[KevinW]

Someone sent this question in via email, and I thought it would be smart to answer it publicly (with his permission of course).  It will eventually find its way into my FAQ.  Any comments on whether this makes sense, confuses or confounds you would be appreciated.

 Kevin I been following the post of your amp with great interest.
I am not very knowledgeable about audio. So I wonder how is it possible for your amp which has an A/D converter in it to sound analog(better than your SET amp), as I am used to thinking that digital sounds inferior to analog...  (This is a sincere questions  )

This is a good question. The answer is that digital amps and digital sources share little in common. Yes, they encode waveforms in a digital format, but that's about it.  A digital source uses a low sampling frequency (44.1kHz) compared to digital amps, which run around 500 kHz.  So Class D amps have a much higher resolution representation of an analog signal.  SACD samples at 2.8 Mhz, and this is one of the primary reasons why SACD sounds so much more "analog".  The LC Audio modules I am using can be clocked up to 1 MHz.

Class D amps and digital sources have even more differences... A big one is that the Redbook digital format is designed for transmitting signals over distance (e.g. a digital cable).  As a consequence, there is error correction information in the signal, and the signal is processed in a way to make it more fault tolerant.  Decoding this requires a very sophisticated device, and it requires very precise clock matching.  Errors in the clock (e.g. jitter) cause huge problems in sound quality, and this is why installing an ultra-low jitter clock is one of the best tweaks you can do to a CD player.

Class D amps do something entirely different with the digital signal.  They amplify it in place, and then simply pass the signal through a lowpass filter.  This filter acts as the "decoder" and usually consists of one inductor and one capacitor. There's no clock involved there!  So Class D amps are essentially immune to jitter.

Regarding sound quality, Class D amplification takes the best characteristics of both  SS amps and Tube amps, while leaving out the worst problems.  The key concepts here are harmonics and linearity.  

Harmonics are the spice of life as far as music is concerned.  The complex interplay of harmonic frequencies and levels is what makes music sound so beautiful (or awful). The ear is much better at understand harmonic information than any instrument, or any spec sheet which horribly condenses all harmonics information down to the total average harmonic distortion.  This is probably the second most misleading spec in audio (after total RMS power ratings).

Tubes have a strength in harmonics, because the harmonic distortion waveforms from a single note tend to die out quickly.  Solid State has a big weakness in that the harmonics from a note tend to "ring" and persist much longer after the note has passed. The amplitude of these rings are very very small, but they still can intermodulate with successive notes and cause a harsh, piercing anharmonic sound. This is why no-negative feedback amps are preferable, because feedback increases the "intermodulation distortion".  Tubes also avoid discordant and unnatural odd-order harmonics that are inherent in SS devices. So even though the Total Harmonic Distortion of SS is much lower than tubes, it's not the total amount that matters, but rather what kind and when.

Single Ended Triodes, in particular, have extreme amounts of 2nd order harmonic distortion, up to 5%, but yet people absolutely love the sound.  Why??  Well, the answer is that 2nd order harmonics are especially pleasing, almost like seasoning on food.  If food is bland, add spice!  The right amount of seasoning can bring the taste to a zenith, but too much can obscure some of the inner detail of the taste. Everyone has different opinions on what tastes or sounds best to them.

Linearity is the big problem that analog amplification devices have.  A 20 Hz signal should have exactly the same amount of amplification as a 20,000 Hz signal, and everything in between should be the same too.  This is not as easy as it sounds.  Tubes are notoriously non-linear, especially the classic SET tubes like the 300B and 2A3, and a really good designer knows how to build circuits that utilize only the linear operating points of tubes.  SS transistors are usually a lot more linear than tubes, but finding high power varieties is not that easy. It is in fact entirely possible that they are no longer commercially produced at all in the near future, because of the prevalence of digital.  To sum up, SS devices are typically more linear across the entire frequency band.

Class D Advantages
Since Class D amplifies at only one frequency (the carrier freq ~500kHz), it has NO problems with linearity!  Also, harmonics of the carrier frequency (mulitples of 500kHz) are so far out of the audio band that there is no effect whatsoever on the music. A Class D amplifier is capable of reproducing music without any harmonic intermodulation distortion, and zero non-linearity problems.  This is why it can sound so good.  The resulting accuracy, detail, imaging, and emotional involvement is stunning.

The problem with Class D is that sometimes it can be TOO accurate, and sound dry or analytical.  We figured out a trick that completely solved this problem for the LC Audio modules, and now there is no discordant information in the music.  It is only the purest representation of the original analog signal.

