Differences in sound: DAC vs. DAC + Pre-amplifier

[barrows]

OK, here is my take.  Please feel free to disagree with me, but do not tell me I am wrong.

 

I first tried no preamp, with a Buffalo DAC (DIY) using the first generation "Legato" I/V stage, and the preamp was a highly capable Ayre K-5xe-MP.  The results were:

 

With preamp, higher noise floor, a little less transparency, but better dynamics and fuller, more robust sound.  I preferred with preamp, but wanted to figure out DAC direct as it was clear there was an advantage in transparency with DAC direct.

 

Then, the designer of the Legato output stage changed it, tripling its output current capability.  I now tried DAC direct vs. preamp again:

 

DAC direct: lower noise floor, better detail retrieval, and now, no loss of dynamics either.  So now that the DAC was driving the input stage of the amplifier optimally, there was only loss by adding the preamp.

 

After about 3 weeks of testing, i sold my Ayre preamp, and have not looked back.

 

It is clear to me that DAC direct, for those who do not need source switching. results in the best fidelity to the source, as long as the DAC's output stage is capable of driving the amplifier input stage robustly.  Additionally, most DACs have the capability to drive most amplifier input stages entirely adequately.  Look for a DAC with low output impedance (less than 100 ohms is nice, but up to a couple hindered ohms is going to be OK for most amps) and enough output voltage.  Most IC opamps in DAC output stages can drive amps directly with no problem.  Occasionally there might be a DAC with a bit weaker output stage, but these are rare.

 

I would advise that taking advice direct from manufacturers who sell preamps might not be the best way to make decisions on this, as those manufacturers have an interest in selling preamps.  One really needs to listen in their own system, to the DAC, Preamp, and amplifier one is considering, to decide for sure.  And, definitely do not take advice from "old skool" audiophiles who just claim that any serious audio system must include a pre amp (for no good technical reason), this is a quaint and antiquated notion which has no basis in reality.

 

There is one addition i would make here:  I have heard the claim that adding a preamp can reduce the high frequency noise from the DAC, leading to "blacker" backgrounds.  This supposition relies on the idea that preamps typically will have low pass filters on their inputs, to avoid RF getting into the circuitry.  OK, this is possible, but a couple of things to consider if you think you are hearing this:  DACs all have low pass filters on their outputs, and well designed DACs should not have a lot of high frequency noise on their outputs.  Power amplifiers also usually have a low pass filter right at their input to avoid RF getting into the amp.  My conclusion is that with competently designed DACs and Amplifiers, there should not be any noise problem going DAC direct which would benefit from adding a preamp as an additional filter element between DAC and Amp.  But, there are probably cases of poorly designed DACs, and/or poorly designed amplifiers where adding a preamp actually helps because of this low pass filter effect.  If this happened to me though, i would want to determine what component was poorly designed and allowing this to be the case.

[barrows]

3 minutes ago, craighartley said:

Does your DAC have a volume control, or are you using fixed output and something like HQPlayer to control volume?

I have more than one DAC, some have volume controls, and sometimes I may control volume in software.  

 

So, on volume controls:  Digital Volume control, in software on a computer, or in in hardware on most DAC chips, is generally going to perform better than most analog volume control implementations.  As long as one does not need huge amounts of attenuation (more than -40 dB or so).  If one is listening at very low levels, small details will be below the noise floor anyway, so at high levels of attenuation one is not missing anything.  Despite many audiophile myths to the contrary, there is nothing about digital volume control when well implemented (at 32 bits or more, with a properly matched system gain structure).

 

What a lot of people seem to ignore in the discussion of volume controls is the analog volume controls are far from "perfect", they add noise and distortion to the signal of their own.  Digital volume control is actually more transparent than analog within it s limits:

 

Consider a 32 bit volume control in the digital domain, and a 24 bit source file:

 

32-24= 8 bits, so with a 32 bit volume control one has 8 spare bits to use with no impact on resolution.

 

8*6=48, so one has -48 dB of volume reduction before any resolution (of our 24 bit source file) has been lost, this means there is no problem here!

