Who Makes the Best Sounding Digital Amplifiers?

[AudiocRaver]

I have been looking for a digital amp that can drive my MartinLogan electrostatic speakers. A candidate must be stable into 2 ohms (more or less) with the ability to “double down” and drive greater power into low impedance loads. Full power bandwidth, low noise, low distortion, and low crosstalk at high frequencies, are also requirements, as are low noise and low distortion at low power levels. Crosstalk at frequencies above 1 kHz is rarely even mentioned in spec lists these days.

Many audio companies have embraced switching amp technology in recent years. Commonly found in subwoofers, the implication is that the technology might not yet be suitable for mid- and high-frequency use. Since most tweeters draw little power anyway, a Class A-B or Class B amp is an easy choice there. An electrostatic panel is another animal altogether, though.

Some offerings claim analog-class sound but are ridiculously expensive, and are therefore not on my radar. A few of the moderately priced brands and models have caught my attention. So, the hunt is on sfor a reasonably priced digital amp able to drive a MartinLogan electrostatic speaker to an SPL level of 95 dB RMS with uncompromising two-channel sound quality.

I have done a lot of listening and taken some measurements with the Crown Audio XLS 1502, a model in Crown’s XLS DriveCore™ 2 series, originally chosen largely on specs. I have really enjoyed it, although in comparing it to the NAD T 758 v3 surround receiver, I found the NAD (with analog amp technology) to sound cleaner. The XLS 1502 used in a vertical bi-amp arrangement (one stereo amp driving the woofer and electrostatic drivers via the speaker’s internal crossover) is a configuration that might provide a more perfect result, with more power available to each of the drivers. I have not yet given this configuration a detailed test, although initial trials have been promising.

I have purchased several low-cost chip-based digital amps that claim the ability to drive in the range of up to 25 or 50 or 100 watts cleanly, for various non-critical applications around the house (upstairs TV room, etc.). None of these has received critical listening attention yet.

The Cherry digital amps that Jeff Permanian has used for driving his JTR speakers at recent audio shows are impressive, or should I say the combination of the Cherry amps with the JTR speakers has sounded impressive. I never heard a hint of any “digital amp” sound. The JTR speakers are quite efficient, so the Cherry Amp’s ability to drive a weird impedance like an electrostatic is an unknown. The technology is proprietary. I admit I have yet to dig deeply into the available specs or evaluate in greater detail.

Mark Seaton uses ICEpower technology, from B&O, in his powered speakers, which I have auditioned numerous times and listened to in detail for review. This combination of amp and speaker technologies is simply marvelous. The ICEpower specs are more detailed than can be found for most digital amp technology, or for most amps in general, and address all my spec hot buttons. I am also looking at Hypex.

To date, ICEpower and Hypex look to me like the most interesting digital amp technologies. I hope to acquire a few DIY modules in the near future for further investigation. Then I will have to figure out a housing for them. I am not the metal worker that some audio DIYers are.

QUESTION: What experiences have AV NIRVANA members had with digital amp technology? Do you own a digital amp that you normally use and like the sound of?

 

[Tony V.]

Ive got D3 amps in my new Onkyo receiver and so far have found them to be very clean and seem to reproduce the sound as it should without any "color" Its neither warm or sharp as some manufactures tend to make them sound. When I upgraded from my old Onkyo 805 to this 920 the one noticeable change was that it seems to be very clear sound not bright but clear if that makes sense.

EDIT: I should also add that Monoprice has some rack mountable class D amps available now so they may be worth a look as well.

 

[Matthew J Poes]

While expensive Hypex NCore is about as good as it gets. I’ve heard and own the diy modules but amps have heard and used the NAD versions. Excellent.

 

[Sonnie Parker]

Now don't you know all amps sound the same?

 

[Matthew J Poes]

Now don't you know all amps sound the same?

What! What! Blasphemy! You must be deaf Sonnie.

Hah ok being serious, the real claim being made is that all amps sound the same within their linear range. I actually do agree with this for the most part. The issue is the definition of linear range. Most people argue it’s the point before clipping. That isn’t the definition, and amplifiers distortion profile isn’t flat with frequency. Especially modern digital switching amps. Look at the ICEpower modules and you can see that at high frequencies distortion rises dramatically. There are a number of 500-1000 watt amps which may have just .05% THD at 5 watts but between 10 watts and say 1000 watts the distortion jumps to .1% or even .5%. The graphs provided don’t show the harmonic spectrum so we don’t know how likely that is to be audible either. Given the most likely cause, I believe it is higher order and thus more audible.

