Bridging / low-impedance / high-reactance capabilities

[Boomzilla]

I borrowed a pair of Emotiva XPA-1L amps to run some ML speakers. The MLs have a 20KHz impedance of 0.6 ohms with a very high phase angle. After hookup, and with no signal present, one of the two amps apparently went into oscillation and immediately blew its output transistors (before either the fuses or the protection system could prevent it). The amp was defective, and Emotiva has repaired it.

But during the repair, the owner had some discussion with the repair tech. In short, two significant factoids emerged:

1. The XPA-1L is a bridged amplifier
2. The technician specifically warned against hooking the XPA-1Ls to those ML speakers again

The fact that an Emotiva technician recommends NOT using their amplifiers with a widely-known and respected commercial speaker (like the Martin-Logan) seems disturbing. Dan, Lonnie, and Keith have been on the Lounge multiple times defending the Emotiva amplifier designs and saying repeatedly that they're fine with specific speakers (such as Infinities) whose impedance is below the minimal "nominal impedance" given in the specifications.

Is it important to the consumer? Only if you happen to have a speaker that the Emotiva amplifier isn't designed to drive. But if you DO have such a speaker (electrostatic? planar-magnetic? DIY with multiple drivers in parallel?), Emotiva gives no particular warning that their amps may not be suitable.

So which amps are bridged? Which (other than the discontinued XPA-1Ls) are unsuitable for low-impedance, high-reactance speakers?

[garbulky]

To add to that.....what's the deal with the bridging? Is this a standard method of doing differential amplifiers or is it a less desirable method? I am confused....
As for if the amp is suitable they do give warning - with the 4 ohm rating. It is not rated for anything below that - though it will do it to 2 ohms unofficially.

[Boomzilla]

From what I understand, less desirable.

[AudioHTIT]

Certainly a concern, but any specific amp not working with an electrostatic speaker is nothing new. Many stats have sub 1 ohm impedances somewhere in their range, my Beveridge Model 3s hit a half ohm in the high end as well;, tube amps with output transformers have always been one solution.

Nelson Pass always designed his amps to drive electrostatics, in the early days using the Dayton Wright as the acid test. One of his design criteria was to use very large numbers of output devices, the Threshold 4000 which delivered 200 WPC in Class A, had 36 devices per channel; along with a beefy power supply that gave it great peak power as well. The 4000 drove the Beveridges very well.

The information about the 1L bridging is interesting, but what you have identified is a very difficult speaker to drive, maybe the XPA-1 would be a better match.

[Boomzilla]

But is the XPA-1 ALSO a bridged amplifier? The XPR-1 was:


www.audioholics.com/amplifier-reviews/emotiva-xpr1-amplifier
"Unlike their original MPS-1 amplifier that employed a Class H topology where only one high voltage rail was used while the output modulated above the input signal, the XPR-1 employs two rails. The low rail is set for +-36V allowing the amp to operate as pure Class A/B up until about 200 watts or so. Once the power levels go higher, the output rail modulates above the incoming input signal to achieve over 1 kilowatt of power into an 8 ohm load off of +-72V rails. I’m sure some forum sharpie will note that +-72V isn’t a high enough supply rail to reach the 1000 watt power level but they would be mistaken once they considered the fact that the output stage of the XPR-1 amplifier is differential, essentially bridging two channels together. Thus the output stage feeding the loudspeaker sees no ground reference allowing the amp to swing virtually the fully difference between the +- rails (144V – minus losses). This effectively doubles the total voltage swing which has the potential of increasing the power output fourfold in the process. You never really achieve a 4x increase in power for bridged amplifiers, else it would either burn out the output devices or be current-limited by the power supply or wall voltage. In reality, the bridged configuration doubles the power in this case. As you can see, there is plenty of rail voltage to do well over 1 kilowatt of power in this configuration. "

[KeithL]

Just to clarify....

A balanced amplifier IS a bridged amplifier.
The two terms basically mean the same thing..... two identical amplifier channels, each driving one phase of the signal.
(The distinction is that ALL balanced amplifiers run in bridged mode, while some amplifiers can be run as a single bridged amplifier or as two separate channels.)
Whenever you run a pair of amplifier channels in bridged mode, each channel sees an impedance that is 1/2 of the total.
So, if you connect an 8 Ohm speaker to a balanced amplifier, each channel sees a load of 4 Ohms.
And, if you connect an 4 Ohm speaker to a balanced amplifier, each channel sees a load of 2 Ohms.
Therefore, when designing a balanced amplifier, each half is usually optimized to drive a lower load impedance (2 - 4 Ohms, rather than 4 - 8 Ohms).

