As many of you know I find this 'digital amplifier' stuff to be cool (just look at my avatar, it has been like that for a while). Sharp now has some new 1-bit DigiReceivers out that upsample SACD, DVD-A, DVD-V, MP3, DTS, DD 5.1, CD, and Alarm Clock to a 1/5.6 MHz DSD bitstream (Pulse Density Modulation/Delta Sigma Modulation) for amplification.

NOTE: Sharps method of digital amplification is different than that of other brands. Other DigiAmps use PWM (as seen in my avatar). Sharp uses PDM (DSD). Both have flaws and advantages.

The first new model is the SD-PX2 ($379 street).




It is only 4.5 inches think and is wall mountable. It includes a built in SACD, DVD-A, DVD-V, CD player and a AM/FM radio tuner. It has 5 digital 1-bit amps running at 5.6 MHz and capable of producing 35 watts into 4 ohms (FTC rates it as 25 watts min. into 4 ohms with 2% THD max) in all of them. If it is like other 1-bit amps, it should have 0.02 THD at 1 watt.

A lot packed into a small area.

The second unit is the HX-500 ($740 street). It is basically the same as the SD-PX2 but does 100 watts per channel and has two seperate wall mountable boxes (an amp and a player) connected by a cable. There is a HX600 which adds an av selector box.


I like it would be interesting to hear SACD/DVDA upsampled to 5.6 MHz. What do you think?

Quote:



Where did you see this? I know the Sharp units don't use a Burr-Brown chipset, but I've looked at BB's datasheets for their digital amp ICs (used in the Panasonic XR-45, etc.) and the distortion vs. power at low levels is nowhere near that good. 0.02% THD at 1 watt would be awesome for a low cost non-tube amp. I'd be very interested in any measurements you've found.

As for the Sharp amps, I ended up buying two of the older series when they were on sale dirt cheap. The sound was pretty good up top but they were lacking in bass control. The Panasonic units that are getting good reviews on AudioAsylum apparently handle the bass better. I haven't heard any of them.

Wodgy, here are some measurements at 1 watt for other Sharp units. I have yet to find the THD at 1 watt for the new generation ones.


http://www.sharpusa.com/products/Mod...58,302,00.html

SM-SX100 - 0.02 THD
~$15,000


http://www.sharpusa.com/products/Mod...58,917,00.html
SD-EX111 - 0.02 THD (at 20 watts 2% THD)
~$250

Since other Sharp units seem to have 2% max THD at near full wattage I hope the other units share the 0.02 THD.

www.classd.org reams the 1-bit method as too noisy, so its ironic they measure well. The guy who runs classd.org hates the 1-bit method. (Wodgy, please read read what the say about it under 'OEM' and tell me what you think of his arguments against it. Are they similar, IYO, to the arguments against SACD?)

I have been telling myself to get a bedroom stereo system for the past 3 years.

I am thinking of either the old EX111 (which lacks SACD/DVDA/DVDV and has only 2 channels) or one of the new ones. I hope they all have 0.02 THD.
Which would you get?

BTW, Wodgy, I know you don't like NAD. How did the old Sharp of yours compare with NAD receivers?

I believe Xanadu has a AKG K1000 plus the old $4500 Sharp 1-bit combo (he paid something like $1000 for it).

Quote:



That's interesting. I guess that's why I felt the highs on my Sharp units were smooth. As the tube guys say, the first watt is the most important. (Tube amps generally start with low distortion and it increases with output power. Solid state amps generally have moderate to high distortion at low power, low distortion at medium power, and their highest distortion at high power.)

Quote:



Some of his arguments are similar to the arguments against SACD (e.g. high frequency noise) while others aren't (e.g. efficiency). Noise is better than distortion, though, and it's a reasonable tradeoff.

Check out his comments on TI's chipset (used in the Panasonic amps); they're interesting and useful. It turns out that I was looking at the wrong TI chipset earlier. Here's a document that contains full measurements of the same chipset used in the Panasonic digital amps:
http://focus.ti.com/docs/apps/catalo...tName=slaa117a

THD+N for a 1kHz signal at 1 watt is 0.03%, which is seriously good. The graph of THD+N vs frequency is very interesting and unusual. Take a look yourself. I believe the "blue" line is the middle one (the flattest). At only 10 watts there is a remarkable amount of distortion at 10kHz. (I believe they've added some kind of filter to the measurement equipment after 10kHz.)

As predicted for a digital amp, the frequency response curve varies according to the speaker load, but not too badly. The FR rise at 20kHz could make 8kHz speakers sound a little bit tipped up, but just a little.

Altogether very interesting, though. I wish they provided a graph showing the distortion spectrum.

