The biggest difference I hear between digital and analog

[NorthStar]

Tim, I have quoted you a response as it is part of the answer to your comments. But its not all- there is another factor.

Many preamps employ active EQ to do their job. The problem is that there is a propogation delay (the very measurable time it takes for a signal to move from input to output) in the phono (or tape) preamp. If active EQ is involved, the feedback signal will arrive slightly late at the input. Now at low frequencies this is not a problem, but as frequency goes up it becomes more and more pronounced (because the propagation delay is a fixed value). The phenomena varies from circuit to circuit depending on internal speed (slewing) of the circuit.

What can happen, and happens more than the industry really ever seems to talk about, is that a tick or a pop (which is a high frequency event due the sharp risetime they require) can 'ring' in the circuit. A simple way of looking at this is that the tick might appear, it gets amplified and fed back, but at the input the tick is already gone. But we now have a (weaker and negative) copy of the tick that is once again at the input of the preamp, offered by the feedback network. It has to make its way through the preamp too! This is a simple model of how this ringing can occur.

This problem seems to be exacerbated by poor high frequency response/slewing rate; of course there is a direct correlation to propogation delays in such circuits being longer.

The solution is to use no feedback (a circuit with no feedback is more stable than one with feedback and stability in this case is important) and passive EQ. You can still get the bandwidth (we spec 100KHz) but now the source of ringing has been eliminated. Since the circuit is zero feedback, methods other than feedback have to be employed to keep distortion down. (Keep in mind also that the preamp cannot be overloaded by the tick or pop; it must have very good IMD at high levels as previously implied.) The supply in particular must be quiet to prevent IMD. The tick or pop is then presented in its actual, non-emphasised amplitudes, occuring over a shorter period. The result is they are far less audible! I had this demonstrated to me in spades about 25 years ago; the difference can be quite dramatic, one where you would have thought the same record to be defective with one preamp and fine with another.



Bandwidth and high slew rate are really one and the same. If you want a circuit to have a high slew rate, it will also be capable of wide bandwidth. You can't "get the real audible audio band right in the first place" if you don't have the bandwidth (as Holman was fond of pointing out decades ago), you need that to reproduce phase relationships correctly! 20Hz to 20KHz just will not do it. You need 2Hz to 200KHz to be able to reproduce all audible frequencies without phase shift. This is because group delays which are an artifact of the cutoff frequency will manifest 10X higher than the lower cutoff, and 10X less than the high cutoff.

---- Great post Ralph, and very interesting your last point about proper frequency response.
...I was, for a very long time, under the assumption that a frequency response (bandwidth, slew rate) of 5Hz to 88kHz was sufficient (roughly 4X at both ends the standard of 20 to 20). ...But you say 10 times.

 

[mep]

I think this latest turn of events in the way this thread is headed is a tempest in a tea pot. People with good tables, arms, cartridges, and phono sections who use an RCM should have very low noise floors and damn few pops and ticks. The way this thread has headed, it makes it sound like everyone who is listening to vinyl is listening to a Rice Krispy Symphony (snap, crackle, and pop) which simply isn't true. For the digital lovers who have long since abandoned analog because their systems really did produce Rice Krispy Symphonies with basically every record they played, this latest direction of the this thread is playing into their hands and giving them a reason to smirk. Are LPs perfectly quiet 100% of the time? No. Does the average record sound super quiet the majority of the time? Yes. If you answered "no" to the last question, something is amiss in your system and/or you have records in poor condition.

As your system gets better, your noise floor goes down and that holds true for both digital and vinyl. It's easier and cheaper to have a very low noise floor with digital, but it doesn't mean that analog set up correctly doesn't have a very low noise floor that would easily satisfy all but the most analog hating digital diehards (and we have some here on this forum and that's ok).

 

[jkeny]

An interesting thread which I expect will converge into the oft-quoted phrases "adequately designed" & "transparent". Oh, sorry I see these phrases have already been used.

