or Connect
AVS › AVS Forum › Display Devices › Ultra Hi-End HT Gear ($20,000+) › Analog architecture of Theta vs. Digital solution of Meridian (questions!!)
New Posts  All Forums:Forum Nav:

Analog architecture of Theta vs. Digital solution of Meridian (questions!!)  

post #1 of 30
Thread Starter 
Hi folks,

In my never ending quest to decide on a HT processor, I'm debating the Theta CBIII or Meridian 861. I am still leaning toward the Meridian, but someone pointed out the fact that Theta has made significant upgrades to their analog section over the last few years, whereas that may not be the case with Meridian.

I know the Meridian controls everything, including the volume control architecture, through the digital domain.

Theta's architecture manages the volume control through the analog domain.

What inherent pros/cons are there to either solution when dealing with HT, or is it irrelevant for HT?

Is Theta's advancements in their analog domain predominantly for 2-channel, or is there significant improvement in HT sonics due to this advancement in technology?

Please help!!
post #2 of 30
take this comment in the spirit it is meant. I would forget about the technology and go and buy something you like the sound of. You should spend considerable time with both of those units before you lay down $$$ on either one. I know that may sound like a cop out, but they really do sound very different.

I demo'd a fair number of processors a year or so ago, including the 861 (previously owned 568) but not the Theta (heard the MkII before and was put off by the reports of bugs and terrible UI) and eventually settled on the EAD 8800Pro which is a fantastic sounding pre-pro. One big advantage IMHO was the EADs ability to bypass 2 or 8 channel analogue signals through without redigitising, I use this with good results with my 2 channel CD player.
post #3 of 30
Stieger says:

Quote:
I know the Meridian controls everything, including the volume control architecture, through the digital domain.
Not entirely true.

Wherever the volume control is (on the analog output) or on the loudspeaker (DSP loudspeaker) it is a hybrid volume control. All calculations are done in 48-bit processing. Volume is controlled digitally until the point that you would affect 24-bit resolution where analog kicks in.

Meridian is often overlooked when used with conventional speakers, however you are doing yourself a disservice by not checking them out.

Cheers,
post #4 of 30
Thread Starter 
Howdy,

I'd love to listen to both units, but my local meridian dealer doesn't have any Meridian gear in stock - it's at the president of the company's house. It doesn't do me any good there....

Regarding Theta, my local dealer just recently cancelled their license.

Thus, I'm in a "pickle" of sorts.

Regarding the last post - I have analogue speakers/amps, so not sure how that differs with regard to Meridian's technology surrounding digital volume control.

Best,

Eric
post #5 of 30
Quote:
I'd love to listen to both units, but my local meridian dealer doesn't have any Meridian gear in stock - it's at the president of the company's house.
If you can't even get a demo of the equipment then imagine how bad the service will be after you've purchased the unit...
post #6 of 30
Eric,

It doesn't matter. Whether it's a DSP loudspeaker, or an analog output to an external amplifier it works the same way.

Volume control is digital until you start sacrificing a bit of resolution, at which point it is transferred to analog.

Greg:

Frankly, if the M kit is at the president's house it's a fairly profound statement ;) Think about it for a second or two!

Cheers,
post #7 of 30
Thread Starter 
I'm buying from a dealer who's a friend in LA. The local guys don't have a clue. At least they were honest when I asked them questions, they told me they didn't know the answers...

stieger
post #8 of 30
"Wherever the volume control is (on the analog output) or on the loudspeaker (DSP loudspeaker) it is a hybrid volume control. All calculations are done in 48-bit processing. Volume is controlled digitally until the point that you would affect 24-bit resolution where analog kicks in."

Then that would mean the volume in the Meridian is more analog then digital. The processing is run at 48 bits precision but what the DSP passes to the DACs certainly isn't 48 bits... at most it would be 24 bits.

Shawn
post #9 of 30
Shawn,

And that leaves 24-bits for volume control.

If memory serves me correctly it was a setting of 52/-35 dB depending on which volume display option you're using. So most of the volume control is indeed analog...

However I rarely listen with the volume set "that low" and my default volume setting is 65 (-22 dB).

Cheers,
post #10 of 30
Stieger

Check out you local LINN Specialist Retailer. He will have on demo the solution from LINN ( Kisto or Kinos ).