There are also other factors such as noise floor, low level hum, channel separation, and switching distortion that are audible non-linearities and can be quite high in some modern amps.

Some examples of things I’ve found in my own DIY amplifiers.

1. A 4 channel module with a poor layout had less than 30dB of channel separation by 5khz. That’s no better than a phone cartridge. The best case scenario was over 80dB so I found that a big problem.
2. An amplifier was picking up 120hz line noise from the power supply due to the amplifier modules grounding scheme and my wiring. It was fixable but a pain.
3. Amplifier modules with high shot noise making the noise floor on sensitive speakers highly audible as a hiss. It could not be fixed. The design put too much dependence on the noise of the resistors and the low noise metal film I used were still audible noisy. The cost of something like an ultra noise Vishay was more than rebuilding the amplifier.
4. Switching amplifiers with audible distortion (heard as an edgy hardness) st high frequencies. These amps were based on Ti chips and the distortion is readily measurable.
5. A kind of pumping that happens with switching amplifiers which don’t use a full bridged configuration. It’s a common issue and happens when high current demands are out on the diodes. Even ICEpower and Hypex modules can suffer this depending on the module and configuration. It’s the audible equivalent of the lights dimming every time the bass hits. Basically the music distortion rises and the overall sound volume quiets when the bass beat hits.
6. Undersized switching supplies causing dynamic limiting. This was technically my fault but I’m still shocked by it. When I studied amp design and was taught to build amps, or even power supplies, the rule of thumb was always to upsize between 20% and 50% of the reasonable limits of your device. This was for linear supplies. Regulated switching supplies could go closer to the margins. I bought an audio grade switching supply from a Spanish OEM that rated my custom module at 1200 watts rms and 1500 watts peak. I hooked up Two amplifier modules that each put out 300 watts rms into 8 ohms and about 550 into 4 ohms. With an 8 ohm load the power supply could be made to shut down and eventually I damaged it. I learned a bunch from that but the simple lesson is that switching supplies need to be overrated by a lot when used with linear amps.

 

[Tony V.]

Oh Oh....we opened the can of worms here LOL

 

[GFOviedo]

I ran out of popcorn!

 

[ddude003]

Let me toss a worm out there... ;^)

Many Class D amps behave as voltage sources, they can double power as the impedance is cut in half. This is not necessarily a good thing for driving Electrostatic Panels... How hard will the amp be driven at less than one ohm as Martin Logan ESLs can do... -Waves hands in the air and mumbles some math- And where are we with distortion and heat out here...

 

[JBrax]

Perfectly content with my Emotiva. Although I haven’t heard many high end amps. Ignorance is bliss I suppose?

 

[AudiocRaver]

Let me toss a worm out there... ;^)

Many Class D amps behave as voltage sources, they can double power as the impedance is cut in half. This is not necessarily a good thing for driving Electrostatic Panels... How hard will the amp be driven at less than one ohm as Martin Logan ESLs can do... -Waves hands in the air and mumbles some math- And where are we with distortion and heat out here...

I guess I am missing what is bad here? Won’t an amp that can drive a low impedance load without limitation be less likely to clip than an amp that can not drive such a load? This “double down” capability, as it is sometimes called, is a good thing for driving ES panels, as I understand it, and is a quality I am looking for.

 

[AudiocRaver]

What! What! Blasphemy! You must be deaf Sonnie.


Hah ok being serious, the real claim being made is that all amps sound the same within their linear range. I actually do agree with this for the most part. The issue is the definition of linear range. Most people argue it’s the point before clipping. That isn’t the definition, and amplifiers distortion profile isn’t flat with frequency. Especially modern digital switching amps. Look at the ICEpower modules and you can see that at high frequencies distortion rises dramatically. There are a number of 500-1000 watt amps which may have just .05% THD at 5 watts but between 10 watts and say 1000 watts the distortion jumps to .1% or even .5%. The graphs provided don’t show the harmonic spectrum so we don’t know how likely that is to be audible either. Given the most likely cause, I believe it is higher order and thus more audible.


There are also other factors such as noise floor, low level hum, channel separation, and switching distortion that are audible non-linearities and can be quite high in some modern amps.