Energy Usage of Class D
One more huge advantage of Class D over analog is the method in which an amplified waveform is created.  With Class D, the output transistors are switch from full-off to full-on to essentially "reassemble" the analog signal in a high-voltage state.  Transistors (and tubes) "like"[/] that.  In contrast, analog amplifiers have to be on all the time, and perfectly respond to the subtle variations in level and frequency of the musical signal.  This uses a lot more energy, and puts more stress on the amplifier. In contrast, Class D amps are incredibly efficient, and should last a long time because of the low stress on the components.

[JohnR]

The problem with Class D is that sometimes it can be TOO accurate, and sound dry or analytical. We figured out a trick that completely solved this problem for the LC Audio modules, and now there is no discordant information in the music. It is only the purest representation of the original analog signal.

Um, so this trick adds spice, or does it remove non-linearities that aren't there in the first place?

[KevinW]

Well yes and no to both...  Class D amps do have non-linearities of a different sort, even though the audio frequency band is ruler flat.  You can see this in the cookbook published on the LC Audio site.  We figured out a way to reduce this non-linearity one step further, and got amazing results.  The net result was adding spice by eliminating something that was masking the emotional beauty of the pure sound.

[JohnR]

Um, the point is that your spiel really doesn't make a lot of sense. (Or your response either, for that matter...)

[KevinW]

Um, the point is that your spiel really doesn't make a lot of sense. (Or your response either, for that matter...)

Unfortunately for those who are interested in the more analytical aspects of what I'm doing, obscurity is precisely my point.  If I said too much about what that circuit does, then it could be replicated by the competition. In the interests of actually making money, I need to keep this little bit of information secret and I'm going to be cagey when discussing it.  I'm quite happy to discuss most every other aspect about this amp, and if DIY'ers want advice on building their own, I'll give them plenty.  I may even sell finished and disguised copies of the secret circuit.

I've already given more clues about what this circuit does that I should have.  I think given JohnR's established technical ability, he might be able to make an educated guess as to what it does.  

Folks, this circuit does work, and I am still quite amazed at how unbelievably good the amp now sounds.  I've tried to explain its function without giving the secret away, and there are some clues in there. If you don't believe it or understand it, just sit back and wait for the finished product to knock your socks off.

[KevinW]

Another question from a PM:

I guess you mean that you are NOT coloring the signal, but are eliminating somekind of harmonic distortion?  

Quite correct.

If the design now is quite pure, then why are you designing a tube input stage for this amp? Wouldn't it color the signal?  

"Coloring" is not the term I would choose, as it implies a negative effect. True, the added harmonic distortion from SETs and tubes is information that was not part of the original signal, but as I tried to describe above, this is GOOD, not BAD.  Or else, SET's with 5% THD would be unlistenable.  But we all know how revered low powered SETs are by those who have used them, myself included. They are not for everyone because of the inherent problems: low power, loss of linearity in the frequency extremes (sub 25Hz bass and 18+kHz treble), and some simply may not like the extra "spice".  For those last few, then perhaps coloration is a proper term.

The goal of putting in the added tube stage is to add in a little bit of the tube magic, while not losing any of the linearity.  This will make the emotional impact of the amp even greater than its current form, which I cannot wait to hear.  

[JohnR]

Well, you did ask people to tell you whether your spiel makes sense or not...

I'd suggest, if you intend to make some of this into a FAQ, split it into two: why is a digital amp different than "bad digital"; and why does your amp sound better than other class D amps. In the latter I'd leave out the  techno-jumbo about ringing waveforms and so on...

[eico1]

...Redbook digital format is designed for transmitting signals over distance (e.g. a digital cable).

redbook definition relates to the data on a cd. The digital data sent over the cable is already decoded from redbook specs and encoded to a bi-phase type signal with no error-control.

Class D amps do something entirely different with the digital signal. ...There's no clock involved there! So Class D amps are essentially immune to jitter.

I'm sure the circuit is sensitive to the oscillator or "clock" performance, especially  PWM system. How sensitive I'm not sure, but immune might be stretching it. Certainly more than a linear amp:)

steve

[KevinW]

Thanks for the clarification on CD format, which I needed.  Regarding clocks on the PWM modules, it makes some difference but not that much.  We tested the performance of various clock speeds this weekend, and were surprised that it didn't make as much of a difference as we had guessed. I think one reason why is that a few clock cycles off at 500 kHz is such a small error that it isn't noticeable to the ears.