 

Then consider that most volume controls in software are running at 64 bits!

 

Then consider that actual i room dynamic range of a system is never better than 16 bits...  So in our above example there are actually 16 bits of resolution we can lose without any audible consequences, so we really have 16*6=96, -96 dB of reductio possible with our 32 bit volume control without any impact on real world resolution in room.

 

One can see that there is no "problem" with digital volume control. 

[barrows]

16 minutes ago, craighartley said:

Thanks for this, which is very interesting. I'd like to use digital volume control, but the problem I've found is that software (eg Roon with HQPlayer) limiting of maximum volume is unreliable, so my speakers are laid vulnerable to potential damage.

I have not found that to be the case with a dedicated machine as the server (not on Windows though, some have reported that Windows can be a problem for this).  With ROON running on Mac OSX or Linux I have never had a volume control "incident".  I have only used HQPlayer briefly, so cannot reliably comment on that, but Jussi (miska) seems to believe HQP volume control is stable on Linux.

If one is using a computer for general purpose and running a software player on it at the same time, that could result in problems I guess...

[barrows]

14 minutes ago, craighartley said:

Thanks for this. I do have a dedicated server (Roon on a Mac Mini streaming to HQPLayer on a dedicated machine running the HQPlayer OS), and I'll have another try. My problem occurred when I changed some settings in HQPlayer and (without me noticing) this wiped the volume limit I had set in Roon.

Yeah, there is no doubt that one has to be a bit careful.  I always have a "standby" function on my amps which is easily accessible, and mutes the output.  So if I ma doing anything which might cause problems, I ma sure to mute the amp first, then I double check everything as a matter of course, then un-mute the amp.  Of course one can make a mistake with an analog volume control as well.  

[barrows]

7 hours ago, ro7939 said:

Yes, as someone stated, one can err Re. an analog VC.  The difference is that with many preamps, a visual glance can confirm the volume setting, whereas in the digital realm some invisible software setting can be off, and take a lot longer to find and fix than a glance across the room.  

 

Well this really depends on set up...  I would submit that many people only have to glance at a tablet in their hands when using digital volume control, and not a small screen across the room.

Digital volume control can work with DSD player, as HQPlayer does it, and so does the ESS chip when playing back DSD.  And i would not claim either approach turns it into "PCM" although the debate would be on semantics.  Of course if one wants DSD to playback with absolutely no processing, then an analog volume control in the DAC will be required, as is done in the newer Bricasti DACs like the M21,  Which sounds fantastic direct into an amplifier (especially a Bricasti amplifier).

[barrows]

3 hours ago, The Computer Audiophile said:

Technically a preamp could improve the signal to noise ratio with a powerful output stage, correct?
 

That’s a serious question, no snark. 

Chris that would only be relevant if one used a higher level during playback:  So, say we made a preamp capable of outputting 10 volts, and then based our SNR test off of full output.  Well yes, then the SNR spec could look better.  But in use we woudl have no better performance, because we would not be using the preamp at 10 volts output.  So, in actual use we probably are only runinng the preamp at a level which produces 2 V output, where the SNR would be much lower.

 

As SNR is a ratio: signal/noise, where the "noise" element is basically a constant (noise floor), the only relevance to performance in actual playback is the SNR at the level one uses during playback.  Inflated numbers based on high playback levels (which are never actually used) is misleading as a "spec" or measurement.

 

We can then see how a adding a preamp degrades performance: no matter what, any preamp (even a passive) will add to the noise of a system.  All noise elements are additive.  So we see from our consideration above, that adding a preamp, at the least, adds more noise, and there fore reduces SNR in use.  Look at the ratio, our amp always requires the same level input foe our listening level, so "S" is constant, and "N" is increased, back to our ratio Signal/Noise one can see that the actual SNR i use will be lower becasayuse of the additional noise of the preamp.,

 

 

[barrows]

11 minutes ago, ro7939 said:

A state of the art analog preamp is positively always required for maximum system flexibility, not a luxury nor performance impediment.  This is always true

The above is false.  the statement: "nor a system impediment" is what I am referring to.