Some examples of things I’ve found in my own DIY amplifiers.

1. A 4 channel module with a poor layout had less than 30dB of channel separation by 5khz. That’s no better than a phone cartridge. The best case scenario was over 80dB so I found that a big problem.
   
   
2. An amplifier was picking up 120hz line noise from the power supply due to the amplifier modules grounding scheme and my wiring. It was fixable but a pain.
   
   
3. Amplifier modules with high shot noise making the noise floor on sensitive speakers highly audible as a hiss. It could not be fixed. The design put too much dependence on the noise of the resistors and the low noise metal film I used were still audible noisy. The cost of something like an ultra noise Vishay was more than rebuilding the amplifier.
   
   
4. Switching amplifiers with audible distortion (heard as an edgy hardness) st high frequencies. These amps were based on Ti chips and the distortion is readily measurable.
   
   
5. A kind of pumping that happens with switching amplifiers which don’t use a full bridged configuration. It’s a common issue and happens when high current demands are out on the diodes. Even ICEpower and Hypex modules can suffer this depending on the module and configuration. It’s the audible equivalent of the lights dimming every time the bass hits. Basically the music distortion rises and the overall sound volume quiets when the bass beat hits.
   
   
6. Undersized switching supplies causing dynamic limiting. This was technically my fault but I’m still shocked by it. When I studied amp design and was taught to build amps, or even power supplies, the rule of thumb was always to upsize between 20% and 50% of the reasonable limits of your device. This was for linear supplies. Regulated switching supplies could go closer to the margins. I bought an audio grade switching supply from a Spanish OEM that rated my custom module at 1200 watts rms and 1500 watts peak. I hooked up Two amplifier modules that each put out 300 watts rms into 8 ohms and about 550 into 4 ohms. With an 8 ohm load the power supply could be made to shut down and eventually I damaged it. I learned a bunch from that but the simple lesson is that switching supplies need to be overrated by a lot when used with linear amps.

As Matt points out, it is complicated. Amp nonlinearity and the onset of distortion come in many flavors and variations. Power supply rejection, common mode rejection, channel separation, various types of noise… We have SO many ways to be annoyed by the sonic imperfections of amplifier circuits, and none of them are simply bad or good, wrong or right, transparent or awful, all show up in shades and degrees, sometimes they grab our attention and sometimes they are subtle and only audible when comparing to another design. And sometimes they are simply not audible AT ALL (EVERY amp has SOME distortion).

Part of what I am led to is that the Hypex and ICEpower modules are thoroughly specified - similar to the kind of detail usually available for integrated circuits (ICs). This gives me at least some HOPE of finding a model with desired characteristics.

 

[GFOviedo]

I’m having flashback when I was taking laws of physics in college.

 

[AudiocRaver]

Sorry, the techie stuff is getting a little thick here.

The goal is the best sounding moderately priced digital amp, Thing is, the right specs can go a long ways toward helping one find it.

 

[ddude003]

A review of Voice Coil Design driven Cone Speakers versus Electrostatic Panel Speakers is interesting to see several "worms" presented to the amplifier(s) powering such speakers…

One such "worm" is the electrical impedance which an ESL present to an amplifier varies with frequency since these speakers are nothing but large capacitors... Unlike a cone speaker, the ESL speaker has the same efficiency from bass frequencies all the way up to 20KHz or wherever it rolls off... An amp which will double power as impedance is cut in half will be making much more power than it should at high frequencies... And high frequencies are _the_ area that ESLs excel in... This is why solid state amps, including class D, often sound brighter and/or more distorted on ESLs than tube amps and/or solid state amps with custom wound transformers or autoformers...

Some other worms to consider are amplifier damping factor and amplifier feedback designs...

 

[ddude003]

Sorry, the techie stuff is getting a little thick here.

The goal is the best sounding moderately priced digital amp, Thing is, the right specs can go a long ways toward helping one find it.

I apologize if this seems like too much tech... Bottom line is, there are reasons for the "synergy" between various amp topologies and the speakers they drive... It has been my experience that Electrostatic Panel Speakers are counterintuitive in these areas... As always, let _your_ ears drive _your_ choices...

 

[AudiocRaver]

I guess the point I am hung up on is that the amp being a voltage source is a given for a solid state amp with a low output impedance and a high damping factor (most solid state amp designs). The amp’s ability to “double down” and drive a capacitive load without current limiting simply assures that it will remain stable and operate in its linear range under those conditions. An amp that can not do this is likely to clip at a lower level than an amp that can do it.