[Kevin P]

Oh come on you guys... give Kevin a break!  

This is his little corner of AudioCircle and he should be able to engage in a little spiel without getting hassled over it.   After all he is using the world's best digital amplifier module that has no peer!    

If you want to hassle Kevin over the amplifier then come to VSAC and do it.  We will have his digital amplifier in our room.   The only reason we can use it is because Kevin's version will have tubes... and Doc B  has said, that as long as it has a tube in it we can play it in our room.  

We will also have several VERY good SET amplifiers so you can compare them and come to your own conclusions.

[JohnR]

Any comments on whether this makes sense, confuses or confounds you would be appreciated.

???????

[KevinW]

Any comments on whether this makes sense, confuses or confounds you would be appreciated.

???????

Heh, perhaps I should have worded it:

"Any comments on whether this makes sense, confuses or confounds you would be appreciated, except for the portions that are intentionally misleading and/or confusing."

Sheesh!  Leave it to the engineers to nitpick the sticky details.      

[JohnR]

Hey, in your case I will believe the amp sounds the way you say it does... even if I think your spiel sucks

Also, kudos for offering to make the technology available in some form to DIYers, it's on my list ...  

[Sa-dono]

Hey Kevin! Hope you had a great vacation when you get around to reading this. I was wondering if you have any comment about this spiel on digital amps I read in a thread, regarding the Carver and Tripath technology, and if there is truth, what you are doing to account for this in your amp? Thanks for any info you can provide!

One of the problems facing a designer building a digital amplifier is to recognise that what he is actually constructing is a very high powered RF amplifier.This demands a power supply that has very low impedenceat RF frequencies. In the case of the Tripath amp the upper limit is 1.5Mhz, in the case of the XICON Digital Solution 2Mhz. Needless to say most designers either miss mark by a large margin or are prevented from realizing it in the production piece because of costs associated with such a supply.There is currently only one manufacturer that Iam aware of.making large value,high voltage caps with sufficiently low impedence at 1Mhz to be used successfully in a digital amps power supply and that is Jensen in Denmark.In the case of my Tripath based DIY the math says that as much as 100amps maybe being switched at 1MHZ. The power supply wiring has to kept short and local decoulping caps have to be used to offset inductance in the wiring and suppress ringing. This is a round about way of saying that we are not hearing what digital amplifiers can really do in most cases because of design compromises in the power supply.In the case of the Carver amp Jensens won't fit into the case and BlackGates cannot be used everywhere they should be either. It is a tribute to how well the design can do with one hand tied behind its back judging from the satisfaction expressed in Dmason's posts

[Kevin P]

Since Kevin is out enjoying the great outdoors I'll try and give you a broad answer... not based upon his product but based upon the LC Audio Modules that he is using.

The LC Audio Modules used have BlackGate VK Series caps designed into the PCB to be as physically close as possible to keep stray inductance low.  Also there are two arranged as a L-Canceling pair to lower DCR at the higher frequencies.   I'll quote Lars from his web site.  

ZAPpulse is based on an integrating modulator, similar to that of any other brand of PWM or ´Digital´ amplifier. Some makes of this type of amplifier will tell you that they are NOT using PWM technology, and they have invented a completely new special way of building a switching amplifier. This is an old marketing trick, as have been seen in the audio business for decades. We believe the new order of audio enthusiats demand a true and honest information about any new product, as the market holds more and more new options to choose from. We have analysed several different of our competitors technology, and they work surprisingly alike, even if there is obviously small deviations in the actual circuits.

So what differentiates our product from the competitors? We use Audio Grade parts to build our amps. We use ultra fast audio grade opamps in the integrator, we use high grade polypropylene capacitors to power rail filtering and output filtering. We use a low distorsion ferrite coil with about 5 times higher saturation point than our competitors.
Also we offer a non compromise version of our amplifier with Black Gate VK capacitors for rail filtering and sound scaping from lower midrange up, in the absolute reference class!

The result is a super high efficient amplifier, that even in the standard implementation surpasses the sound quality of most conventional amplifiers. The sound is soft, and with a touch of tube like warmt in the voice range. One distinct property of the ZAPpulse is the capability to expand the sound stage. When you increase the power level, most conventional amplifiers tend to narrow the soundstage, and get a more agressive midrange, and less resolution in the mid and top ranges.
A ZAPpulse (like most other PWM amplifiers) have the ability to maintain the same soundstage, dynamic and midrange resolution at both high and low power levels. This is what hifi reviewers label the amplifiers capabilities of soundstage expansion. On this fiels ZAPpulse is extremely capable, and always makes a highly dynamic and precise soundstage, never with agressive voice reproduction.