 

A preamp in the signal path does this:

 

Losses of fidelity from:

 

Preamp input switching devices

Preamp gain stage

Preamp analog volume control

an additional interconnect

At least two additional connections per channel (XLR or RCA jacks and plugs, plus their soldier joints to wires)

 

All of the above mentioned items are lossy, noise and distortion sources which can only add noise and distortion and reduce fidelity.  Remember, that distortions and noise are additive throughout the system.  Removing the preamp (and this applies to any preamp, no matter how good) removes these sources of noise and distortion.

[barrows]

30 minutes ago, Kimo said:

Benchmark will tell you that you will get better performance using their DAC 3 at a fixed setting and inserting their preamp.

 

I would expect a manufacturer who manufactures and sells a preamp to say this.

 

31 minutes ago, Kimo said:

ESS will tell you analog volume controls are better than digital.  I think there is a paper online from them with measurements.

 

You can find Martin Mallinson's video from RMAF on Youtube.  Be careful how you interpret it though. if you pay attention, what he is saying is that analog volume control has an advantage IF you need to use tons of attenuation.  A digital volume control is demonstrably more transparent than an analog one as long as it is not used at very high levels of attenuation.  For example, the ESS chip's volume control runs at 32 bits, so, with a 24 bit source file you have 8 bits extra for volume before you lose ANY resolution.  8 bits means that you can have attenuation up to (8*6dB) -48 dB.  The other potential drawback with digital volume control is thta it does not reduce the noise floor as the volume is lowered, again, this is a non-issue unless you use very high amounts of attenuation, and have a noisy source, both are non-issues in the vast majorities of real world systems.

[barrows]

14 minutes ago, Kimo said:

Call Benchmark and they will describe to you in detail why they achieve better performance with a preamp.  I don't think that they will pitch their preamp.

 

I have no need of any convincing, I have been running amp direct for a few years now, but before I ditched the preamp, I did extensive testing, and only sold my preamp (a rather fine Ayre K5-xeMP) when I was absolutely sure that amp direct was performing better.

[barrows]

@motberg,  Are you aware that LDR volume controls add lots of distortion?  

[barrows]

4 hours ago, bodiebill said:

I had the same experience with the Lightspeed Attenuator, and now with the DIY Stereo Coffee LDR preamp. Closest thing to no preamp, and I love the sound.

From a technical perspective, this is just not true.  For example, a simple passive attenuator like a Goldpoint will be much more transparent than any LDR could ever hope to be.

 

It is OK if you like the sound of an LDR, but please do not make claims about it being transparent, as this is just not the case.  LDRs add large levels of distortion, orders of magnitude greater than that of anty good preamp with simple switched resistor volume control.

 

sandyK has provided the proof.  And this has also been shown every time someone measures an LDR based preamp.

 

Here is some more info on LDR based volume controls:

 

https://neurochrome.com/pages/tortuga-audio-ldr3

 

 

[barrows]

@Blake, agreed.  BTW, I know Jim, he is a really nice man.

 

Of course, what I would say, is that if one prefers the sound with a preamp in the chain, there is probably a "problem" in the system somewhere (given a set up where the DAC does have an appropriate output stage for driving the amp), which the preamp's additional noise and distortion are covering up.  This is kind of like the person who finds their system too bright, and decides to tamp down the high frequencies a bit by using a highly capacitive cable: the problem with this approach is that it is lossy, "fixing" one problem but adding another.  Such an approach to system building results in less and less fidelity as changes are made by adding more losses to cover up problems.

 

I am interested in hi fidelity playback, and not playback which is intentionally colored, or obscured.  Other's may prefer a colored system, and that is fine with me as long as they recognize that and understand that the colored sound is what they are getting.

 

Personally, I find that systems with a lot of their own colorations can sometimes sound "good" (think SET amp with simple music), but they tend to make different music sound the same, rather than maximizing the differences between different music, and often the reduction of small details from his approach can obscure the greater musical meaning.