Does this make sense?

 

[AudiocRaver]

A fairly consistent shortcoming of digital amps is that when driving a capacitive load, there is a frequency response peak at high frequencies, in the 10 kHz to 15 khz range. When using Dirac Live, this is compensated for.

 

[ddude003]

I guess the point I am hung up on is that the amp being a voltage source is a given for a solid state amp with a low output impedance and a high damping factor (most solid state amp designs). The amp’s ability to “double down” and drive a capacitive load without current limiting simply assures that it will remain stable and operate in its linear range under those conditions. An amp that can not do this is likely to clip at a lower level than an amp that can do it.

Does this make sense?

I know you really want to talk about the amp... And why not... You are looking for an amp... And the speakers your amp are going to drive is something that need some consideration... A cone/voice coil speaker(s) in a box is not the same as a piece of mylar in the air... Generally speaking -waives hands in the air-, the two designs are almost at opposites of each other... The cone in a box has a voice coil and cone that has a mass that is 5 to 10 times that of a piece of mylar... The cone in a box has a motor/electromotive energy and follows or maps to an impedance curve that has low impedance at low frequency and a high impedance at a high frequency... A piece of mylar in the air has no such thing as a voice coil and little or no electromotive energy. It does not follow or map well to its impedance curve and has a high impedance at low frequency and low impedance at high frequency...

The things you have said you are looking for in an amp are great for a cone/voice coil speaker in a box and that is what many/most amps are designed to drive... Not so swift for a piece of mylar hanging in the air...

I understand what you are saying... Like many, we diverge on this area of amp to speaker interface/synergy... Vive la difference...

 

[AudiocRaver]

Indeed, no argument that cone speakers and electrostatic panels are vastly different animals in the speaker kingdom, and an ES presents unique challenges to an amplifier.

 

[AudiocRaver]

Appreciate the feedback and ideas.

For now I will be studying the literature and specs on the ICEpower and Hypex designs. It will be a month or two before I make a purchase and give a listen. I will report back then.

 

[ddude003]

Appreciate the feedback and ideas.

For now I will be studying the literature and specs on the ICEpower and Hypex designs. It will be a month or two before I make a purchase and give a listen. I will report back then.

Some other interesting reading:
- Missing Link in Speaker Operation Parts 1 & 2 by D. J. Tomcik - Chief Electronics Engineer, Electro-Voice, Inc.
- Audio, Distortion and Feedback - Nelson Pass 11/1/08
- Paradigms in Amplifier Design - Ralph Karsten - Atma-Sphere Music Systems,Inc.
- Output Transformers In Transister Amplifiers - Sidney Corderman - VP R&D Mcintosh Laboratory Inc.
- An Overview of Electrostatic Speakers - Wes Hatch

It appears there is much room for innovation and improvement in D class amplifiers and the matching of the Amplifier to Speaker Interface...

Good luck in your journey...

 

[AudiocRaver]

Looks like a great reading list. Thanks!

 

[Matthew J Poes]

A fairly consistent shortcoming of digital amps is that when driving a capacitive load, there is a frequency response peak at high frequencies, in the 10 kHz to 15 khz range. When using Dirac Live, this is compensated for.

Thankfully the placement of the feedback circuit in the Hypex modules ensures that isn’t true of those amps. They have load independent frequency response.

ICEpower are pretty good too, response variation with load is minimal. Their biggest issue is the rise in distortion at high frequencies. It’s variable with load as well. This is why I like Hypex modules better.

 

[Matthew J Poes]

Is the doubling of power really so important?

I know it was often a sign of a good amplifier, but most modern amplifiers using class-D topology and SMPS can’t and don’t double in output with half the load. The supplies are highly regulated with a hard current limit. However the other side is that they are still unconditionally stable into low impedance loads and capable of far more output than in the past. Hypex can be made to be stable down to 2 ohms and they have modules up to 2000 watts of output.

 

[JStewart]

Did you look into NAD C268? Can be bridged for 300W (although website doesn't state still 2ohm stable in bridged mode). Seems reasonably priced as amps go. Assuming its the same amp section as the C368 that Matthew Poes reviewed earlier this year as NAD states the C368 can be bridged and used with the C268 bridged for a 2 channel set up.