He has a contractual obligation to have one of these on demo. That does not mean just a sticker on his shop window or a catalogue on the shelf, like most places.

LINN is one of the very few, if not the only, High End manufacturer to operate with a selective distribution policy which ensures you the end user can demonstrate on the correct conditions any of their products. He must carry a minimum demonstration stock and have adequate demo premises.
Included in this contract is staff traing, presentation and in-house housekeeping.

This cannot be said of Meridian, Krell, McIntosh and all the other usual suspects who sell to anyone that can shift their boxes.

Check out their gear, I´m sure you will be surprised how good it is. At least it won´t be in the house of the guy´s boss, which incidentally says nothing about how good or bad the product is.

If you do not like the it at least you have had the chance to demo it.

Gordon
post #11 of 30
John,

"And that leaves 24-bits for volume control.

If memory serves me correctly it was a setting of 52/-35 dB depending on which volume display option you're using. So most of the volume control is indeed analog..."

If they aren't loosing 'bits' it would need to be lower then that. If they digitally attenuate the signal just 6dB that lost a bit of resolution... and 24 bit DACs don't really have a full 24 bits of resolution because of their own noise.

If 0dB was full scale to the dac (which it may or may not be... how much higher above that does the volume control go?) a setting of -35 has lost nearly 6 bits of resolution... meaning the signal to dac noise ratio has gotten 35dB worse then it was before.... unless this was done in the analog domain which would attenuate the dacs noise along with the signal.

Nobody has scoped a Meridian's dacs to see how they really do this?

Shawn
post #12 of 30
Shawn,

Again, you assume that they are hitting the data bits for volume control, and you have literally no basis for this point of view. If it was, we'd hear issues with noise on the processors, and as you well know that decidedly isn't the case.

Meridian isn't willing to give up much on the technical details of their volume control, and I don't blame them. Does Lexicon give up much information on their proprietary details? Nope. Theta? Nope.

If you want to know more specifics, call up Marc Koval and see if he'll give you more information.
post #13 of 30
John,

"Again, you assume that they are hitting the data bits for volume control, and you have literally no basis for this point of view. "

If they are adjusting the volume digitally they are lowing the signal in the digital domain. That means the signal is getting closer to the noise floor. That means they are effectively using less bits... hence they are lowering resolution.

"If it was, we'd hear issues with noise on the processors, and as you well know that decidedly isn't the case."

The noise level wouldn't increase, the signal decreases... IOW gets closer to the noise floor.

If they aren't doing volume control in the digital domain then that isn't the case as analog attenuation after the DACs also lowers the DACs noise floor along with the signal.

Or they could be taking a hybrid approach like Wadia did. They do up to about 6dB digitally then switch in 6dB of analog attenuation and repeat that. That 6dB digitally still lowers the S/N ratio.

Shawn
post #14 of 30
Wadia's don't behave that way, as I recall, at least with my previous 27ix.
They have DIP switches internally where one sets the maximum analog output voltage. You would set this value based on your systems' overall sensitivity, taking into account your amp gain & speaker sensitivity & desired listening level. "They say" a setting of 70 on a Wadia product is where you could start to detect lost bits. The downside of Wadia's approach is that one soon realizes that CDs are recorded at levels all over the map, and that dialing in one setting of internal DIP switches is never optimal for all listening material & situations.

The old EAD DSP-9000-MK3-Pro DAC had a unique volume control ( had one of these too ). There would only ever be 6db of digital volume attenuation at which point a resistor ( of increasingly larger value ) was inserted in the signal path.

VikingBoy: OT -> What CDP are you using into your 8800-Pro ?
I went from Wadia+DCS, to Sim Eclipse, to many others in between, and am exceptionally happy just using the DACs in my 8800-Pro fronted with a Monarchy DIP Upsampler ( for my 2-channel ).

On Topic: Forget the architecture and decide with your ears & wallet :)

- Andy
post #15 of 30
I'm frankly not sure I understand wall the fuss is about concerning digital volume controls. A 24-bit DAC has what, over 145dB of dynamic range? So if you set your reference level is slightly louder than you'd ever care to listen, are you ever going to be able to hear the noise floor? Add even a mild amount of analog control and I can't see how.

EDIT: Well, nevermind. No real system achieves the full 24-bit dynamic range, of course. DAC nonlinearities and analog noise get in the way. So I could see how you don't quite get enough dynamic range out of a practical DAC to just "set it [reference level] and forget it".

