Opinions on the role of the preamp in a modern single source system

[barrows]

Hi folks: Here is my take. I run a DIY DAC using the digital volume control of an ESS 9018 chip. Now before folks get their panties all in bunch regarding digital volume controls lets look at the facts with how the ESS' works. It controls volume using 48 bits of resolution, so there is no truncation of actual data, even with 32 bit files. Additionally, ESS specifies all noise products (in this case that would be quantisation noise) at below -170 dB, so we can be sure this VC is going to be as transparent as is possible. My amplifier is a PASS X150.5, this amp has 15K input impedance per phase, and produces full output with 1.75V of input. My DAC's gain is set such that it reaches 1.9V of output at 0 dB, and the output impendace is a littel high, at 180 ohms per phase, but still well within the 10x margin often recommended. So between DAC and amp I have a reasonable impedance difference, and a very good voltage gain match. The other factor to beware of is current drive.

When I first tried removing my Ayre K5xe-MP preamp, and driving the amp direct, I did notice improved transparency, and a slightly lower noise floor (my DACs noise floor is exceptionally low, even compared to the quiet Ayre pre), but, I did experience reduced dynamics, aspecially micro dynamics. My conclusion was that the (relatively simple discrete MOSFET) output stage of the DAC did not have enough current drive to produce optimum results with my amplifier. Then the designer of my DAC released a new version of the ouput stage circuit. This version paralleled 3 MOSFETS per phase, where previously there was only a single MOSFET, increasing the current drive. With the new version, listening tests confirmed that going amp direct was now at least equal in dynamics to having the Ayre pre in place, and transparency was improved.

Now I run exclusively in DAC direct mode. The answer is, we have to listen in our systems to decide if this approach is going to work for us, but a technical understanding of what makes for these differences is also very helpful. If the DAC has adequate current drive, adequate V out, and a good impedance match, and the volume control implementation is well done, the only reason (with a single source system) to prefer a preamp in the chain woudl be to add a certain desireable coloration to the system. I am not saying that there is anything wrong with adding color if that is someone prefers, but that approach is not for me.

Note that AudioExplorations' Medea+ DAC has a super beefy output stage, which will likely drive any amplifier as well as any preamp in existence. Take a look under the hood of the Medea and you will see what I mean.

 

[mojave]

I voted before I saw the results and was surprised that I am the only one that voted "No (never really tried with a preamp)". I have been a computer as source only guy since I bought my first IBM computer in 1989. Of course, the computer as source as improved a lot since then.

I currently use 64-bit digital volume attentuation, my output impedence is 75 Ohms, and my SS monoblock amps' input impedence is 10K Ohms for a 133x ratio. Both my DAC and amps are balanced.

I went to RMAF a few years ago and heard two-channel systems for the first time. My system is relatively inexpensive (and has changed since RMAF) but I was surprised that I couldn't find a system that I felt had the same level of detail as my system. There were many other things that other systems did better, but detail wasn't one of them IMO. I have often wondered if it wasn't because my system was so simple that I felt it had better detail.

Doesn't the impedance mismatch issue only occur with those who use tube amplifiers. I thought about a getting a tube buffer, which does volume attenuation and impedance changing but no gain, but was told it wouldn't do anything with SS amps except color the sound with the "tube quality." On the Rane website it says,

Impedance matching went out with vacuum tubes, Edsels and beehive hairdos. Modern transistor and op-amp stages do not require impedance matching. If done, impedance matching degrades audio performance.

Modern solid-state devices transfer voltage between products, not power. Optimum power transfer requires impedance matching. Optimum voltage transfer does not. Today's products have high input impedances and low output impedances. These are compatible with each other. Low impedance output stages drive high impedance input stages. This way, there is no loading, or signal loss, between stages. No longer concerned about the transfer of power, today's low output/high input impedances allow the almost lossless transfer of signal voltages.

What then, does impedance matching have to do with unity gain? Well, it shouldn't have anything to do with it. But because of different manufacturer's definitions, it is one way (brute force) of correcting gain discrepancies between products. Impedance matching introduces a 6 dB pad between units.

I guess my question is, "Does a preamp benefit a tube amplifier more than a SS amp?"

Another consideration is that some DAC's, especially pro audio devices, have the D/A output stage permanently at max. My previous DAC had this issue and had a slight hiss when the ear was all the way up to the tweeter. Digital attenuation did not remove the hiss. I was told by the manufacturer that it was expected that attenuation would occur after the output stage. Even a few dB of attenuation would remove the hiss. Since those using preamps probably seldom listen at maximum volume, the post DAC attenuation could be removing some noise.