Quote:



To be honest, I liked the tone of the Sharp much better. Most NAD amps could outperform the Sharp in the bass, except for the NAD C320BEE, which has a serious bass deficiency (unless you bypass its preamp section).

So, you think the Sharp and the TI are both very good technologies? I can see that you seem to like the TI more. Is there anything you feel the Sharp method out does the TI method in?

Also, Sharp seems to insist that the signal stays digital and is not converted to analog first...which is against what Classd.org says. What do you think? If it says digital then the Sharp has an advantage in the SACD department.

This is Sharps demonstration of the signal path (which contradicts classd.org):




Where can I hear about this Panasonic amp (does it have pure digital input)?

Thanks.


BTW other HeadFiers, feel free to chime in. I didn't post this just for Wodgy and me.

AFAIK, the panasonic/ti is supposed to stay digital all the way to the end as well. Everyone dislikes its analog inputs except on the new xr50.

Quote:



Well, I think they're both interesting. It's not clear to me that either of these technologies has any advantages over a conventional analog amplifier, except for greater heat efficiency. (They also have obvious disadvantages, including frequency response curves that vary depending on speaker load -- and the graphs in the TI document I posted earlier were measured into a resistor, rather than a real speaker, so real-world behavior is likely to be much worse.) These digital amps do have a slightly different behavior at low power, but this behavior is not uniform across the frequency spectrum, so it's hard to judge. My guess is that at low power they would sound qualitatively different than a conventional amp.

It's difficult to make real judgments about these amps because even the TI document I posted earlier does not contain a complete set of measurements. What does the distortion spectrum look like? What about IM distortion measurements? What is the damping factor? Does it vary by frequency? etc.

Quote:



Keep in mind that "staying digital" is not the same as saying there are no conversions. Both the Sharp and TI technologies require a PCM->PWM/PDM conversion, and, as the TI document puts it, this conversion is "nonlinear" (i.e. lossy). In fact, as the Classd.org guy points out, the PCM->PWM/PDM conversion with today's chipsets introduces more distortion than a high quality digital to analog conversion in a conventional analog amplifier. (e.g. For the TI chipsets, the conversion has an error of 0.01%, which isn't bad but can also easily be outperformed.)

It's not clear from the diagram you posted whether the newer Sharp amps do an analog conversion first for digital signals, or whether the information stays digital, but it really doesn't matter either. Even with SACD input, the DSD signal will get resampled to the digital amplifier's higher sample rate.

The "it all stays digital" claim is really just misleading marketing, meant to make you think that there are no conversions going on behind the scenes. That's not true, except perhaps in the case of the TI/Panasonic amps with SACD input only. The real question to ask is "is the PCM->PWM/PDM conversion better than a PCM->analog conversion"? At current amplifier sample rates, the answer is no, at least from a measurement perspective.

Personally, Czilla, I'd just suggest going to a store and auditioning some of these units. It's difficult to tell how they'll sound from the measurements alone. Also, unless small size is an absolute requirement for you, don't discount traditional analog amplifiers. Choose what sounds best to you.

Quote:



And that's a bad thing? How does it differ from upsampling a 16/44.1 signal to 24/192 (with the exception of it being PDM - not PCM- in the Sharp)? Whether it is good to upsample DVDA/CD to 5.6 MHz goes back to the whole PCM vs. DSD debate. What's the difference between resampling and upsampling?

I still plan on listening, but I just like to know the technical stuff.

Quote:



You mean the Sharp amps? As far as I can see the PDM must first be converted to PCM and THEN to PWM in PWM digital amps. That is the process in Sony's SACD digital amp all-in-ones. If the world is going to go with SACD (which is unlikely) then doesn't the PDM digital amplification make perfect sense? Amplifying a PDM/SACD bitstream directly (ignoring my first question in this post) via a PDM amplifier would mean perfect accuracy theorically?

THANK YOU!

Quote:



Well, I only have a background in traditional (PCM) signal processing, so take what I'm saying with a grain of salt. PCM-to-PCM resampling is well understood and linear -- in fact, at integer multiples, done correctly, it can even preserve all the original values. It's not clear to me if PDM to PDM resampling can be done linearly, though it seems likely that a doubling of the sample rate can be accomplished easily enough.

There aren't really any logical or engineering reasons to upsample DVDA/CD to high bit rate PDM just for the heck of it. The motivation behind any kind of conversion/resampling is to make the ultimate conversion to analog more accurate. So this isn't really like the DSD vs. high bit rate PCM debate. The real question at hand is how much distortion does each of the following paths generate?
1) PCM -> PDM -> amp -> speakers
2) PCM -> analog -> amp -> speakers
Each path starts in the same place and ends with the same result. The only legitimate reason to choose one path or the other is because one is more accurate than the other. Currently path 2 is measurably more accurate, even though it might not be as intellectually "pure". Staying all digital until the speakers is good for marketing, but if it isn't more accurate, it doesn't make engineering sense.