Still, it doesn't seem like you need much analog adjustment to bury even that noise floor below the hearing threshold.
post #16 of 30
Quote:
Originally posted by John Kotches



Volume control is digital until you start sacrificing a bit of resolution, at which point it is transferred to analog.


Cheers,
John

Is there a simple explanation for the advantages this hybrid volume control has over a complete analog volume control? Is it cheaper? It can't be simpler.

Thanks Rick
post #17 of 30
Andy Lammer: I'm using an Exemplar modified Denon 2900. It has a valve output stage and sounds considerably better than the EAD DAC. The EAD DACS are better than the standard Denon 2900 outputs prior to modification.
Theres a whole load of stuff on Exemplar over at Audiogon.com forums and a review at six moons.
http://www.6moons.com/audioreviews/exemplar/denon.html

I understand the reviewer at six moons didnt disable the video section when he reviewed the unit which might account for why he wasn't as blown away as others. Switching the video display off makes a noticeable difference.
Rather than drag the thread off topic, PM me if you want any more info.

regards
Ian
post #18 of 30
Rick:

It enables Meridian to use a consistent volume control strategy in their processors and DSP loudspeakers.

Shawn:

Sorry, that isn't the case. You will probably find that the architecture for Meridian is similar to Wadia's and I am quite confident that they aren't sacrificing a single bit of resolution for their volume control mechanism.

Why not call Marc as I've suggested, and talk to him about it. I won't be able to talk with Meridian until CES time.

Cheers,
post #19 of 30
John,

"Sorry, that isn't the case."

Digital volume control lowers the signal level down (uses less and less bits the more you attenuate it) without changing the dacs own noise floor. Unless the volume control DSP is also changing the noise by adding more noise (dither) which is what Meridian has done before. The more you attenuate below 0dBfs the less and less bits that are being used.

If you watch the digital bitstream on a scope feeding a DAC the bits literally look like a VU meter as more and more are toggling on/off as the signal level to the DAC is increased up to 0dBfs. When the volume level is decreased so are the number of bits being used. Play a test CD recorded at very low level and only a few bits are being used at all. Play silence and all bits stay low.

That is simply the way it works.

" and I am quite confident that they aren't sacrificing a single bit of resolution for their volume control mechanism."

They did in the 518... they did in the early versions of the 565 which allowed for digital volume control on the digital outputs....

Shawn
post #20 of 30
Shawn,

I'm reporting what I am told. In the time that I have known Bob Stuart (about 5 years now), he has yet to be factually incorrect.

Since I'm handling the Q&A for the Hitchhiker's Guide, I'll have time to discuss this with Bob (and probably Duncan as well).

Cheers,
post #21 of 30
John,

Please do, it will be interesting to know how Meridian does it.

I'm just telling you how PCM works, no neccessarily how Meridian is currently doing their control.

Do you know binary?

If not disregard the rest of this message.

If so that all this really is.

Say for example 65,535 is as loud as a signal can get... 0 is as quiet as it can get.

Pick any number between 0 and 65,535 and write it in binary. As you choose lower numbers (aka lowering the signal level) and write that in binary you will find it takes less bits to reproduce those numbers in binary. IOW, the lower signal is using less bits.

65,535 is 1111111111111111 in binary... IOW 16 bits of data.

20,000 for example is represented by 15 bits..
10,000 needs 14 bits
3,000 needs 12 bits
etc...etc...

Shawn
post #22 of 30
John---clearly Shawn doesn't know you that well, if he has to ask if you know binary :)

Having said that I'm inclined to agree with Shawn, that any use of digital attenuation on a 24-bit signal is going to cause a loss of LSBs. Of course, if you're piping a 16-bit signal to a 24-bit DAC with a digital volume control, that's a different story---but any 24-bit native signal is going to lose resolution upon attenuation. I suspect there must be some sort of semantic disconnect going on here.