Mainly where Kevin is tweaking the amp is in the external power supply, switching frequency used and the preamp/buffer stage.   I'm not sure if he is modifying the LC Audio Modules (he can answer that) but you can see that we use much more expensive parts throughout the board than the Tripath or XICON production units.  LC Audio is small enough and the customer base is much more reflective of high-end audiophile needs.  Tripath and XICON are trying to produce something for the mass market.    Nothing wrong with that but they have to be much more careful about the price point.   We have a little more room to play with the base price because we are aiming the product directly at high-end audio.  If you look at the layout of the board you can see that there has been a lot of care taken in keeping stray inductance low and the caps are chosen and designed into the circuit in such a way as to really keep the ESR low and to prevent ringing.

Based on the subjective performance I'd say that we are getting the technical hurdles conquered and digital amplification is going to be the norm for most amplifiers in just a few short years.

[audiojerry]

Kevin,
Maybe some folks don't appreciate it, but I thought your explanation of the technology behind digital amplification was thoughtful and informative.

I am confused, though. What is this stuff about a Solar Hi-fi version of a digital amp? Have you announced this previously? The only thing I saw on your website was the Spectron.

Can you provide more details?
I am currently auditioning Sony's version of digital, the C70, which is an all-in-one unit. It is remarkable for what it offers for the price, but as you alluded, this type of amp is geared at the mass market, and despite its value, it doesn't approach the hi-rez performance that I accustomed to.

Based on your remarks, it sounds like your digital amp will surpass anything from the SS or tube domain. Is this your belief, or would you qualify this by saying something to the effect that it is a great value, but not the final word in high-end performance?

[JLM]

Audio Jerry:

Check out the fusion amp thread here at Solar HiFi.

Believe it or not, I'll be at another audiofest the same weekend as VSAC!

This one (hosted by Steve Deckert of Decware) is much closer to home.

jeff

[KevinW]

Kevin,
Kevin,
Maybe some folks don't appreciate it, but I thought your explanation of the technology behind digital amplification was thoughtful and informative.

I am confused, though. What is this stuff about a Solar Hi-fi version of a digital amp? Have you announced this previously? The only thing I saw on your website was the Spectron.

Thanks, I am an educator at heart, and I like to provide helpful explanations for people that motivate them towards thought.  I do need to get the Spectron info off my site, and the info on the Solar Fusion amps up.  But I have been very stressed for time in the last couple of weeks!

Based on your remarks, it sounds like your digital amp will surpass anything from the SS or tube domain. Is this your belief, or would you qualify this by saying something to the effect that it is a great value, but not the final word in high-end performance?

There is no doubt that the digital amplifier concept offers some unique solutions to common and inherent problems with tubes and SS.  Being a dedicatetd tube-o-phile at heart, I really didn't think the performance of the protoptype would be anywhere close to this good.  It beats my highly modded Tulip SET in every category, even in terms of emotional involvement, which very much surprised me.  I think my amp will set a new benchmark in terms of sound quality at the $1500 price point.  But I would not suggest that it is the best amp ever produced, which I think is what you are asking.  The opinion that truly matters is that of the customers who buy my amp.  I'm really looking forward to getting it out there after VSAC.

[KevinW]

Hey Kevin! Hope you had a great vacation when you get around to reading this. I was wondering if you have any comment about this spiel on digital amps I read in a thread, regarding the Carver and Tripath technology, and if there is truth, what you are doing to account for this in your amp? Thanks for any info you can provide!

As the "Other Kevin" so ably answered, the LC Audio modules use Black Gates directly on the module, which specifically addresses the problem mentioned in your post.  This goes a long way towards tapping out the sonic potential of these modules, as my ears can attest.

My amp also uses Jena Labs Ultrawire, which has extremely low inductance thanks to the very high strand count.  This definitely helps.  Finally, I will be offering a very special capacitor upgrade circuit that will further lower the impedance in the RF spectrum, as well have additional benefits for the stereo separation of each module.  This circuit was designed by Jena Labs, and it will unfortunately be somewhat expensive due to the cost of the high quality capacitors (~$600).  Performance without this circuit is outstanding, but for those who want the very best... it can be done.  

[Sa-dono]

Thanks for the info both Kevins! It sounds like this should be an incredible amp. Hope to get a chance to hear it sometime