[_scotty_]

WARNING LONG POST    Post edited to reflect cab of classd.org correction on PS AUDIO amp.
This thread is the result of my discussions with Kevin from Tripath Technologies tech support.
   I talked to Kevin for over an hour a couple of days ago discussing
digital amplification designs and the presence and absence of feedback in the various architectures employed by the chip manufacturers trying to get into the digital amp market. My first question was why there were so
few digital amps with feedback in the marketplace.  Kevin had a simple answer, Tripath holds most of the patents pertaining to the use of feedback
in digital amplification chip architectures. There are currently four
manufacturers of digital amps with feedback, B&O Icepower,PS Audio,Spectron,and
Tripath. Licensee's don't count. If there are more someone chime in. The result of this is that in the highend market there are eAR amps,Spectron, Belcanto,Carver Pro, and PS Audio with feedback and TACT without. The difficulties of designing a good sounding digital amp without feedback are formidable.  Stabilityand frequency response variations with load impedance are just two of many.  One of the approaches to controlling distortion in an amp without feedback is to very tightly regulate the power supply voltage.  When this is done properly you have the well received TACT amplfier with its high price tag. This sort of power supply that will power a highend amp is not cheap. The reason for the tight regulation of the power supply is twofold, it reduces distortion and it makes the amplifier more resistant to output response abberations induced by driving loudspeakers with impedance
curves that vary in resistance and phase with frequency.  They are not
simple resistive loads.  When you descend down the food chain to the mass market offerings from Harmon Kardon,Panasonic,Sharp,Sony, link to
list probably incomplete of others http://www.puredigitalaudio.org/digitalamplifiers/index.shtml click on resources,Decisions related to the final cost of the product at the consumer level have to be made. The chipsets and the MOSFET drivers are relatively
inexpensive and the highly regulated power supply necessary for an attempt at state of the art execution of the design is not.  The consequence
is higher distortion and output frequency response variations dependent
on speakers impedance curve. These non-negative feedback amps can show increasing high frequency response deviations as the amp interacts with the load through the output filter. The more a loudspeaker looks like a resistor the better it may sound with one of these less expensive offerings.
This is a partial explanation for the wildly varying reports regarding the
SONY and Panasonic Receivers of late. The Newform speaker looks a lot
like a resistor from the midrange on up. My own DIY speaker looks a lot like a resistor from 100Hz on up. These amps can sound quite good with an easier load. This is where YMMV. This also partially explains the apples and oranges when the Sony or the Panasonic is compared to a Carver Pro
amp or my own Tripath based amp. The big, simple, brute force, unregulated and technologically simple and inexpensive supplies used with the Digital amps that have negative feedback are unuseable with digital amps that cannot employ negative feedback in their design. I am a little less optimistic about massmarket digital amp designs without negative feedback taking over the world than I was a month ago when I bought my
XR25.  

Does anyone have any thoughts or other viewpoints on this subject.

[Andrikos]

Nice explanation Scotty. What is your DIY speaker BTW is it a ribbon of some sort?

[ABEX]

Scotty:
   Good post!
 
    I use to have Magnepans and they are a weird load for most amps. That is a big reason that I went back to a conventional speaker with a load that does not dip as low as most other SOTA speakers I have seen. I got tired of having to try to tailor amps and Pre-amps to fit in whereas a speaker that does not have the dips that exotics have are much easier to find a mate.

I stressed to someone that uses a set of Martin Logans that they might think twice before getting into this new Panasonic Amp.

Not that I dislike the sound of the speakers mentioned ,but I got tired of the sensitive issues which are associated with these speakers and there are many.

What speakers do you see that are of audiograde quality that you would recommend which will not need the Highend dressing which comes with ultra expensive amps like the ones you mentioned in your thread or being a relativly flat 6-8ohm load?

Hope that makes sense as It's been a long day.

TIA

[Rob Babcock]

Very interesting; that does explain why some guys are blown away by digital amps and some can't figure out what the fuss is about.  Synergy is always key, and apparently much more so with digital amplifiers.

[Enrico]

Did I understand correctly: digital amps are best for speakers that are an 'easy load', rather than 'difficult loads' like Magnepan and electrostatic?

[ABEX]

Enrico:
    That might be a good question for those who own those speakers and have the Panny amp. I would get the amp on a trial basis to see. If you have a Tweeeter Etc. they might be able to tell you as  they have the full line of Martin Logans.

     I had a hell of a time once with an HK amp with Maggies I know that much.

Speaking of speakers the amp did not favor I heard Axiom's have a bright tweeter that would not be good for it. I use NEAR Metal speakers which makes me alittle hesitant,but they have modified drivers and are not tipped torwards being bright.

Hope that helps!

[cab]

A few points mentioned in the first post need clarification I think...

Talking about feedback in digital amps, there are several varities, and each have their pluses and minuses.

Tripath may own several patents, but they have not cornered the market on feedback topologies-they use a rather garden variety pre-filter loop, which is in fact the main weakness in their design.

Primarily, there are amps that are "open loop", without feedback.  Tact is of this type. As a result, they must have a sophisticated power supply to help with the problems posed by speaker interactions. The other digital amps, for the most part, take feedback from BEFORE the output filter, including Tripath and PS Audio (which does indeed use feedback). As a result, one sees rising distortion with frequency, and difficulties driving speakers which have impedences which vary greatly. The filters are optimized at a certain impedance, usually 4 or 8 ohms, but real world speakers can present wide ranging impedances, and thus can cause less than optimal performance with these amps. This is one reason why usage of amps of this design has in the past been restricted to subwoofers and bass reproduction-the distortion is much lower at these lower frequencies.

