High end Class D amps?

Based on the photos of the inside of the Rogue “tubeD” amp, the amp uses a standard class D module with a proprietary tube input stage and a linear power supply. 

George - you obviously aren't interested in class D, so why do you keep bothering to post on this and other class D threads?

The D in class D has nothing to do with digital. This amplifier topology just happened to be the fourth (after A, B, and C) to be described and was assigned the letter D. 

It does so happen that class D switches the output transistors, but then so does class E. 

It would be nice if this thread focused more on what differentiates a great sounding class D amp from a mediocre one, instead of the unsubstantiated reasons why class D can't possibly be considered high end. 

My personal experience with class D is limited. I had a pair of Nord NC500 monoblocks for a little while. There was a lot to like about them - the were very detailed, quiet, and powerful, as well as being small, light and cool - but I just didn't feel the emotional connection to the music that I did with the big Krell amp they replaced or the Parasound JC5 that replaced them. 

However, I don't believe my experience with the amps is an inherit limitation of class D. I just don't think these particular amps were a good match for the system I had at the time. 

I think class D has a lot of potential and I'm very interested to understand what makes some class D amps sound better than others.

This argument seems kind of silly. As with class A and class A/B, a class D amp's output power at any given load impedance is limited by its voltage and current capabilities. 

In class A and class A/B amps, the output transistors are dissipating power proportional to the bias current and voltage across them. Most class A amps (although not all) are set up with enough bias to operate in class A only into 8 ohms, switching to class A/B into lower impedance loads. With adequate heat sinking (and a big enough power supply), their thermal capabilities can generally support enough current to double power into 4 ohms and again into 2 ohms (and some even again into 1 ohm), since the voltage across the output transistors carrying the current is lower when they are delivering the maximum current. 

This is where EPDR comes into play, because if the phase angle requires high current while the output voltage is still low, the thermal limits of the amp can be quickly exceeded. 

Class D amps operate differently, since the transistors are almost always either in saturation (where maximum current is flowing, but with very little voltage across them) or turned off (where no current is flowing). The only time they are dissipating significant power is when they are switching from on to off or back (this is obviously a simplification). but this is a small percentage of the time. This is how class D amps can be so efficient. 

But there is nothing about class D amps that would prevent you from building an amp that doubled power as you halved impedance. It's just that there is not the same benefits to do this as their might be in a class A or A/B amp. The amp is not going to run much more efficiently at half the rail voltage like a linear amplifier will. 

Since most speakers are nominally 8 ohms (or slightly less), most class D amplifier designs are optimized for this load.  But that doesn't mean you couldn't build amp to do what you're describing. The Hypex and Purifi modules, for example, can be operated at a lower voltage with the same high-current power supplies to create an amp that can easily achieve what you are talking about.

If you ran a Purifi module at around 35v, it would be limited to about 30w into 8 ohms, 60w into 4 ohms, 120w into 2 ohms, and could probably achieve 240w into 1 ohm. But why would you want to do this since you could achieve almost the same power into 1ohm and get over 200w into 8 ohms without generating much heat by using a higher voltage supply.

10K class d is paying for a name, and aesthetics.

 800$ class D amp is smart buy.

 Don’t let the sheep make a line you can’t help but follow.
 Happy hunting.

As with any audio component these days, it's possible to get pretty good quality for a modest sum (say, under $1K for most component types). Clear improvements beyond this quality require considerably higher investment. 

Consider a simple component like a film cap. A cheap no-name 1uF cap will cost about $0.30 qty 1 at retail. Stepping up to a better quality WIMA will increase the price to $0.50. Moving up to a SoniCap will increase the price to close to $10. A Miflex KPCU copper foil cap will increase the price to close to $75. 

Multiply this by the hundred+ components in a given product and you can see how the build cost for high end products can quickly escalate. 

Whether these "upgrades" result in sound quality improvements that are worth the incremental cost is up to the customer, but since there is a fairly robust market in high-priced gear, many people believe it is. 

But saying that you are just paying for name and aesthetics is just not true. Class D amplifiers can benefit from judicious use of higher quality components and more sophisticated design approaches as much as any other product.

@twoleftears  - I am thinking about trying an input buffer using tubes.

I don't think I would consider using the Korg NuTube like ATM does. This is a pretty high-distortion device - specifically designed to add distortion for synthesizer special effects. It also has a bad reputation for microphonics, and is hard to swap in and out. 

