The Magico S5 2024 - Double Demanding

Following up on my prior blog post on speakers that are hard to drive, I was getting ready to use the Stereophile measurements of the Magico S5 2024 model as an example of a speaker that was hard to drive, but whose issues could be reasonably passed off as "engineering trade-offs."  Then I read this curious bit in JA's text:  

The minimum EPDR values are 1.03 ohms at 42Hz, 1.81 ohms at 406Hz, and 1.05 ohms at 20.6kHz. 

Wait, that last bit has to be an error, I thought.  Typically only electrostatics show this kind of behavior.  I reread and examined the raw graph, look at the red oval: 

 

 Of course, the circles added are mine, not Stereophile's. This makes very little sense, except we get a clue from his text later on:  

There is a very slight lack of energy above 13kHz before the response starts to rise, peaking at 35kHz (not shown in this graph), which will be due to the tweeter's fundamental dome resonance. 

Suddenly I had a very Sherlock Holmes moment.  The corpse went from "accidental drowning" to "potential murder" in an instant.  This now appears to be a speaker whose crossover may not be optimal in the engineering sense any longer.  

What's Correct

The reason I wanted to use this speaker as an example of "hard to drive" but not deliberately designed to be a low impedance speaker is everything below 6 kHz.  This is classic sealed, 3-way speaker behavior.   Lets walk through it from left to right. 

The resonant impedance peak at 30 Hz is what a sealed speaker should look like.  The yellow circle shows the other impedance minima, at 42 Hz.  We of course wish it was higher, but we understand what's going on.  The designers chose 2x 8 Ohm woofers AND wanted a bit of a bass boost around 60 Hz.  Totally normal behavior.  See JA's quasi-anechoic measurements, below: 

 

The only thing that's a little odd is the low impedance at the crossover point, ~200Hz.  We'd expect it higher, but it's not awful.  The rise in impedance between 200 Hz to 3 kHz  shows filtering/EQ applied to the midrange to match the levels of the woofers and tweeters.  Again, perfectly normal, and not a sign of anything other than excellent driver matching.  Notice also the steep filter behavior in the response plots at 200 Hz.  Pretty close to ideal LR4 behavior.  Bravo. 

The resonance peak at 3-4 kHz is going to be pretty close to where the midrange to tweeter crossover is, and again, absolutely normal.   Any lower and you risk endangering the tweeter.  The horizontal off-axis plots confirm Magico did not set this too high  for the midrange. 

 

Do You Smell Almonds? 

What's wrong with the impedance plot above 3 kHz is the phase angle.  It should be closer to 0 and going positive and end above 0.  This is the typical behavior of a tweeter with a voice coil.  The further up you go the more the impedance is dominated by the inductance.  In the S5 however the tweeter impedance (above crossover) is VERY negative, implying there's a lot of capacitance there. 

While the specific resistance and inductance (Re and Le) of a done tweeter vary a great deal the curve has the same shape and it's just not capacitative.  It's not an electrostatic. 

About the only way to cause this behavior, which lines up with the observed mechanical resonance is if you put in an RC circuit in front of the tweeter.

In the drawing below we emulate a typical tweeter with R1/L1.  3 Ohms and 0.03mH inductor.  We've drawn a parallel circuit of C2/R2 to simulate the RC circuit we think is in the Magico S5: 

 

 

 

As we mentioned, R1/L1 simulate the tweeter impedance above resonance.  C1 and R2 are EQ.  They would dampen the tweeter's output above 10 kHz.  OK, so this is just an EQ circuit, nothing nefarious about this, right? Well, yes and no.  The issue here is the choice of EQ circuit.  By having C1/R2 in parallel with the tweeter we've essentially chosen a very low EPDR.  The designer could have, for instance, put in a series circuit to achieve the same thing and maintain high EDPR.  It's not that the EQ wasn't necessary, it may very well have been.  The issue is the way in which it's implemented in the example above. 

An argument might be made by Magico that hey, we wanted to avoid anything in series with the tweeter.... well, OK, but you have all those other parts in series with the tweeter, midrange and woofers, why go all puritan here?  

Bottom Line

This Magico speaker appears to be unnecessarily difficult to drive in the tweeter range, which may then give it an air of discernment.  Any number of amplifiers are well suited for a 1 ohm load in the bass, but a 1 ohm load in the treble is another thing altogether.  You can forget low feedback amplifiers, and almost any amplifier without several pairs of output transistors per channel.  In the treble range these speakers will be as hard to drive as almost any electrostatic, and if you've ever compared the sound of an average tube amp vs. a decent solid state on a pair of Martin Logan or Sander's speakers you know it's night and day. 

 

On the one hand, it seems like Magico made an unfortunate set of choices in the crossover, though it cannot be proven without looking at the full schematic and individual driver characteristics.  On the other, at $82,000 US or more, it's unlikely any user would be pairing this with anything less than a very big, beefy amplifier. Still, the sleight of hand used to create a speaker which on paper appears to be relatively benign, but upon further inspection is actually quite a beast.  

 

In the modern audio marketplace I don't now how many potential buyers would really care at the end of the day, but for some of us, watching an organization with an enviable reputation take a shortcut like that is depressing.  

 

For me, having seen what Focal did with the Profile 918, and now this I've at least learned a great deal about how a speaker which looks good on paper can be made to behave badly in store.