The phenomena that was rejected earlier in this thread that the sound could be in a different plane to the rest of the music, now has a technical explanation - where tics and pops & surface noise are correctly reproduced by the "good" phono stages (& cartridges & amplifiers & speakers) at their correct frequencies & therefore occupy a perceived plane different than the music and not within the fabric of the music itself. The brains sensory gating is then better able to tune them out. This differs to the spray of IMD products often produced which perceptually intermingle with the reproduced music !

 

[jkeny]

Mep, ultrasonic noise will not be heard directly so just because it sounds quiet does not necessarily mean there is no noise above 20KHz. Modern cartridges can produce frequencies well over 20KHz.
The possibility is still there for these frequencies to cause slew rate limiting, ringing & IMD products within the audible range.

BTW, this is also possible in digital systems - so not exclusive to analogue but it's sonic consequences are probably very different between digital & analogue reproduction?

 

[microstrip]

An interesting thread which I expect will converge into the oft-quoted phrases "adequately designed" & "transparent". Oh, sorry I see these phrases have already been used.

The phenomena that was rejected earlier in this thread that the sound could be in a different plane to the rest of the music, now has a technical explanation - where tics and pops & surface noise are correctly reproduced by the "good" phono stages (& cartridges & amplifiers & speakers) at their correct frequencies & therefore occupy a perceived plane different than the music and not within the fabric of the music itself. The brains sensory gating is then better able to tune them out. This differs to the spray of IMD products often produced which perceptually intermingle with the reproduced music !

jkenny,
After reading your post I feel much better ...

 

[jkeny]

jkenny,
After reading your post I feel much better ...

But is it real or do you have a measurement to prove it - otherwise it's anecdotal & will not be treated seriously

 

[MarinJim]

All I know is the "goose-bump factor" Like an opinion, everyone has their own. With the right ss amp, Concert Fidelity tube pre, Venture speakers (coming from Focal Alto Be's and 802D's), GNSC Wadia S7i ( the best cdp I have heard, but so-so with usb) I have embraced digital. I really appreciate all the comments, very entertaining.

 

[mep]

An interesting thread which I expect will converge into the oft-quoted phrases "adequately designed" & "transparent". Oh, sorry I see these phrases have already been used.

The phenomena that was rejected earlier in this thread that the sound could be in a different plane to the rest of the music, now has a technical explanation - where tics and pops & surface noise are correctly reproduced by the "good" phono stages (& cartridges & amplifiers & speakers) at their correct frequencies & therefore occupy a perceived plane different than the music and not within the fabric of the music itself. The brains sensory gating is then better able to tune them out. This differs to the spray of IMD products often produced which perceptually intermingle with the reproduced music !

I think this portion of the thread has been misguided and misleading. First of all, this portion of the thread that I started long ago makes it sound like tics, pops, and surface noise are a predominant part of analog LP listening which simply isn't true. Another thing that has been missing is how different stylus shapes (and cartridges) can further reduce noise to even lower levels than the low levels we are already accustomed to. Somehow this has gotten off track and the phono stage has taken prominence which I don't find to be the truth. I do believe that phono stages can lower the overall noise floor by having a better S/N than their lesser brethern, but if you have crappy LPs that were either bad pressings or they were never taken care of properly, nothing is going to quiet those down short of throwing them in the trash where they belong.

Those of us that love the sound of analog and realize what we don't hear with digital (assuming you aren't a one-trick pony and you listen to both on a regular basis) don't need to try and convert digital diehards over to loving analog. It's not going to happen and frankly, I don't care. You either get it or you don't. Discussions like the ones that are happening now on this thread are a mere smokescreen meant to poke fun at analog technology and those who listen to it in case you haven't figured that out by now.