The 6DJ8/6922 tube that VTV uses is much better, but I'm not sure it's the ideal tube to use with the relatively low voltages I have available (approx +/-65v if I use the main supply for the Purifi module). I don't want to add another high voltage supply to drive a tube section, although I could consider a DC-DC converter as part of the buffer board design if I have to.

At any rate, I'd prefer to avoid the hassles of tubes, so I'm going to see what's possible with various solid state approaches first.

I think I stated earlier in this thread that I used to own a pair of Nord Hypex NC500 monoblocks. There was a lot to like about them, and I think they delivered compelling performance for the price, but were not as engaging as I was looking for in the system I had at the time. But they convinced me that class D holds a lot of promise. 

In the past year, I've had a lot of fun building DIY electronics (two pairs of monoblock amps and a preamp so far), so I've decided to try building a class D amp as my next project. I like monoblock amps for a number of reasons, so this project will be another pair of amps. I'm using the Purifi modules, and using a high-current linear power supply (1500VA) for each amp, and leaving room in the chassis to experiment with different input buffers. 

This probably wouldn't be cost effective for a commercial product, but that's what's fun about DIY. 

I'm going to use the Neurochrome buffer to start, but I've already started designing a buffer with a single-ended class A output that I'll try for round two. I've got a few other ideas in the works for other input buffers as well, so I can hear for myself how they affect the sound. 

I hope to be playing music with the Neurochrome buffer by the end of the month. 

From an article on Audiophile Style - interview with Bruno Putzeys. 

@Sagittarius: Class D has achieved very low levels of distortion, but is it possible for class D amplifiers to continue their evolution into something close to a straight wire with gain, i.e. minimal phase shift in the audio band? (A similar question from maty).

 

Bruno: The 1ET400 module has the frequency and phase response of a 2nd order Butterworth filter cornering at 60kHz. If you look at the phase shift of that, it’s very nearly “linear phase” in the audio band. To take some rough numbers, it if you have a circuit that has a 0.2 degree phase shift at 200Hz, 2 degrees at 2kHz and 20 degrees at 20kHz, that’s the same as saying it has “0.001 degree per Hertz” phase shift. That’s another way of saying that the whole signal is simply delayed by 2.8 microseconds. If you plot phase shift on a linear frequency scale that’s immediately obvious because you get a straight line. Of course a simple delay doesn’t change the sound. It’s literally the same as starting your music a few microseconds later.

 

Lars: My dad used to say that if you left a CD in its case without playing it back, it’d just sit there accumulating massive amounts of phase shift as time went by.

 

Bruno: What that matters to sound is how much phase shift differs from a pure delay. Anyone who’s ever done phase measurements on speakers will remember that you have to remove the time-of-flight delay from the data, for instance by marking the leading edge of the impulse response. Otherwise the linear phase shift corresponding to the distance between the speaker and the mic completely clouds the picture. In the case of the 1ET400 module it’s just under 1 degree at 20kHz. There never was a phase shift problem in class D, it’s simply a trick of the light that happens when you plot the phase response on a log scale without removing the fixed delay.

I believe Bruno in that modern class D amps with the output filter in the feedback loop do not have an issue with phase shift in the audio band.  I actually thought the high frequencies on the Nord NC500 amps sounded very good (although being in my early 60s, high frequencies to me are not the same as they were 40 years ago). 

My issue with the Nords on the system I had then was that they lacked body in the upper bass/lower midrange. I tried both the Sonic Imagery and Sparkos discrete op amps in the buffer stage. The Sparkos had a bit more warmth and was closer to the sound profile I was looking for, but gave up too much detail. 

The Bel Canto Ref600M is pretty similar to the Nord in it's design approach, using a SMPS power supply and a custom input buffer. There aren't too many class D amps built with linear power supplies. The LKV Veros (mentioned earlier in this tread) is one of the few that does and has received good reviews, so I'm going to give it a try (partly because I'm comfortable building a linear power supply myself, and I don't want to just drop in a pre-built SMPS supply).

@atmasphere -nice explanation. Thanks.

@twoleftears - I actually think that 1KVA toriodal is undersized for a hi-end stereo Purifi amp. I've generally found that using a larger power supply improves the dynamics and authority in the mid-bass and results in a more pleasing sound. I don't know how much this translates to class D designs, but I would expect the same applies. 

The Purifi modules' over-current protection doesn't kick in until 25A, so it seems like there's value in using a beefier supply. The DIY Purifi amps I'm building will use a 1.5KVA transformer for each monoblock, although a small portion of this is allocated to the lower voltage supplies.