Internal details aside, all of our amplifiers are in fact specified to work very well with loads between 4 Ohms and 8 Ohms.
In reality, MOST well designed speakers stay somewhere near their stated impedance, and, if they vary widely, vary on the high side.
And, because typical speakers do vary considerably, all of our amps are rated to be absolutely stable into a load as low as 2 Ohms.

If you want to wonder anything, then you should wonder why "a widely-known and respected commercial speaker" would drop to less than 15% of its rated impedance.
The short answer there is that ELECTROSTATIC speakers, like many Martin-Logan models, are widely known for totally unreasonable impedances that drop to very low values at high frequencies.
(Electrostatic panels offer an almost purely capacitive load, which is very difficult to drive, and it requires some tricky engineering to prevent a drop to very low impedance at high frequencies.)

To put it bluntly, all of our amps are rated to drive 4 Ohm and 8 Ohm loads easily, and to tolerate loads as low as 2 Ohms with no problems; they are NOT rated to drive a 0.6 Ohm load.
In reality, because there is normally very little energy at such high frequencies, most of our amps will tolerate even those M-L's.
(However, make no mistake that, their claims notwithstanding, those speakers are well known as being "very difficult to drive".)

But, no, I would probably recommend NOT using any of our fully balanced amps to drive a load that falls below 1 Ohm....
Very low load impedances do present a little more difficult challenge for a fully balanced amp....
So, for speakers with very low impedances, like M-Ls, you're probably better off with any of our amps that aren't fully balanced.

NOTE that only certain ELECTROSTATIC speakers, including some Martin-Logan models, have impedances that drop to ridiculously low values at high frequencies.
(Some other electrostatic speakers take more care to present a more reasonable load impedance curve.)
PLANAR MAGNETIC SPEAKERS, like Magneplanars, are quite different; they usually present a somewhat low but purely resistive load and are quite easy to drive.
(Magneplanars only have a reputation for being difficult to drive because they require lots of power.)

And, yes, if you're planning to make DIY speakers, then you should design them to present a NORMAL load impedance - like 4 Ohms.
(Lower impedances are only appropriate for amps specifically designed for them, like some subwoofer plate amps, and some pro equipment.)

[Wideawake]

A more pertinent question that comes to mind is: "WTF is a speaker manufacturer doing creating speakers that cannot be driven by 99% of the amps out there?"

What is the nominal impedance that these speakers are rated at? The IEC standards for specifying nominal loudspeaker impedance states that minimum impedance should not fall below 80% of nominal. For an 8 ohm speaker this would be 6.4 ohms minimum, for 4 ohms it would be 3.2 ohms and for 2 ohms it would be 1.6 ohms. Not 0.6 ohms!!!

Perhaps ML needs to take their design back to the drawing board.

[novisnick]

@keithl thanks for your input, would you be kind enough to educate me, or point me in the right direction on your statement,
"A balanced amplifier IS a bridged amplifier.The two terms basically mean the same thing"

Thaks for all you do here! Hope you has a great Birthday Too!!

[garbulky]

Another question Keith - since it is bridged - yes the impedance is halved ....but is there any advantage that can come from this design other than the differential advantage with cancelling disadvantage? For instance I know the XPA-2 had 45,000 mf effective capacitance but it used something like 180,000 mf wired differently to provide more power in some other way at 45,000 mf.

[KeithL]

A "regular" amplifier has one amplifier channel which delivers a signal - which is a voltage referenced to ground.

A "balanced" amplifier uses two identical amplifier channels to deliver equal but out-of-phase versions of the same signal.
Taken separately you have two identical but out-of-phase signals relative to ground.
When you connect the load ACROSS the two outputs, you end up with double the signal across the load.
Because the two amplifier channels are identical, but inverted relative to each other, errors and distortions common to both tend to cancel out.
Because you have two amplifiers driving the load, what you have is electrically equivalent to if each amplifier was driving HALF the load (so half the load impedance).

If I were to draw you a very general schematic of "a bridged amplifier" and "a balanced amplifier", they would be the same picture.
You would have two amplifiers, with the load connected BETWEEN their outputs, and equal but out-of-phase inputs driving each one.

