What is the advantage of ultra bandwidth?

I'm looking into buying a new amp (Spectral DMA 100S), which is ultrawide bandwidth, and I'm wondering, why are wide bandwidth amps made? What are the theoretical advantages of this approach? and what are the possible pitfalls? I've heard the amp with my current speakers and loved the sound. But I don't know much about what wide bandwidth has to offer


To achieve wide bandwidth, the circuitry must be "fast". This means improved transient response, better dynamics, etc... It also means that the signal remains linear well beyond what we can hear, making in-band response much cleaner and purer. Limited bandwidth typically results in phase shifts within the audible frequency range along with high frequency roll off.

A good spec to look at amps with is called "power bandwidth". This is FAR more important than "frequency response". Frequency response is measured at 1 watt of output and power bandwidth is taken at the rated power output. If an amp is truly "fast" and "wide bandwidth", it should be able to do full power WAY, WAY beyond what we can hear. Too bad not that many manufacturers actually print this spec anymore : ) Sean
In general, the wider the linear bandwidth of a system the better. Spectral makes very good equipment, but I would recommend that you start with their preamps and also be sure to use their recommended cables.
Sean, as usual, gives the concise & layman-friendly rendition of the matter. Onhw also gives good advice, I dare say.
I can add that music sounded VERY different to my wife's & my ears when we switched to wide-bandwidth amplification! The differences mainly concerned "background" (for wont of a better word) musical detail. We also felt that there was subjectively more time to listen to (more details of) the musical piece. Objectively, our previous equip is comparable, if not slower, than the wide-bandwidth -- or so the manufacturer claims.
Finally, as these designs amplify beyond 1MHz, ANYTHING in that spectrum will be duly served to your speakers for further... processing: this includes even harmonics (which we like) and trash (which we don't).
The DMA 100S: great product!
(I like Spectral, so my comments on their products are partisan!)
Thanks for the kind words Greg. You also brought out some valid points that i forgot to mention. As such, i hope that you don't mind that i highlight or expound upon them a bit. I've tried to keep this "simple" so most folks can follow along. If i've oversimplified or stated something that needs correction or further clarification, PLEASE do so : )

Since wide bandwidth circuitry can respond faster, the difference between peaks and quiet passages seem even more drastic. If there is no signal present or notes that are spaced apart from each other, the circuitry can drop back down to "zero" and come back up to full output fast enough to show just how "separate" each note really is. This is what separates "average" components from those that have a "really black background".

Having a "blacker background" gives you higher resolution and the ability to more easily listen "into" the recording i.e. able to hear smaller details that are otherwise covered up. They could be covered up either by a higher noise floor or due to signal loss as the slower, "unprepared" circuitry TRIES to respond to each and every dynamic contrast as it occurs.

Some of the losses might be attributed to what is called SID or Slewing Induced Distortion. Slew Rate is what measures how much voltage can be passed in a matter of a phenomenally short period of time. This is expressed in Volts per Microsecond or V/uS. A higher figure is always better ( in my opinion ) with 350V/uS* being "better" or "faster" than 35V/uS. Think of slew rate as being how much amplitude signal the circuitry can develop at the drop of a hat.

Rise time is how fast a circuit can respond from zero to a given level of ( rated ) output. It is also a sign of how fast it can respond to dynamic changes, so a lower figure that takes less time is always better. Like slew rate, it too is also measured in microseconds ( uS ). A figure of .01 uS* is more desirable than a figure of 1 uS given the same level of output.

A shorter rise time also (typically) signifies that there will be less ringing or that the circuitry has "good brakes". This is called "fall time" but most manufacturers don't publish such a spec. Think of rise and fall time as a car that can accelerate like gangbusters AND stop on a dime. After all, such is the nature of music.

If the signal is so "dynamic" that the circuitry can't respond fast enough or in great enough amplitude, the circuitry will:

A) lose some of that signal's resolution B) introduce distortions of its' own to the signal. Obviously, both of these are "bad things". That is why one would want to have BOTH a low rise time ( to respond quickly to the signal ) and a high slew rate ( to produce enough volume to handle all of the signal ).

In order to better understand what i mean, think of the way that a fast, speedy and organized person and a sluggish "i'll do it later" person respond to tasks. The person that rushes to get things done at the last minute overlooks small details ( that can add up to major repurcussions ) simply because they don't have the time that it takes to look at every aspect of what they are doing or involved in. Not only are small details left out, but the quality of the main task also suffers. On the other hand, someone that is speedy and efficient can actually do more work of higher quality without getting "slowed down" by details. In fact, if they are truly fast and efficient and beyond the task presented to them, the small details will become even more apparent without them running over the main theme of the task.