Finally, I also wonder if some preamps attenuate the line level signal before it amplifies it as part of the impedance matching or impedance changing process? If so, this could change the noise floor and cause the amplified signal to actually sound "better" with the preamp than without it.

 

[dallasjustice]

Hi folks: Here is my take. I run a DIY DAC using the digital volume control of an ESS 9018 chip. Now before folks get their panties all in bunch regarding digital volume controls lets look at the facts with how the ESS' works. It controls volume using 48 bits of resolution, so there is no truncation of actual data, even with 32 bit files. Additionally, ESS specifies all noise products (in this case that would be quantisation noise) at below -170 dB, so we can be sure this VC is going to be as transparent as is possible. My amplifier is a PASS X150.5, this amp has 15K input impedance per phase, and produces full output with 1.75V of input. My DAC's gain is set such that it reaches 1.9V of output at 0 dB, and the output impendace is a littel high, at 180 ohms per phase, but still well within the 10x margin often recommended. So between DAC and amp I have a reasonable impedance difference, and a very good voltage gain match. The other factor to beware of is current drive.

When I first tried removing my Ayre K5xe-MP preamp, and driving the amp direct, I did notice improved transparency, and a slightly lower noise floor (my DACs noise floor is exceptionally low, even compared to the quiet Ayre pre), but, I did experience reduced dynamics, aspecially micro dynamics. My conclusion was that the (relatively simple discrete MOSFET) output stage of the DAC did not have enough current drive to produce optimum results with my amplifier. Then the designer of my DAC released a new version of the ouput stage circuit. This version paralleled 3 MOSFETS per phase, where previously there was only a single MOSFET, increasing the current drive. With the new version, listening tests confirmed that going amp direct was now at least equal in dynamics to having the Ayre pre in place, and transparency was improved.

Now I run exclusively in DAC direct mode. The answer is, we have to listen in our systems to decide if this approach is going to work for us, but a technical understanding of what makes for these differences is also very helpful. If the DAC has adequate current drive, adequate V out, and a good impedance match, and the volume control implementation is well done, the only reason (with a single source system) to prefer a preamp in the chain woudl be to add a certain desireable coloration to the system. I am not saying that there is anything wrong with adding color if that is someone prefers, but that approach is not for me.

Note that AudioExplorations' Medea+ DAC has a super beefy output stage, which will likely drive any amplifier as well as any preamp in existence. Take a look under the hood of the Medea and you will see what I mean.

First of all, I am so glad Barrows is posting here. Thanks for joining us!

My DAC and amp are designed to work without a pre-amp. I use a BMC DAC1 to BMC monoblocks. Gain is controlled with a digital signal sent from the DAC directly to the amps. Therefore, gain is controlled inside the amps. I believe this sort of minimalist design heavily contributes to a super low noise floor which is exceptional IME. I can't say that pre-amps don't help is some systems, it's just never been demonstrated to me that they are beneficial in all systems. I believe the assumption should be that less is more, unless proven otherwise.

 

[barrows]

You bring up one interesting point about using a digital volume control: that this way of attenuating the signal will not attenuate the noise of the following analog stages, unlike using a preamp, where, at least, the noise generated by the DACs analog stage(s) will be attenuated, along with the signal, when reducing volume. This is a good reason (and so is truncation of the bit depth) to make sure the voltage output level of ones' DAC is well matched to the input sensitivity of the amplifier, reducing the amount of attenuation required, and rendering the noise (and bit truncation) a non-issue. If one is using a DAC with 6 or more volts of output into a 2 volt amplifier input stage, digital volume control may not be a good way to go. As mentioned by AudioExplorations, Daniel Weiss addresses this in his products by having a course output adjustment in the analog domain on his DACs, allowing for proper matching with the amplifier. With my DIY DAC, it is possible to adjust the gain of the outptu stage to match the input sensitivity of the amplifier, achieving good matching.

 

[AudioExplorations]

This is a good reason (and so is truncation of the bit depth) to make sure the voltage output level of ones' DAC is well matched to the input sensitivity of the amplifier, reducing the amount of attenuation required, and rendering the noise (and bit truncation) a non-issue. If one is using a DAC with 6 or more volts of output into a 2 volt amplifier input stage, digital volume control may not be a good way to go. As mentioned by AudioExplorations, Daniel Weiss addresses this in his products by having a course output adjustment in the analog domain on his DACs, allowing for proper matching with the amplifier. With my DIY DAC, it is possible to adjust the gain of the outptu stage to match the input sensitivity of the amplifier, achieving good matching.