Other designs (Spectron, Icepower, Mueta) take the feedback from AFTER the output filter. As a result, they can drive any common speaker load without ill effect. Their distortion is much flatter with frequency since the effects of the ouput filter are actively being corrected. This is  fundamentally a better design, in my mind.  

Mueta's digital amp design using both current and voltage feedback is both ingenious and a leap ahead of the rest. Their specs are much better than all the rest, and once these chips are released, have the potential to leapfrog all currently available digital amps. See their website for more info, mueta.org.

Also, see this site for a more detailed description of the various topologies:

classd.org

Happy listening....

CAB

[_scotty_]

cab, I stand corrected the SDAT output stage does indeed employ negative feedback, I was thinking of the analogue front end which PS audio claims has no negative feedback. Do you know when the Phillips DSL SODA,or the Mueta, may make it to market.  Did you happen to notice the type of power supply used by the design lads 2X100W sample
amp.  Any construction details would be helpful to the DIYer.

[cab]

The Phillips SODA will not be brought to market as it has been superceeded by their UDC amp topology, which is purportedly simpler, cheaper, and even better sounding. I do not know when or if they will ever bring it to market as an eval type product. Seems like they are only interested in licensing the technology to deep pockets. I believe it runs on a traditional power supply, as does the Mueta. That is the beauty of these amps with after filter feedback-they have a huge power supply rejection ratio and can therefore work quite well with garden variety power supplies.

I have been hounding Mueta for nearly 2 years for a sample. They promise me they will have something available early next year, around Febuary...Don't hold your breath! If it lives up to its specs, it should be worth the wait though....

[_scotty_]

It would be nice if Phillips would release a line of Receivers and amps
employing the UDC technology it would potentially blow the non-negative feedback offerings right out of the water.Enrico, ABEX I have a retired set of Magnaplaner SMG's which I will have to try out on the XR 25. While the Magnapan speakers have traditionally been viewed as a difficult load
by audiophiles they are actually not that big a problem from an impedance standpoint,they look like a 4ohm resistor.  Any speaker with stable impedance characteristics and a reasonably flat impedance curve will sound better. Zobel networks on drivers help stabilize load impedance and flatten curves. The rising impedance curve of an unzobeled dome tweeter will negatively interact with the digital amps output filter and result in a rising response. Electrostatics may indeed pose problems
when used with a Panasonic.  I am clueless when it comes to recommending speakers that might be compatible with the Panasonic
or the SONY Receivers. I haven't looked at or seriously listened to any commercial loudspeakers for fourteen years.

[Val]

I used to own Magnepans, and they aren't difficult loads by any means, being mostly resistive and flat at 4-5 ohms. If an amplifier can't drive that load their designers need to go back to the drawing board. What Magnepans need is high power because of their low sensitivity and because they need relatively high volumes to sound alive.

[ABEX]

"traditionally been viewed as a difficult load
by audiophiles they are actually not that big a problem from an impedance standpoint,they look like a 4ohm resistor."

I believe the load is likely to be in the range of 5.5 with Maggies and yes they are a stable load ,but having them be in between the value or something makes them play havoc with some amps where they will sound strident or something.Atleast that was my experience when I had mine.

That was not the reason I sold them though. Difficult loads are usually associated with loads being like APogees where they dip down to the 1.5 range. I just like having better loads that are higher now because I have had better success getting amp synergy without having to spend more than $1K plus to get adequate sound or synergy.

YMMV

[guest1721]

These are another example of a PWM based amp using local feedback.  Fortunately, the levels (if I'm not mistaken ) are resonably low, and with the switching behavior, it would seem to be a better way to implement feedback without introducing other goofy behavior.  Which leaves you with any non-linearities caused by the output filter (LC).  These should be minimal, the choke being reasonably small and the cap as well- it shouldn't increase speaker filter slopes too much so as to be audible, and in keeping the feedback local rather than global, it offers a faster loop response.

[J North]

Are these their own design or are they using someone else's technology?

Where do they stand on the digitial amp technology curve as compared to Tripath, Ice, PSAudio, Spectron, TI, Sony, etc etc.?

[J North]

Any thoughts on this?:

http://www.audioholics.com/techtips/audioprinciples/amplifiers/digital_classD_amplifiers.html

[guest1721]

Their own design, it's quite nicely implemented, thought (as with any amp) not flawless.  The Audioholics site is full of stuff to bother the audiophile and cause extra hits.  I strongly recommend skipping that site until it stops the dirty junk and plays nice.  

I just implemented balanced interfacing with it, and while it needs some ground tweaking now, it sounds unbelievable despite this.

[J North]

badman,

I have heard many good things about the LCAudio modules as well.

In what areas do you think it needs additional improvement or tweaking?

[randytsuch]

Here is a new one, just showed up at DIYAudio
http://diyaudio.com/forums/showthread.php?s=&postid=261205#post261205

Not enough information available yet, but it looks interesting to me.

Randy

[ABEX]

What do you think should be more important!
 

One thing I have learned is that one mans good\great sound is another mans sonic hell.
 

There is always differing opinions in audio. You need to find out what you think before weighing in on another's point of reference or opinion.
 

Anyways that is what I thiink,You asked remember?

[kana813]

Having spent some time with both the Spectron and lastest Tact amps,
I would urge anyone who's interested in digital amps to take a listen
to these products.

IMO, plugging a digital source directly into a digital amp, eliminating a
preamp, external DAC and the associated cabling is the way to go.

Tact amps can do this now, Spectron next amp the MIIID will also
have this capability.

Amps with digital processing power can add features like bass management, crossovers and room correction.