[barrows]

31 minutes ago, Blake said:

 

I don't know that all preamp designers are intentionally coloring the sound, in fact I am sure that is not the case with some preamps, but of course, there are probably plenty of designs out there with intentional colorations.  But again, you are speaking from a relative point of view here, right?  Your current system for example is not 100% transparent, you are just attempting to get as close as possible to a theoretical ideal of transparency.

 

Sorry, i guess I was not clear.  In no way do i mean to suggest that those who design preamps are trying to add colorations.  What I am saying is that any component addition (like a preamp) can only result in loss of fidelity; by adding a preamp, where one was not present before, the only possible result is that doing so adds more distortion and noise than there was before.  If one then prefers the sound with that preamp in place, what that person is preferring is the coloration added by that preamp's noise and distortion profile.

 

And, of course no system is 100 % transparent to the recording (electronics can get very close these days, but certainly not speakers!), but to me it is the goal of a good system to be as transparent to the recording as is possible.  Adding unnecessary components to the system is counter-productive to that goal.

[barrows]

6 minutes ago, craighartley said:

I so want to take my expensive transformer-based ‘pre-amp’ out of my system, but if I do that I get a prominent click out of the speakers on the start of music playing, which is negligible with the ’pre’ inserted. Miska has suggested that this may be due to DC offset from the DAC analogue output, but it disappears if I stream from Roon direct (via MicroRendu in RoonReady mode) rather than Roon>HQPlayer>NAA. 
I know my amps have too much gain (39db) for the software volume solution to be ideal, but it’s the click that prevents me pursuing it. 

Wow Craig, 39 dB is really high and quite unusual gain for a power amplifier.  Have you ever contacted the manufacturer and asked if the gain could be lowered?  That much gain is probably also make the amp(s) noisier than they could be.

 

Of course your "click" problem is something other than preamp related.  If i had that "click" problem i would track down the ideal solution for the problem, rather than covering it up by adding a transformer based pre/volume control.

[barrows]

@craighartley, cool looking amps.  but I ma confused, looks like the Leema is an integrated with its own volume control?  Why not just use the amp's VC?

[barrows]

28 minutes ago, motberg said:

But I think to use software volume control, you need modify the source file digitally. So isn't the choice digital manipulation of the source vs. analog manipulation of the DAC output?

Yes, but software volume controls typically run at bit rates much higher than that of the source file, so there is no loss.  For example, with ROON volume control, it works at 64 bits, so if your file is even 32 bits, you retain all the bits of the original file with no change with attenuation of up to 32 bits.  32 bits equates to about -192 dB, so as you can see, there is no loss of information.

[barrows]

1 hour ago, Kimo said:

Dynamic range is specified as 138 db.

 

I am curious as to what this refers to?  DR of what exactly, as this number is at the extreme end of what is possible in the analog realm (and only for line level, amplifiers, not at all...)

[barrows]

4 hours ago, CG said:

Distortions can cancel

Hey CG, I certainly get what you are saying, but just as easily they can by additive as well, as the operative word here is "can".  As far as the rest goes, totally agree on impedance, etc.  But there are plenty of DACs these days with the exact same type of output buffer as one would find in the output of a preamp, with no difference.  These DACs have the exact same ability to drive the input stage amplifiers.  I would suggest one is getting very, very, lucky if they have preamp which just happens to actually cancel the distortion of their DAC, I mean, are we not talking about a very small possibility here?

There are of course some very unusual exceptions, but for the most part, DACs have output z at 100 ohms and often less, and amplifiers these days are usually over 50 kOhms input impedance, so no problem.  Now one could, if they really tried hard, manage to find a DAC/Amp combination thta is problematic, but it woudl be a rare exception rather than the rule.

[barrows]

Thanks CG, i realize it is "one circuit".  Point being that most DACs have enough drive to drive the inout stage of most amplifiers perfectly.  And, as I mentioned, one can find exceptions, but one does have to work a little bit to do so.