Quote:



You're right about the first two points. As for the last point, I don't think it makes perfect sense right now, but maybe one day in the future once the amplifier technology itself improves. The TI chipset is supposed to be the best currently, but from the measurements in the document I posted a link to, its clear that these amplifiers are not competitive with conventional amplification except at low levels of power (around 1 watt) where they might even be better than conventional amps.

By the way, you might be interested in Stereophile's rather complete measurements of the PS Audio HCA-2, which is a digital amp priced at $1700:
http://www.stereophile.com/amplifica...29/index8.html

These completely suck. Even ignoring the absolute distortion, all of it is odd harmonics!

I hope this is not typical of digital amps in general. Odd harmonics generally do not produce a pleasant sound. Perhaps this is why it's so hard to find distortion spectra measurements for these amps?

HCA-2 is popular with many audiophiles. I guess it goes to show that either audiophiles like distortion and that perhaps our current measurements can't define sound -or- measurements aren't everything.

I am kind of intellectual soup-Nazi when it comes to measurements, so I have trouble believing the latter. I believe that everything in this world can be measured.

As Lord Kelvin said:

Quote:



Quote:



However, while everything may be measurable, its importance must be weighed by principle and reason.

If digiamp technology expands at the rate of Moores Law it should surpass analog amplification. Just compare the THD numbers of the past generations of Panasonic digiamps. I remember seeing on audio circles the the XR10 has 0.9 THD at one watt. I believe the XR50 has 0.03 THD.



I have to wonder why push DVD-A or SACD technology if digital PWM is the future. Why not do like the LaserDisk and create a hi-rez format in PWM, avoiding all conversions. Of course, Sharp could use a similar argument with SACD.

Quote:



I definitely agree with that.

Quote:



You're comparing apples and oranges. The 0.9% THD measurement is almost certainly at full output, not at one watt. The chipsets have not improved significantly in much but total power.

Amplifiers are governed by a set of physical laws. IMHO it's as unlikely that they'll double in performance every 18 months as it is to expect the maximum speed of aircraft to double every 18 months, or the fuel efficiency of cars to double every 18 months.

If indeed digital amplifiers have almost entirely odd harmonic distortion, and if this was audible, I'd find it sad if digital amps are "the future." It would be better if technologies were selected on the basis of how well they match human perception rather than selected on the basis of clever marketing. We're still trying to recover from the inferior sound of CDs, which won out partly because they were digital and digital "must be better." The truth is usually in the details.

Quote:



For the average consumer it is probably a step up.





One last question...is having an amplifier sample rate (for PDM) of 5.6 MHz better than an amplifier sample rate of 2.8 MHz. Is higher better (up to a point)?

Basically, is converting PCM to 5.6 Mhz less lossy than converting it to 2.8 MHz. With SACD, could 5.6 MHz conversion be beneficial as upsampling is with PCM?

Is their any chance that converting PCM to 1/5.8 Mhz is good, like converting 16/44 to 24/96 (assuming that PDM *is* better than PCM which is debatable)?

I know these questions are similar to the ones I have already asked but I like to be clear. Thanks.

I just remembered that many audiophiles pay thousands of dollars for PCM to DSD conversion - just look at the dCs Purcell and EMM labs products. The Purcell is considered the best in the world.

From dCs Purcell's site:

The Purcell may be ordered with a PCM to DSD upsampling option, which allows all sample rates from 44.1kS/s to 192kS/s to be converted to a DSD data stream running at 2.822MS/s. Many people regard the CD to DSD conversion as superior to even the CD to 24/192 conversion. Typical comments are that it is even more musical and natural, with better low level resolution. As always we advise you to try this out for yourself, but we are confident that you will be impressed.

From this angle the Sharp is an excellent value (esp. at 5.6 Mhz). I wonder if the analog outputs (RCA) of the Sharp are upsampled.

(Of course audiophiles spend thousands on cables to make of this what you will)

Quote:



Almost certainly. In fact, the higher sample rate would negate a lot of the bad feelings I have about SACD (it's too close to "just good enough," which is a poor engineering philosophy in general).

Quote:



Yes to the first. I can't see any same benefit to the latter, though the conversion I have in mind in this case (doubling the sample rate) is also lossless.

Quote:



The reason we upsample PCM is to ease the work of analog filter design and, in the asynchronous case, to reduce jitter. If the PCM to 5.6 Mhz conversion helps ease the work of amplifier design (that's why they're doing it), then it would be worthwhile.