Would you rather have your audio system performing like a well trained professional or a "shlep" ??? Obviously, these are two extremes and most equipment falls somewhere in-between. Where you draw the line should be done with your ears though and not with a sheet of specs. Hope this helps Sean

* These figures were taken from one of my preamps and represent pretty "fast" equipment.
Ummm... I have a different viewpoint.............................

Anyhow, rise time does not mean that the amp in question will have more or less ringing than another amp with slower rise time. It is perfectly possible to have fabulous rise time and be horribly under damped with lots of very fast ringing - in fact I see this all the time on some scary fast opamps, which is one reason that they often sound like dog poop when used in audio gear.

Rise time and slew rate tend to be closely coupled.

Along with that both are then related to bandwidth.

Wide bandwidth amps like the Spectrals can have benefits. However they also can have deficits. One of the biggest deficits is that they will amplify within the AM broadcast band (550kHz - 1700kHz.). This means that there is the *potential* to send significant out of band signal through the amp/preamp combo and have it amplified substantially. In essence this is almost like having a parasitic present in the amplifier - it could potentially effect the way things sound rather dramatically.

Secondly, wide bandwidth amps can become a bit unstable when hit with the "wrong" load and the "wrong" frequency input to excite them. (I'm not saying that this happens with the Spectrals, just that it is a consideration in wide bandwidth designs)

There is some question as to the benefit of a 1 Mhz. bandwidth in terms of the signal within the nominal audio range, especially given the RFI issue. My product, the Symphony No.1 amplifier is spec'd to about 250kHz. at about 75V/us. I think this is more than sufficient, and avoids getting significant gain in the broadcast band.

One of the potential advantages of wide bandwidth is the potential to do "justice" to a square wave. But, this is a false goal, since *nothing* that we have today as a signal source does much better than a ratty square wave! In practice, regardless of the bandwidth I have found that the damping (not damping factor) of the amp design is far more important to the resulting sound than is anything else!

In fact, I can "tailor" the sound of most amps using this criterion alone. Conversely, it is pretty accurate to predict the sound of an amp based upon a look at the square wave response on a bench! (especially into a reactive load)

So, the ultimate determinant of what you hear from a Spectral or other amp PROBABLY has more to do with factors other than bandwidth, frequency response, slew rate or rise time than you might imagine. Indeed, I'm pretty confident that I can mod a Spectral without changing those specs very much at all and give you a *completely* different subjective sound, and probably without any significant topology change!

So, what I'm saying is that a very large component to any given amps sound is the *implementation* details, not the raw specs. (Of course there are some amps that are not going to sound good no matter what you do, but that's a different topic)


Thanks for your responses! I was out of town and just got back. Any thoughts on high frequency interference or distortion distorting audible frequency signals?
Robert, i think that Bear touched base on that subject. Going "mega-bandwidth" CAN open up a whole new can of worms if your not careful.

On the other hand, one of the gentleman that was first responsible for "super-fast, wide-bandwidth" designs ( Dave Spiegel ) stated that his findings show that slower designs of limited bandwidth are more susceptible to RFI based problems. As to who is "right" is up to whom you want to believe.

Other than that, I basically agree with what Bear has to say. I will also add that what one considers to be "wide bandwidth" is obviously a subjective thing. To my way of thinking, anything that can reproduce 100+ KHz while remaining linear is "wide bandwidth". I say this because most components DON'T make it out this far or do it "gracefully". I will also say that i think that active components "should" be able to make it out past 200+ KHz for best results. The slew rate that Bear quotes of 75 V/uS is quite reasonable with most EE's and audio engineers thinking that 50 V/uS is sufficient. My personal "preference" is that it should be as fast as possible.

As to his comments about fast rise / fall times NOT equating to "good performance" due to severe ringing and lack of damping, i don't doubt his findings in the least. As one might assume, there are always "oddballs" that tend to break certain "rules" while basically conforming to others. My comments were only meant to be a "generalization" at best and i hope that they were taken as such.

Speaking of integrated circuits, Stereophile had an article a while back pertaining to the differences in IC's ( Integrated Circuits ) from manufacturer to manufacturer. Even though they were supposed to be built to the same spec and carried the same part number, there were differences of appr 70 dB's from one brand to another in terms of noise floor !!!

Obviously, some parts / components are WAY better regardless of published specs. As such, one could see how substituting just a few raw parts could drastically alter the "sonic flavor" of a component. The fact that i've experienced such changes first hand leads me to agree with such a logical deduction and Bear's statements. After all, if changing passive parts can make such a difference, why shouldn't changing active parts do even more ??? Sean