 

Here is my example, of my direct experience back when I used a preamp:

 

I had a DAC with a discrete IV stage, followed by a single small signal MOSFET for each signal leg (so four small signal MOSFETs to drive the output).  This DAC did not do well enough going amp direct (my preamp at the time was an Ayre K5-xeMP).  I noted lower noise floor and better low level detail retrieval amp direct, but there was a problem with dynamics and presence-this were a bit wimpy, if you will.  Then the designer of this DAC changed the output stage design, to where the output buffer used 3 MOSFETs per signal leg (12 in total for balanced stereo output).  When I made this change to the output stage I tested DAC direct again (amplifier at the time was a Pass Labs X150.5), with the output stage change, now the DAC direct approach was clearly better in all parameters, lower noise floor, better detail retrieval, and perfectly fine dynamically.  I tested this to be sure, over long term as well, and eventually sold the preamp.

Since then I am totally committed to amp direct and have no need of a preamp.  I just make sure my amps are not to crazy difficult to drive (easy to do these days), and make sure the output stage of any DACs I use have plenty of current capability (which is easy to do as well given the nature of current IC OPAs).

 

As mentioned, one can find combos which will have compromises, but those are pretty rare these days, requiring a DAC with a rather "weak" output stages and an amplifier which is a bit difficult to drive (less likely to find?).

 

An (possibly) even better approach, which I am hoping will become more popular for consumer audio, could be to lower the amplifiers gain, and have higher voltage output from the DAC (as well as high current capability), as this could make the connection less subject to interferences (noise pick up) and cable differences.  Such is in pro audio where line level voltages are typically higher.  this approach is used by Benchmark in their gear.

[barrows]

@CG, thanks for the links.  I do not have time now, but will read up later.  I try to keep my system as simple as possible, and use just a 1 m pair of interconnects (Iconoclast with low capacitance and inductance) between DAC and Amp.

 

My main point being, though, that these same problems you mention can occur between a preamp and amp as well.  Having a preamp in the system is no guarantee of avoiding what you are mentioning.  On most DACs and preamps I look at these days, the output buffer is often the same (an IC OPA of some variety, with a lowish value series resistor).  Given that the output stages of either DAC or preamp are often the same, the notion that there is actually a "problem" going DAC direct to amp is just not valid (given certain rare exceptions).

 

I guess the real "problem" here is that there really are no standards for home audio in terms of output to input of various components, impedances vary quite a bit from product to product, as do voltages, as do the stabilities of various circuit topologies.  And audiophiles are constantly mixing and matching various components and brands.  

[barrows]

16 hours ago, sandyk said:

 

 Actually, many Power Amplifiers these days have an Input Impedance of 10K to 20K which usually results in improved S/N as well as contributing to a lower DC offset.  e.g. Audio Power Amplifier Design Handbook 5th ed - D. Self.

I am not aware of "many" high end power amps with input impedances of 10-20 K?  Maybe a few, but not many, and certainly a minority.  A quick perusal of amps measured by Stereophile, taken at random showed this:

 

Parasound: 90K

CH Precision: 90K

McIntosh: 40K

Luxman: 300K

Constellation: 20K (specified as 200K)

Sim Audio: 32K

Pass Labs: 100K

PS Audio: 96K

Bricasti: 170K

Ayre > 1M (JA has difficulties measuring input impedances above 1M)

 

All are for balanced connection, which I consider the norm these days, as most DACs we are talking about here have balanced outputs.  So we see a couple of amps with lower numbers, but most are up there.  Certainly even the Constellation (20K measured) is easy enough to drive by a typical DAC with 100 ohms output impedance, as long as the output stage has plenty of current capability.  Most of my DACs have output impedances of around 33-100 ohms (per phase that is).