The only significant difference is really one of semantics.....
When you design a balanced amplifier, the two channels are "permanently joined"; and, as in some of our designs, there may be additional "cross connections" that help to further reduce distortion.
When you "bridge" two separate amplifier channels, they are connected exactly the same way, except that a little piece of extra circuitry MAY be required to invert the signal going into one.
(f you have a balanced input, then all it takes is reversing two of the wires on one channel; if the input is unbalanced, a simple little circuit is used to create the inverted signal from the original one.)


And... 
Thanks

Oct 18, 2016 11:16:22 GMT -5 novisnick said:

@keithl thanks for your input, would you be kind enough to educate me, or point me in the right direction on your statement,
"A balanced amplifier IS a bridged amplifier.The two terms basically mean the same thing"

Thaks for all you do here! Hope you has a great Birthday Too!!

[Boomzilla]

Thanks, Keith -

I suspected what you said was so prior to reading your comment. It would seem, if I understand you correctly, that ANY truly balanced amplifier would have to be FAR more robust than an unbalanced one to have the same stability at ultra-low impedances. Either the amp can weld, or it can't?

Since there's little information at 20KHz, and since what's there will never be at high power output, does the low impedance at 20KHz really matter? It would seem that most amplifiers would just roll off their frequency response due to the high phase angle, and go to 20% distortion (to pick a number).

Yes, I also agree that Martin-Logan might have given more comprehensive amplifier recommendations (use either a transformer-coupled amplifier or else one guaranteed stable to 2-ohms or below). I think that the transformer coupling makes the most sense (which is why Best-Buy demos their ML speakers with all McIntosh auto-former amplifiers, I suspect).

Although some planar magnetic speakers like Magnepans have benign impedance curves, other planar-magnetics (Infinity IRS, Carvers, Eminent Technology, etc.) can, like the electrostatic speakers, have serious impedance dips I think.

And a final question, please? Would running an otherwise "bridged" amplifier such as the XPA-1 via its RCA (unbalanced) inputs make it more robust into low-impedance loads?

Thanks - Boom

[KeithL]

The only advantage is that distortions common to both channels will be cancelled out (at least to some degree).

For a given output power, a bridged/balanced amplifier will also require a lower operating voltage.
This really doesn't make any difference for amplifiers with transformers and power supplies.
However, for car equipment, which runs on a limited low voltage, it means that you can get more power from a bridged amplifier without bothering to step-up the supply voltage.
(This is also true for those cute little chip-amps that run off a 12V wall wart.)

Capacitors are a more complex question because there is simply no set amount of capacitance that is required for an amplifier.
The amount of capacitance required depends on the characteristics of the individual circuit you're using.
The capacitors in the power supply get "recharged" at each peak of the line voltage waveform.
The capacitors must be able to store enough energy that the amplifier doesn't run out of power "between recharges".
As a generality, too little capacitance will reduce the maximum available power but, once you reach "enough", extra capacitance won't make a significant difference.

You're best NOT trying to compare different amounts of capacitance between different circuits.
The math that describes how capacitors add together, and how much energy they store, is somewhat complex.....
(If I parallel two 50,000 mF@ 50V caps I end up with 100,000 mF@50V; if I series them I get 25,000 mF@100V; both sets can store the same amount of power.)

I've got a 3000 FARAD capacitor on my desk (that's 3,000,000,000 mF).....
A set of a dozen of them would be very useful in an electric car, but it wouldn't be any use at all in an audio amplifier.
(But it sure sounds impressive )

Oct 18, 2016 11:20:00 GMT -5 garbulky said:

Another question Keith - since it is bridged - yes the impedance is halved ....but is there any advantage that can come from this design other than the differential advantage with cancelling disadvantage? For instance I know the XPA-2 had 45,000 mf effective capacitance but it used something like 180,000 mf wired differently to provide more power in some other way at 45,000 mf.

[Boomzilla]

Oct 18, 2016 10:56:45 GMT -5 Wideawake said:

A more pertinent question that comes to mind is: "WTF is a speaker manufacturer doing creating speakers that cannot be driven by 99% of the amps out there?"

What is the nominal impedance that these speakers are rated at? The IEC standards for specifying nominal loudspeaker impedance states that minimum impedance should not fall below 80% of nominal. For an 8 ohm speaker this would be 6.4 ohms minimum, for 4 ohms it would be 3.2 ohms and for 2 ohms it would be 1.6 ohms. Not 0.6 ohms!!!

Perhaps ML needs to take their design back to the drawing board.

Several misconceptions here, I think - First the vast majority of amps on the market (including some AVRs) seem to drive ML speakers just fine. Why? Because the "nominal impedance" across MOST of the audio band is 4 ohms. Only at 20KHz and above does the impedance fall dramatically.