At the same time, I frankly don't see what the big deal is. What matters is not the SNR of the resulting digital output, it is the absolute noise floor level determined by the combined analog and digital volume settings and the amplification. You could do 30dB of digital attenuation, dropping the digital SNR by 30dB, and it wouldn't matter---if the DAC noise floor has been set comfortably below the hearing threshold.
post #23 of 30
Michael,

"Of course, if you're piping a 16-bit signal to a 24-bit DAC with a digital volume control, that's a different story-"

I've only scoped a few different DACs but all of them were 24 bit and on all of them when fed 16 bit data the 16 bit data was padded with zeros so that the datas MSB was also the MSB of the DAC. That makes sense as it will keep the relative playback levels consistant between different bit depth sources and also maximize the S/N of the 16 bit data relative to the dacs noise.

Shawn
post #24 of 30
Yes, of course that makes sense, but what I'm talking about is adding a digital volume control to that---so that the 16 bit signal is scaled to anywhere between the 16 MSBs to the 16 LSBs. (Of course for non-trivial scalings you'll have some roundoff error and therefore new quantization noise.)

Still, what I'm saying is that as long as the 24-bit DAC's noise floor remains below the hearing threshold (via analog volume control and amplification) then digital attenuation should be benign---even if you do lose resolution.
post #25 of 30
Michael,

The latest generation BBs are hittting 132dB in SNR, so it's starting to get wayyyyyyy up there in performance. I would have thought 22ish bits of resolution wasn't doable 3 years ago when 20 was a struggle.

My guess, and it is only a guess, is that they are padding either side of the signal, and using the digital control as a bias to set maximum signal voltage on the output.

Pad a 24-bit signal, with 8 below for rounding, 8 above for rounding and 8 for volume gets you suspiciously close to the 52 volume setting for the transfer. 99 is maximum output level, 52 is 47 offset. Hmmmm.......

Shawn,

I prefer hex to binary, as it's a lot less typing. I'm quite comfy with binary if I have to be.

Apparently you've never seen me use shorthand like 2^10 etc to illustrate the difference between decimal K and binary "convenience K". 2^24 is about 16 million discrete voltages, or 8 if we are not considering +/- flagging to be discrete voltages.

I'm working mostly with silly stuff like Data Center Infrastructure automation these days, and I don't spend as much time as I used to fiddling with the lower level bits.
post #26 of 30
John,

"My guess, and it is only a guess, is that they are padding either side of the signal, and using the digital control as a bias to set maximum signal voltage on the output.

Pad a 24-bit signal, with 8 below for rounding, 8 above for rounding and 8 for volume gets you suspiciously close to the 52 volume setting for the transfer."

This is the part you keep getting stuck on... the AD 1852 DACs in your 861 are 24 bit DACs. You can not pad a 24 bit signal with 8 more bits on each side and feed that to the DAC, it simply has no mechanism for accepting/using anything more then 24 bit data. Look it up in the engineering notes for the DAC if you don't believe me.

http://www.analog.com/en/prod/0,,761...1852%2C00.html

"I prefer hex to binary, as it's a lot less typing. "

Certainly is, but when you are watching the bits toggle on and off to a DAC binary makes more sense then hex since you are literally watching binary in action.

Shawn
post #27 of 30
Yep, that's why I suspect there's some sort of semantic disconnect here. At some point you're down to 24 bits, and unless you've done no digital attenuation you've necessarily lost some LSB's. If done right, it won't matter, but you will nonetheless.
post #28 of 30
Another digital player is TacT Audio -> the room correction folks.
As I recall, the TacT converts everything to 48-bits before doing their math.

A TacT RCS unit, without any TacT digital amps connected, will do solely digital volume control. Wherease a TacT RCS connected to a TacT digital amp will NOT use any digital attenuation, as the TacT RCS will control the TacT amp's voltage rails -> so purely analog control.

On the subject of TacT, have you considered a TacT TCS unit ?
http://www.tactlabs.com/Products/TCS/TCS_Frames.htm

- Andy
post #29 of 30
If the TacT could do multichannel high-res digital in, it would be on my short list, that's for sure. They make some sweet gear.
post #30 of 30
Quote:
Originally posted by John Kotches


Meridian is often overlooked when used with conventional speakers, however you are doing yourself a disservice by not checking them out.

Cheers,
That is not true as most people who have the Meridian processor don't care to buy their digital speakers.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Ultra Hi-End HT Gear ($20,000+)
This thread is locked  
AVS › AVS Forum › Display Devices › Ultra Hi-End HT Gear ($20,000+) › Analog architecture of Theta vs. Digital solution of Meridian (questions!!)