For the reasons mentioned, voltage matching functionality to me is one of the most useful and important characteristics of a DAC or source component. I really struggle to understand why almost none of today's DAC designers implement this in their designs (Weiss products and DIY being the only options I know of today). Adding a properly dithered 32/48-bit digital volume control is almost a freebie as this can be implemented within the DAC chip and then with a robust analogue output stage you enable the flexibility of eliminating the preamp. Seem's almost a no-brainer to me, even for those who use preamps for source switching and those who are single-source "preamp-believers" (majority of the people here), properly matched output voltage means less detrimental analogue volume control attenuation is needed within the preamp.

 

[dallasjustice]

Why stop with the pre-amp? Why not get rid of the amp as well. You certainly could do so with high efficiency horns.

 

[egidius]

My option is wadia s7i into audio consulting silver rock passive into mipa amp also by audio Consulting, it works.

 

[barrows]

Yeah DallasJ. But some of us just cannot listen to horns. I know I prefer classic box speakers myself, and usually ones which are not all that sensitive as well, so I know I'll always need a few watts...
Not that horns, panel speakers, and open baffle designs do not do some things well though, they are just not for me.

 

[MarinJim]

I like a tube preamp in my system because it causes an emotional response for me to the music I listen to. And that, ladies and gentleman, is all I care about.

 

[DonH50]

For the reasons mentioned, voltage matching functionality to me is one of the most useful and important characteristics of a DAC or source component. I really struggle to understand why almost none of today's DAC designers implement this in their designs (Weiss products and DIY being the only options I know of today). Adding a properly dithered 32/48-bit digital volume control is almost a freebie as this can be implemented within the DAC chip and then with a robust analogue output stage you enable the flexibility of eliminating the preamp. Seem's almost a no-brainer to me, even for those who use preamps for source switching and those who are single-source "preamp-believers" (majority of the people here), properly matched output voltage means less detrimental analogue volume control attenuation is needed within the preamp.

I am not sure if by "DAC" you mean the actual DAC chip or the component that conatins the DAC, power supply, interfaces, etc. In any event I suspect most designs do not have extra signal processing in their DAC boxes so if it ain't in the DAC chip itself they ain't gonna' do it. Data converters (ADCs or DACs) achieve their highest preformance near full-scale, so reducing the signal digitally will cost SNR and depending upon how the data reduction is performed will add distortion as well. This means any digital attenuation will reduce the performance of their box. Analog step attenuators add little in the way of "detrimental" things to the signal path so I am not sure why you think a digital implementation would be better than an analog version? I would expect the opposite, at least in this application.

FWIFM - Don

p.s. 32/48 bit resolution does not matter if the DAC chip itself is 16 or 24 bits; to reduce the volume, you have to reduce the range of (digital) signals into the data converter.

 

[barrows]

I am not sure if by "DAC" you mean the actual DAC chip or the component that conatins the DAC, power supply, interfaces, etc. In any event I suspect most designs do not have extra signal processing in their DAC boxes so if it ain't in the DAC chip itself they ain't gonna' do it. Data converters (ADCs or DACs) achieve their highest preformance near full-scale, so reducing the signal digitally will cost SNR and depending upon how the data reduction is performed will add distortion as well. This means any digital attenuation will reduce the performance of their box. Analog step attenuators add little in the way of "detrimental" things to the signal path so I am not sure why you think a digital implementation would be better than an analog version? I would expect the opposite, at least in this application.

FWIFM - Don

p.s. 32/48 bit resolution does not matter if the DAC chip itself is 16 or 24 bits; to reduce the volume, you have to reduce the range of (digital) signals into the data converter.

