, and not quite so flat:
http://www.diyaudio.com/forums/multi-way/279278-4-way-series-crossover-accuton-acoustic-elegance-drivers-15.html#post4533770
 
Ah see what you mean. was confusing impedance with damping and got thrown.

What I'm getting from this write-up is that what's being touted is the unremarkable result that a constant impedance load (wrt frequency) will have a constant voltage across it irrespective of the source impedance. IOW, it follows Ohm and Kirchoff.

There's a marketing disadvantage, though- reduced sensitivity to amplifier output impedance and drive capability. Why is that a disadvantage since it's a technical advantage? Take the Wilson speakers, for example. The impedance curve swings mightily and dips down to brutally low levels. So the reviewer will note that the Wilson is far more revealing of amplifier differences than Joe's speaker!
 
Interesting. It looks like you fixed the very ugly looking response at the crossover point though, making direct comparisons between impedance-flattened and conventional designs hard to really discover.

Based on the Stereophile charts this isn't a speaker I would have bothered with at all, so that's impressive.

Could you post the raw driver responses, and CSD's of the post-modified system if you have them?

Thanks,


Erik
 
Last edited:
Status
Not open for further replies.