Honestly, I believe that if analog was as silent as digital that digital people still wouldn't like it because analog requires more money, more time invested, more space, less convenience, and a much higher skill level in order to pull it off at the levels that can be achieved. Complaining about analog noise levels is obscuring the real truth of digital's shortcomings compared to the beauty of great analog. Those that hear both digital and analog on a regular basis in their home on their systems realize that DSD has closed the gap, but a gap still remains. As you go below DSD quality, the gap widens no matter how much you don't want to believe it's true. But then, the ironic thing is that some digital lovers argue against each other on how much resolution you need with some claiming that MP3s are as good as it gets and you don't even need RB. In the digital world, according to some digital lovers, it's good to be at the lowest level of resolution and listen to lossy MP3s which has thrown information out the digital window in order to save storage space and swear you can't hear any difference between MP3s and CDs. I have never heard an analog lover saying they wish they had less analog resolution to listen to.

 

[opus111]

As your system gets better, your noise floor goes down and that holds true for both digital and vinyl.

Yep - this is most probably the single metric that correlates best with SQ. But its not the noise floor as traditionally measured - i.e. with nothing playing. Its the dynamic noise floor while music is being reproduced.

It's easier and cheaper to have a very low noise floor with digital, but it doesn't mean that analog set up correctly doesn't have a very low noise floor that would easily satisfy all but the most analog hating digital diehards (and we have some here on this forum and that's ok).

Its only easier and cheaper to have a very low noise floor with digital if we adopt the traditional way of measuring noise floor - with zero stimulus. If we adopt the method that matters, then its easier to do this with analog simply because the bandwidth is narrower with analog and there are no aggressor sources (clocks) either. The problem with most digital kit currently available is that it would not measure well against analog in terms of its dynamic noise floor but as yet there's no standardized measurement to check this. We still have to listen.

 

[mep]

Yep - this is most probably the single metric that correlates best with SQ. But its not the noise floor as traditionally measured - i.e. with nothing playing. Its the dynamic noise floor while music is being reproduced.



Its only easier and cheaper to have a very low noise floor with digital if we adopt the traditional way of measuring noise floor - with zero stimulus. If we adopt the method that matters, then its easier to do this with analog simply because the bandwidth is narrower with analog and there are no aggressor sources (clocks) either. The problem with most digital kit currently available is that it would not measure well against analog in terms of its dynamic noise floor but as yet there's no standardized measurement to check this. We still have to listen.

You need to clarify what you mean by bandwidth because the audio bandwidth with digital is severely limited compared to analog.

 

[opus111]

Yep, good point. Happy to clarify that I mean the bandwidth that's needed beyond the DAC. A digital signal coming out of a DAC chip is an RF signal. Take a multibit DAC - Lynn Olson measured his PCM63 and found its output waveform had a bandwidth that extended beyond 20MHz. If you happen to use an ESS Sabre DAC, I reckon you could easily multiply this figure by 5.

 

[MarinJim]

Digital at higher resolutions sounds pretty damn good to me. However, I have found the source is not the most important factor in system quality.

 

[NorthStar]

-- Indeed, this is an excellent thread.

Analog is organic.
Digital is euphemistic.

 

[opus111]

However, I have found the source is not the most important factor in system quality.

Interestingly I've found the total opposite. The source is the primary factor, GIGO. If the recording sucks (DSD examples I've heard fit the bill) then its pointless to build a decent system to replay it. Beyond that, the DAC in a digital system is where its at.

 

[Atmasphere]

because the bandwidth is narrower with analog .

Just to be clear here its digital that has the limited bandwidth compared to analog. Typical LP record/playback has bandwidth to 30KHz on a routine basis, often much higher.