Don, have you read the thread? I see you are a technical expert here, so clearly you must know the reality of this? Take a DAC chip which has an onboard volume control at 32 bits, this is a common part these days (like a Wolfson 8741). OK, now, there is no real world audio system which will achieve more than 20 bits of S/N in room, in fact, I have not even seen a test result of any DAC (meaning the entire component) by John Atlkinson which achieves better than 20 bits of resolution. So, with our 32 bit volume control we have 12 bits of leeway in attenuation which will not affect the real world resolution of the system. 12 bits is over 60 dB of attenuation, so we have no loss of resolution, even up to 60 dB down. Now in any decently matched system one should not need anywhere near that much attenuation. Even if we consider a 24 bit volume control, we have still have 4 bits, and 24 dB of attenuation before we even have to consider any loss of resolution.
Audio Explorations has pointed out that it is a good idea to make sure that the voltage output of one's DAC is well suited to the sensitivity of the amplifier, so as to not have to use too much digital attenuation, a good point, and one which should be paid attention to by anyone considering going this route.
An example in my system: I use a DIY DAC with an ESS 9018 chip. The ESS volume control actually operates at 48 bits, and the gain of my DAC is set such that 0 dB results in ~1.9 volts output. My Pass amplifier has 26 dB of gain, and reaches full output at ~1.75 volts input. With this set up I use around -22 dB of digital attenuation with contemporary (highly compressed) recordings, and only a few dB of attenuation with typical "audiophile" stuff (like RR's HRx). With a well set up system using digital volume control there is no concern about loss of resolution. Now, S/N ration will drop as attenuation is added, but as long as one does not use a noisy DAC to begin with and does not have a system with way too much gain, this will not present an audible problem either. Indeed, eliminating the preamp actually can reduce the overall noise floor, as it eliminates at least one active stage and the inherent noise and distortion of that circuit. Now, you speculate that using the digital volume control will "probably" introduce distortion, of course this is true, there will be quantisation noise, but the question is how much? Well ESS specs state that all quantisation noise products will be below -170 dB! I think we can agree that these are irrelevant. Additionally, we know that even the best, traditional, volume controls are not without their own problems: the generally accepted "best" approach is a switched resistor ladder, which will also add noise and distortion from both the switches and the resistors themselves, when we add that to the noise/distortion of the preamp circuit we see that preamps are somewhat compromised components (and this is why a really great preamp is hard to design and build, and expensive to purchase)-but it is kind of like a capacitor in the signal path: the best cap is no cap, and the best preamp is no preamp.

Of course, if one implements the system poorly, the DAC direct approach is doomed to failure: a badly designed digital volume control could be a problem, a bad system match could result in the need to use too much digital attenuation, and/or an inadequately designed DAC output stage may not drive the following cable and amplifier input stage adequately; hence those who are interested in going this route need to pay attention to the details.

 

[DonH50]

I have been following the thread only off and on. I have never claimed reading comprehension as one of my strengths. Nor to have all the answers. My expertise does include a variety of data converters but mostly at much higher frequencies. However, I was not addressing the use (or not) of a preamp, just the idea of digital vs. analog signal attenuation.

Digital attenuation I have used in other systems but I am not sure I am following the implementation here. If the DAC itself has 20 bits of SNR, then I am not sure what a 48-bit digital volume control does for you. I must be missing the boat on this one (would not be the first time); perhaps I do not understand how the digital volume control works, or at least how it is implemented in the DACs you are describing. I assumed it would shift (divide, truncate, whatever) the bits so that the lsbs fall off, reducing the overall (digital) signal level into the DAC itself. Say you have a 16-bit DAC and 32-bit volume control; what do the extra 16 bits buy you? The DAC sets the resolution of the system at 16-bits; to reduce the volume, you must reduce the range of bits going into the DAC. Reduce the signal by half (one bit), and you also give up 6 dB in SNR (with respect to the quantization noise floor).

The other schemes I have seen to adjust the output level are to incorporate a multiplying DAC, use another DAC to adjust the signal DAC's reference level, or use a DAC to adjust the gain of an analog gain stage (e.g. biasing a device or analog multiplier cell). I am sure there are others.

I agree that an active stage adds noise and all that, but am not sure why lopping off bits does not? Or that a simple analog attenuator adds more artificats than reducing the digital signal level? I understand the SNR loss might not be audible in the real world, but as I understand it the loss is there nevertheless. Assuming the same bandwidth, SNR goes as roughly 6N and SFDR as 9N in dB where N is the number of bits, when both are referenced to quantization noise (not including other noise sources).

I am confused, yes.

 

[AudioExplorations]

Don, perhaps the following slides from ESS would help - they talk about how a 16bit signal is handled by both a 16bit and 32bit DAC chip, truncation error, signal to noise ratio, and comparison of both to analogue volume control.

ESS - Digital vs. Analog Volume Controls, October 2011

 

[DonH50]

Thanks very much, I think I understand what they are doing. I happen to disagree with some of it, and wonder if they really do have a 32-bit DAC internally, but at least I understand your argument now. I need to research a bit more (no pun intended!) to see how or if they get around the issue of quantization noise at the output since if the output is only 16 (or 24 bits) they still have that to deal with (outside the analog noise floor in the article).