There is NOT much information at 20 KHz

What information IS at 20KHz is at very, very low power

Because the amplifiers don't have to work too hard (if at all) at that high a frequency, they're generally tolerant of the low impedance there.

In other words, a 0.6 ohm impedance in the middle of the audio band (or even worse, in the bass) would be impossible for 99% of the amplifiers on the market. Such a low impedance MIGHT be tolerable if it were a VERY narrow impedance dip, but if it were over a broader range of frequencies, it would make the speakers "amp killers" for sure.

But at 20KHz, the dip is tolerated by most amps without a hiccup.

[KeithL]

Sort of....

It's not so much a matter of being "robust" as being designed with very low impedances in mind.
It wouldn't be difficult to design an amplifier intended to run speakers between 0.5 and 2.0 Ohms instead of ones between 4 Ohms and 8 Ohms; the latter is simply the industry standard.
(Arc welders are designed to weld; amplifiers are designed to play audio into speakers.)

The tricky part is not the low impedance so much as the fact that we're talking about an almost purely capacitive load.
Again, the issue is that this is simply not the normal load that most amplifiers are designed to expect.

To be blunt, it's not a matter of "providing more comprehensive recommendations" - although that would help.
It's a matter of designing their speakers a certain way and totally ignoring the standards that everyone else follows.
(It's sort of like buying a new $200,000 sports car, and then finding out that you can only buy gas that it will actually run on at your local airport.)
I happen to like Martin Logan electrostatic speakers.... but, if you buy them, you have to accept that the number of amplifiers they work well with is somewhat limited.
(And that's even more true for this particular model - which is why they DO tell you than number.)

The problem with transformers and autoformers is that they are big, heavy, expensive, and they distort.
They increase weight and cost and limit performance.

Our amplifiers are perfectly stable with loads down to 2 Ohms - which covers about 99.9% of all speakers......
Planar magnetic speakers have an almost purely resistive "motor assembly".....
Any extreme excursions in impedance they may have are due to odd crossover designs.
Out of the thousands of speakers in the world, I've only ever heard of a half dozen that go below 2 Ohms that aren't electrostatics.
And there are many electrostatics that control their impedance to remain within more normal limits as well.

The XPA-1L (and the XPA-1) are not designed to drive a 0.6 Ohm load
(the input chosen has very little to do with the output load, so I can't imagine it would make any difference).

Oct 18, 2016 14:13:01 GMT -5 Boomzilla said:

Thanks, Keith -

I suspected what you said was so prior to reading your comment. It would seem, if I understand you correctly, that ANY truly balanced amplifier would have to be FAR more robust than an unbalanced one to have the same stability at ultra-low impedances. Either the amp can weld, or it can't?

Since there's little information at 20KHz, and since what's there will never be at high power output, does the low impedance at 20KHz really matter? It would seem that most amplifiers would just roll off their frequency response due to the high phase angle, and go to 20% distortion (to pick a number).

Yes, I also agree that Martin-Logan might have given more comprehensive amplifier recommendations (use either a transformer-coupled amplifier or else one guaranteed stable to 2-ohms or below). I think that the transformer coupling makes the most sense (which is why Best-Buy demos their ML speakers with all McIntosh auto-former amplifiers, I suspect).

Although some planar magnetic speakers like Magnepans have benign impedance curves, other planar-magnetics (Infinity IRS, Carvers, Eminent Technology, etc.) can, like the electrostatic speakers, have serious impedance dips I think.

And a final question, please? Would running an otherwise "bridged" amplifier such as the XPA-1 via its RCA (unbalanced) inputs make it more robust into low-impedance loads?

Thanks - Boom

[Wideawake]

Oct 18, 2016 14:20:06 GMT -5 Boomzilla said:

Oct 18, 2016 10:56:45 GMT -5 Wideawake said:

A more pertinent question that comes to mind is: "WTF is a speaker manufacturer doing creating speakers that cannot be driven by 99% of the amps out there?"

What is the nominal impedance that these speakers are rated at? The IEC standards for specifying nominal loudspeaker impedance states that minimum impedance should not fall below 80% of nominal. For an 8 ohm speaker this would be 6.4 ohms minimum, for 4 ohms it would be 3.2 ohms and for 2 ohms it would be 1.6 ohms. Not 0.6 ohms!!!

Perhaps ML needs to take their design back to the drawing board.