Agreed that bandwidth is important for phase accuracy. My point was and is, that most all this phase information, once you process the signal from the mic, is shot to hell. Its gabage already, and bandwith does not account for simple RC phase shifts and transformers and tape and cutter filters and mixers and lp RIAA reconstruction nor CD bandwidth issues, in other words, while I dont disagree wi th your theory the reality is phase is shot to hell with original analog and redbook anyway. There were mixers out there with 3micro second slew rates by the time the signal trhaveled all the wy through them, and with tape bounces and everything else. It still "sounds" good. Your widebandwith amps are great as far as hopefully reducing the odd distortions and slew issues to a point. But my experience shows widebandwidth not to be a solution for real world audio signals. Don't get me wrong, I aint sayin its not good to have bandwidth, just it can be overkill, that all. Audio and ears are bandwidth limited, but as to IMD, as I have said numerous times, really, if you look at what just two tones look like in those stereophile test reports, imagine 50 or so occuring at once in real, bandwidth limited, recorded music...they are there right from the master mixers console and just get worse and somehow, we still get "sound"..that part has always amazed me.

Not all recordings are limited bandwidth. Some are some aren't. (RCA did a recording of Also Sprach Zarathustra with Riener that had plenty of HF bandwidth, even though it was done in 1954. It was recorded at 30 IPS. I have some of the mic preamps that they used for that recording, and they have bandwidth to 50KHz.) However, what it sounds like you don't realize is the group delay artifacts show up if there are bandwidth limitations, regardless of signal content. Its just something that will happen. So it happens on the record side and also in playback. The better recordings have wider bandwidth, and if played back on a better system, will demonstate this in that the result will be a more spacious presentation with more impact. This is because phase has a lot to do with the 3D effect of soundstage. If HF phase information is lost or foreshortened, the soundstage will be diminished. So it does not matter the source; the bandwidth helps even if its only in playback.

 

[opus111]

Just to be clear here its digital that has the limited bandwidth compared to analog. Typical LP record/playback has bandwidth to 30KHz on a routine basis, often much higher.

Yes, as I explained in a reply to mep, I wasn't speaking of the bandwidth of the audio, rather the bandwidth necessary for the signal coming out of the DAC, prior to low-pass filtering. That signal is much wider bandwidth than the audio and hence its much more difficult to maintain sufficient dynamic range at this point in the signal chain. However, your comments apply only to RBCD, hires digital has no trouble exceeding 30kHz audio bandwidth.

 

[Atmasphere]

LP does not either It seems to be other things in the signal chain that limit bandwidth.

 

[DACMan]

I would disagree that analog really has more bandwidth than digital......

Sure, a really well recorded LP can theoretically have SOME information up to 40 kHz or so, but there's more to it than that. First, let's remember that RIAA curve - where the high frequencies are boosted... Don't try and get a 20 Khz cymbal crash onto that record at 0 dB, or at -20 dB.... you'll overload something awful. If you don't overload the amps, you;'ll either melt the lathe cutter head, or bounce the needle right off the record. And, since you bring it up, exactly how much phase shift does that inductive electromagnetically-driven cutting lathe have? And how about the playback cartridge? To suggest that a record can reproduce ultrasonic frequencies with accurate phase and amplitude characteristics is rather a stretch.

Now, how about digital? Well, a CD can only store information up to about 20 kHz - but at least it can do it at 0 dB (or close to it). But we can go up to 24/96 - and now we can record up to 44 kHz or so. Not enough? So, let's go to 24/192.... now we can go to 90 kHz or so. The point is that we can always go up to a higher sample rate - whereas records are pretty well maxxed out (there hasn't been an improvement in the basic technology of vinyl in a very long time).

Maybe we could record an analog recording in a groove, cut at 10,000 RPM in the surface of a glass platter plated with diamond-like-carbon - but I don't see anybody doing it.

While we're on the subject, what exactly IS the phase-response of a microphone? How many degrees of phase shift at 10 kHz does a "good" microphone have? For that matter, what is the total THD of a cutting lathe + vinyl + cartridge + phono preamp signal chain? I'm betting it's nowhere near the 0.03% or so of a decent A/D and D/A.