[barrows]

1 hour ago, sandyk said:

 

Hi Barrows

These are more recent higher performance designs as per the research by Douglas Self, and variations used in the published designs from Silicon Chip (Au.) magazine etc. , which has published designs with a typical distortion of .0006% ,and a further improvement when using matched devices in the front end. The lower input impedances also markedly reduce the DC offset at the output due to the bias currents of the LTP ,  meaning that no adjustment is normally needed to reduce this provided that the LTP ( Differential Pair) devices are reasonably close in HFE. 

Please check your PMs.

Regards

Alex

Sure, but is not this discussion about real world amplifiers people generally own and can purchase, not theoretical designs, or obscure DIY efforts only available to a few hobbyists?

[barrows]

@sandyk,

 

Yes, but that does not make this statement any closer to being accurate:

 

"many Power Amplifiers these days have an Input Impedance of 10K to 20K"

 

I would also suggest that for the purpose of this discussion, even an amp with 10K input impedance can be driven properly by many, many DACs available these days.

[barrows]

6 minutes ago, sandyk said:

 

 I would put it to you that the amplifiers that are in common use by the vast majority of consumers do not use balanced topology, as do the vast majority of affordable commercial DACs, some of which MAY provide this facility , but is often not used by typical consumers.

You appear to be taking an elitist position here, as there is very little, if any, advantage to Balanced Topology in most domestic situations,where relatively short interconnects and speaker leads are used.

 

Alex, I just did not understand what you keep posting these things?  I mean, the "vast majority" of consumers are probably not using a Doug Self designed amp.  But why even mention the average "consumer", this is an Audiophile Forum, read and contributed to by Audiophiles; I do not think the "vast majority" of consumers are reading this, and most of them are listening to cheapo bluetooth speaker via Alexa and are not concerned whether they should use a preamp or not...  There are many, many, many amplifiers which do feature balanced input stage topology, it is not uncommon at all.  Now balanced all the way through to the output stage is less so, but that has nothing to do with the DAC/Amp interface anyway.  Contemporary dual IC opamps also make implementing a balanced input stage relative easy and affordably. 

As for DACs, again, there are tons of truly balanced DACs out there, and not at high prices either.  For just one example, I am playing around with a Topping D-90 right now, it is well under $1K USD.  There are many other DACs <$1K with true balanced outputs.  Especially with a DAC it is important to have balanced outputs, as almost all DAC chips used these days have balanced outputs, and it actually takes more circuitry (if one is going to do it right via balanced to single ended conversion stage) to convert the output to single ended than it does to have balanced output.

Nothing elitist about this, balanced topology is relatively common place, and easily produces real benefits for the output stage of a DAC (a simple example of this advantage is the cancellation of power supply noise).

 

Anyway, not really sure what these posts have to do with the interfacing between DAC-Amp and/or preamp?

[barrows]

3 hours ago, sandyk said:

 You appear to be under the impression that the majority of members are using balanced topology.

Nope, I would never assume anything like this.

 

3 hours ago, sandyk said:

3 hours ago, sandyk said:

I would suggest that they are only owned by a relatively small percentage , probably no more than 20% of our members and readers, as they are simply not necessary in a typical domestic situation.

 

And such a suggestion just reflects more about what you think and your own approach, rather than anyone else's.

 

3 hours ago, sandyk said:

Digital Attenuation, the quality of which will vary depending on the way it is implemented

Nope, any notion of digital attenuation being a "problem" only exists in the past.  With the possible exception of a poorly put together system with absolutely terrible gain matching-point being that in any system with bad gain matching, there are considerable benefits to fixing the real problem (the poor gain structure) regardless if one uses analog or digital attenuation.  Higher levels of analog attenuation come at a price, higher noise levels due to resistor noise.

 

In my opinion, you are living in the long past: single ended DAC outputs are a really poor idea in this time of balanced output DAC chips, and compromise the possible performance of any DAC.  And, digital volume control is not compromised in its current implementation in any contemporary DAC of which I am aware (32 bit volume control and higher is transparent).

 

If one is interested in high fidelity to the source recording, the only valid reason for having a preamp is if one has multiple sources, including analog ones, otherwise going DAC direct has a distinct advantage in fidelity to the source.