However, unless I missed it, from a performance standpoint the ESS approach makes a digital design that matches an analog, albeit without aging effects, so I am still not sure I understand how analog is so much worse than digital volume control? I only took a quick look and will try to spend more time later. Probably off-line as I think this is off-topic as far as this thread goes.

Thanks - Don

 

[Saturntube]

I have not read the whole 20 pages of the thread (I do have a life) but I have some hands on experience on passive preamps and dacs.

I had an EAD DCS 9000 III DAC wich uses a mixed digital and resistor loaded volume control, looosign only one bit of resolution. I used it for a long time, I also had the smaller brother EAD 7000 III Dac whic is similar without the output satge and volume control, and separate power supply; regardless I liked the EAD 7000 III better connected directly to a resistor loaded volume control... I guess theory only gets us to a certain point and then you have to try the passive and see how it sounds.

About the original question, yes I like the Active preamp better: but only an extremely good active preamp and there is only one or two that I would use! It does add more meat to the bones, so to speak and dinamics and makes the music lively: again a very good preamp, sadly most preamps just damage the signal, and I mean most! In that case I would prefer a passive preamp, passive pramps are hipnotic, detalied, transparent: You must have a passive pre (volume control) in the drawer just to keep checking against your active preamp and see what you are missing, as a "pied a terre".

Now if you have a low efficiency system I would have to admit an active preamp is a must, and a curse!
If you have a high efficiency system you can do wonders with passive preamps and never miss an active preamp!
If you are using only digital, I think there are dacs out there that can put out enough power to drive a high efficency system, and do without even the amps! ja say good bye to those expensive DHT!

 

[barrows]

this is actually:

"(an active preamp) It does add more meat to the bones, so to speak and dinamics and makes the music lively: again a very good preamp, sadly most preamps just damage the signal, and I mean most! In that case I would prefer a passive preamp, passive pramps are hipnotic, detalied, transparent: You must have a passive pre (volume control) in the drawer just to keep checking against your active preamp and see what you are missing, as a "pied a terre"."

Factually incorrect. You really should go back and read the thread! Many here have compared with and without preamp, and the experiences vary greatly. But the fact is simple: If you have a DAC with an output stage which is capable of driving the chosen power amp's input stage properly, there will be no more meat on the bones or better dynamics with a preamp in the chain. That is just a fact. Now, it is clear there are many DACs which do not have an output stage which is as good as the best preamps, but some do. Consider: if there is no loss using a digital volume control, as evidenced above, and the output stage of the DAC is identical to (or better than, either is possible) the output stage of the preamp, then using the preamp can only add noise and distortion to the signal.
Now, I am not claiming that DAC direct is going to always be best, it is a system thing, and some DACs are not good at driving some amplifiers directly, so everyone has to try this in the context of their own system to find the best approach-but I am claiming that with the right DAC and amplifier, DAC direct is superior to adding a preamp, this is non disputable.

 

[Saturntube]

DAC direct is superior to adding a preamp, this is non disputable:

Depending the system....

 

[Phelonious Ponk]

DAC direct is superior to adding a preamp, this is non disputable:

Depending the system....

Yes, I suppose you can put it that way, but it is not some broad systemic synergy that you need. It is not complicated. Barrows put it simply and clearly:

If you have a DAC with an output stage which is capable of driving the chosen power amp's input stage properly, there will be no more meat on the bones or better dynamics with a preamp in the chain.

What might there be? Noise. Distortion. Loss. And you might like it, particularly if you're used to listening to analog. And you might hear it as "more meat on the bones." Enjoy.

Tim

 

[Saturntube]

Impedance needs to b at least 10 times lower on the recieving end, and of course the neeed voltage.

I have been using a passive preamp for a few years now, I do listen to digital and analog, through the same passive general volume control.
Multiamp.

But I have to say the best preamp I listened to did kill the passive, analog or digital: Metaxas Opulence.

I did not believe it myself!!

 

[Phelonious Ponk]

Impedance needs to b at least 10 times lower on the recieving end, and of course the neeed voltage.

I have been using a passive preamp for a few years now, I do listen to digital and analog, through the same passive general volume control.
Multiamp.

But I have to say the best preamp I listened to did kill the passive, analog or digital: Metaxas Opulence.

I did not believe it myself!!

How do you explain it?

Tim