Honestly, I've heard some very good vinyl systems, and some of them sound DIFFERENT than a good digital system, but I'm simply not convinced that they are BETTER - or more accurate, or lower noise, or lower distortion, or anything else *I* use as a metric of "sound quality".

 

[opus111]

For that matter, what is the total THD of a cutting lathe + vinyl + cartridge + phono preamp signal chain? I'm betting it's nowhere near the 0.03% or so of a decent A/D and D/A.

THD is something of a red herring in audio, seeing as it barely correlates with SQ. 0.03% THD is fairly poor for an A/D and D/A these days but plenty good enough for decent sound.

 

[DACMan]

I agree with you in basic principle - but I think you are overstating the claim.

First off, we DO have a defined maximum bandwidth - and that is the limit of human hearing. We don't have to consider 1.2 mHz noise because we can't hear it - so we can call our "bandwidth of interest" the limits of human hearing (even though we may disagree slightly at the precise numbers involved). Second, there are indeed different types of noise, with different characters. I would even "give you" that analog noise (in the form of white noise hiss) is relatively innocuous whereas at least some forms of digital noise (especially some data correlated types) sound quite discordant. So, I think we both probably agree that we would rather hear a nice smooth hiss noise floor at -70 dB than a ragged, jagged digital one at the same -70 dB.

BUT, that's not what we're comparing. What we're really comparing (when we compare a quiet vinyl pressing to a 24 bit digital recording) is a nice smooth -70 dB analog noise floor (if we're lucky) with a nasty, awful sounding digital noise floor at -130 dB. (If you play a 24 bit recording so the loudest points are approaching the noise level of a fighter jet taking off - and absolutely threatening your hearing - the noise floor will be somewhere below the sound of the blood rushing through the veins in your leg.)

I think what you're really talking about when you talk about "the dynamic noise floor" is the distortion floor..... the fact that the, let's say, 0.05% "THD+N" in the analog recording is really harmonic distortion plus innocuous white noise, while the 0.003% "THD+N" coming from that DAC is really "digital error" - which isn't harmonically related to the content, AND isn't innocuous white noise either. If that's what we're really talking about, then I would agree that they aren't comparable..... and that, in order to sound good, the digital source had better have a total signal error (whatever you want to call it) which is inaudible since, if it is audible, it will probably be much more discordant than the "error" in the analog signal.... This certainly does suggest that, in order to sound really good, we must require significantly lower distortion figures from a digital audio reproduction. (Luckily, it isn't that difficult to achieve this.)

I would, however, disagree that this is difficult to measure. That "THD+N" spec is really just the total error in the signal - the difference between the original signal and the new signal. It is usually calculated by subtracting the input signal from the output signal (appropriately scaled) - and so the result includes any and all digital error, noise, or whatever else you want to call it. So, turn that number into dB and you have the "dynamic floor of everything that doesn't belong there". If you do the math, for a good digital system the result should be very near the calculated "digital S/N ratio" and VERY small.

I personally have another theory altogether....

Virtually all DACs use digital oversampling filters, which do measurably alter the time response of the output (when transients are involved). I personally suspect that THIS is usually what you are hearing as a difference. (In other words the difference is the audible difference between all the high frequency and phase anomalies caused by cutting lathes, and vinyl, and phono cartridges in the analog signal chain vs the frequency and phase anomalies and timing errors caused by various types of filters - digital and otherwise - in the digital signal chain.)

Yep - this is most probably the single metric that correlates best with SQ. But its not the noise floor as traditionally measured - i.e. with nothing playing. Its the dynamic noise floor while music is being reproduced.

Its only easier and cheaper to have a very low noise floor with digital if we adopt the traditional way of measuring noise floor - with zero stimulus. If we adopt the method that matters, then its easier to do this with analog simply because the bandwidth is narrower with analog and there are no aggressor sources (clocks) either. The problem with most digital kit currently available is that it would not measure well against analog in terms of its dynamic noise floor but as yet there's no standardized measurement to check this. We still have to listen.