Several misconceptions here, I think - First the vast majority of amps on the market (including some AVRs) seem to drive ML speakers just fine. Why? Because the "nominal impedance" across MOST of the audio band is 4 ohms. Only at 20KHz and above does the impedance fall dramatically.

There is NOT much information at 20 KHz

What information IS at 20KHz is at very, very low power

Because the amplifiers don't have to work too hard (if at all) at that high a frequency, they're generally tolerant of the low impedance there.

In other words, a 0.6 ohm impedance in the middle of the audio band (or even worse, in the bass) would be impossible for 99% of the amplifiers on the market. Such a low impedance MIGHT be tolerable if it were a VERY narrow impedance dip, but if it were over a broader range of frequencies, it would make the speakers "amp killers" for sure.

But at 20KHz, the dip is tolerated by most amps without a hiccup.

I'm not so sure that you have that quite right. I doubt that "the vast majority of amps" and AVRs can drive these speakers and here's my reasoning: The 0.6 ohm dip does not suddenly appear at 20KHz, rather the impedance dip likely follows a slope over a certain range of frequencies. I have not seen the impedance slope of this speaker but it is conceivable that at 10KHz (1 octave below 20K) the impedance is too low for most amps to handle, even as it hurtles down to 0.6.

Also, when you consider high resolution audio recordings the filters are set to cut off audio at about 48KHz (PCM) to 50KHZ (DSD). This means that there may likely be considerable energy at 20KHz, even if that energy is noise or distortion. Additionally, as an owner of these speakers I would not like to take into consideration every source material that I play on them to determine whether or not there is likely to be 20KHz frequencies present. That to me is ridiculous. If I'm buying a high resolution system it is likely that I have high resolution material to play on it.

[jackpine]

When an amplifier goes into oscillation it is in an uncontrolled feedback loop. Some of the output is fed back to the input where it gets amplified. This amplified signal now gets fed back into the input again at a higher level. This will keep happening until the amp or the speaker or both die. Same thing happens when a mic feeds back and a small noise turns into a ear piercing squeal in about 2 seconds. Either the mic preamp or the audio amp must have some sort of protection built to eliminate amp or speaker failure.

[Boomzilla]

Oct 18, 2016 15:40:16 GMT -5 Wideawake said:

I'm not so sure that you have that quite right. I doubt that "the vast majority of amps" and AVRs can drive these speakers and here's my reasoning: The 0.6 ohm dip does not suddenly appear at 20KHz, rather the impedance dip likely follows a slope over a certain range of frequencies. I have not seen the impedance slope of this speaker but it is conceivable that at 10KHz (1 octave below 20K) the impedance is too low for most amps to handle, even as it hurtles down to 0.6.

Also, when you consider high resolution audio recordings the filters are set to cut off audio at about 48KHz (PCM) to 50KHZ (DSD). This means that there may likely be considerable energy at 20KHz, even if that energy is noise or distortion. Additionally, as an owner of these speakers I would not like to take into consideration every source material that I play on them to determine whether or not there is likely to be 20KHz frequencies present. That to me is ridiculous. If I'm buying a high resolution system it is likely that I have high resolution material to play on it.

You may be right. And that may be the reason why ML doesn't publish any impedance curve at all for this speaker.

But I AM right that any transformer / autoformer coupled amp will have less problem driving these than a "direct-drive" amplifier. As the frequency goes up, the transformer impedance does too. So the driver tubes / transistors see the transformer impedance AND the speaker impedance in series. As KeithL says - transformers are expensive and can cause distortion - but in the case of the ML speakers, amplifier transformers' impedance is a blessing.

[garbulky]

This is the impedance for the old version of the Montis (solid line)

www.stereophile.com/content/martinlogan-montis-loudspeaker-measurements#qeSMFofP1AVHzf6W.97
Note that the new version is somewhat different.

[Boomzilla]

Amps that happily drove the ML speakers:

Dynaco ST70 (tubes with output transformers)
Heathkit EA2 (tubes with output transformers)
Crown XLS-1000 ("pro" class-D & rated to 2 ohms)
Cary integrated (massively overbuilt & heavily heat-sinked)

So two "direct-drive" amplifiers DID drive the MLs without problems - but one didn't sound very good and the other costs $6K.

The two "best sounding" of the four were the Cary & the Heathkits.

[Boomzilla]

And an audio amigo is going to loan me his vintage McIntosh MC250 amp to try. Since the outputs are auto-former coupled, and since they have 4 ohm taps, they should be fine with the ML speakers.