@ricevs - Thanks for the offer, but I think there is lots of low hanging fruit to focus on before attempting to modify the Purifi module. At some point, I might consider it. 

I think the Weiss op amps could be worth a try. From what I've read, these are the best discrete op amps available and have the ability to bias the output stage fairly high (adjustable with a small trim pot). I plan to use these on one of the input board designs I will build for my DIY amps. 

My experience with the Sparkos op amps is that they dulled the sound a bit to much, although did add a bit of welcome body (this was on a hypex amp). The Sonic imagery op amps were very clean sounding but sounded a bit thin and sterile in my system. 

George - You've brought up this notion of EPDR in several class D threads, but haven't explained how it relates to class D. This is a term that was made up by a Stereophile reviewer to try to explain why linear amplifiers had a hard time driving speakers with a reactive load. Specifically, speakers whose current demands are not linearly related to the voltage across the speaker terminals. 

In a linear amplifier, this will cause a higher than expected thermal load on the output transistors because the high current demand can exist even with a high percentage of the rail voltage across the output devices. 

In a class D amplifier, the output transistors don't care about the voltage across the speaker since the voltage output only affects the duty cycle, not the power dissipation. They are constantly switching on and off at their oscillation frequency regardless of the voltage across the speaker. As long as the amp can handle the current demands, the fact that the load is at all reactive is irrelevant. 

If you disagree with these statements, please explain why. (And don't just repeat your previous statements since so far you haven't described how EPDR relates to class D). 

Obviously, you can contrive a situation where a given class D amp does not have the current capability to drive a particular load at the desired voltage, but this true with a class A or A/B amp as well. And, in fact, the thermal challenges are much more difficult with a class A amp, and even a class A/B amp, to meet a specific desired current capability.

There are class A/B amps that have been designed to output very high current levels at the expense of cost and weight, and thus far, class D designers have chosen to focus more on the mainstream market where current demands are not as extreme. But this has nothing to do with EPDR.

I'll grant you that there may be limitations in the number of output devices that can be paralleled and still maintain appropriate switching speeds. With current output devices, this may limit the maximum current that class D amplifiers can deliver to less than is possible with a class A/B design. I have not tried to design a class D amplifier myself (other than using an off-the-shelf module), so I don't fully appreciate the limitations, But I think the potential is still quite high even with current devices. 

For most reasonable speaker loads, the current demands are well within the capabilities of mainstream class D amplifiers, such as those built using the Purifi, Hypex, and ICEpower modules. The Purifi module, for example, can output 25 amps before the current limiter will kick in. Obviously, this still requires that you have a power supply that can deliver this level of current, as it does with a class A/B amp. 

@noble100 - I'm wondering if part of your improved satisfaction having inserted the ML preamp in the chain is a lower impedance, stronger drive capability of the preamp compared to the Oppo. 

I've found that even a very neutral preamp with a fairly high current, low impedance output stage can sound much smoother and more natural, while still sounding very detailed, than when it is out of the system. 

I don't have access to the documentation for the Hypex modules used on these amps, but the connector is the same (except for one status line) as the Purifi module. There is no provision on the Purifi connector to bypass the driver circuitry. I suppose it could be possible to have jumpers on the card which would do this, but it seems unlikely since this circuitry (on the Hypex board) is an integral part of the feedback design that makes the Hypex module function. 

The Hypex and Purifi modules do expose the outer-most feedback loop, which is designed to provide a feedback connection from the speaker connections (or as close to the speaker connections as is practical). This is generally connected to the speaker output connections on the buffer board. It could be possible that Rogue is adding circuitry to this path to alter the Hypex module's transfer function, or someone integrating their front end buffer inside this feedback loop. 

@djones51 - if your definition of an ideal amp is a perfect voltage source, meaning an amp with zero Ohm output impedance, then a class D amp gets closer to this ideal than any other amplifier topology. Class D amps have much lower output impedance than class A or A/B amps. 

You make it sound like the Wilson Alexia is the most popular speaker on the market. How many of these has Wilson sold?

 I’ve owned a lot of speakers over the years and none had an impedance curve that dropped below 2 ohms. If you own Wilson Alexias then maybe the class D amps that have been discussed aren’t the best choice. But if I owned a speaker this expensive, I probably wouldn’t be driving it with an amp that costs well under $10k. 

Looks like this one to me. 

https://www.pascal-audio.com/product/s-pro2/

I wish it was a simple as determining a single test parameter like SINAD to determine what's going to sound good and what's going to sound mediocre. But then this hobby would be rather boring.