Benchmark introduced the DAC1 in 2002 and it quickly became the best-selling 2-channel professional D/A converter. To this day, the DAC1 is a standard fixture in many recording studios, and it is also a central component in many high-end hi-fi systems. In August of 2015, Enjoy the Music.com selected the DAC1 as one of the 20 most significant digital audio products from the past 20 years.
After 13 years the DAC1 still outperforms most 2-channel D/A converters, and it still has a legitimate place in top recording studios. The DAC1 is far from obsolete, and most DAC1 converters are still in use today. But technology has changed in the last 13 years, and it is now possible to build a better DAC. Benchmark added the DAC2 in 2012 and discontinued the DAC1 in July of 2015, after 13-years of continuous production.
The DAC2 looks a lot like the older DAC1, but it adds many important features while offering higher performance. Many people have asked if the DAC2 sounds different or better than the DAC1. This is a hard question to answer because both products are designed for transparency. In other words, the noise and distortion produced by each product is well below audibility in normal operating conditions. Nevertheless, there are some situations where the DAC2 can provide an audible improvement. One reviewer, Gary Galo, recently had the opportunity to hear a DAC1 and DAC2 side-by-side. He noted some audible differences and we agree with his conclusions. We have had a great deal of experience listening to these converters side-by-side in our own listening room and we are familiar with some subtle differences.
This paper examines the audible differences between the DAC1 and the DAC2. It will also include measurements that may help to explain these differences.
It is easy to show that the DAC2 measures better than the DAC1 in almost every way. From a marketing perspective it would be tempting to claim that all of these measured differences make audible improvements, but this just isn't the case. The truth is that many of these measured differences should be inaudible when the playback system is properly set up with an ideal gain structure. Both products measure very well because both products are designed to be transparent. This means that they are designed to get out of the way and deliver a clean, uncolored, and quiet output.
The measurements show that the DAC1 and DAC2 add very little noise and very little distortion. Both products have an analog frequency response that extends from 0.1 Hz to more than 200 kHz while inter-channel phase is precisely matched. Both products are reference-grade professional products that are specifically designed for studio monitoring. Accuracy is a primary goal in most professional monitoring chains, and the measurements show that both products perform this role well when running in their ideal operating range.
When the DAC1 and DAC2 are pushed outside of the ideal operating range, the DAC2 begins to shine. The DAC2 remains transparent over a wider operating range than the DAC1. To understand this difference, we need to look at a side-by-side analysis of the measured performance.
We connected a DAC1 and a DAC2 to our Audio Precision 2722 test station and ran a series of side-by-side tests to document the performance differences. Most of these tests show small differences that should be well below audibility. But two tests showed some differences that could be audible: One test showed some potentially audible differences in the lower half of the volume control range. A second test exposed some potentially-audible differences when the digital audio input contained inter-sample overs.
The DAC1 and DAC2 each have high-quality volume controls that are intended to be the primary volume control in the playback system. The DAC1 has an analog volume control and the DAC2 has a 32-bit digital volume control. These controls provide peak performance in the upper half of their rotation. In the lower half of the rotation, the performance is somewhat reduced as the control is turned down. But, this only becomes an issue when the the volume control is turned way down to compensate for an amplifier that has too much gain. The playback system should always be configured such that a normal listening level is achieved while the volume control is in the upper half of its rotation.
If a normal listening level is reached while the volume control is below 50% rotation, then the gain structure of the playback system is misconfigured. These lower volume control settings should only be used to reduce the the music to a background level. The system is incorrectly configured if a normal playback level is achieved when the volume control is below 50%. There is really no excuse for this configuration error because Benchmark converters have programmable passive output attenuators on the analog outputs. These attenuators should be configured to set a reasonable operating range for the volume control on the DAC1 or DAC2. Gross misconfiguration of the gain structure can produce audible noise with either product. Moderate misconfiguration can expose a L/R imbalance on the DAC1 but not the DAC2. In general, the DAC2 is more tolerant of misconfigured gain structures.
The ratio between the audio signal and the background noise or "Signal to Noise Ratio" (SNR) is reduced when the volume control is turned down too far. At full volume, the DAC2 has an SNR of 126 dB, while the DAC1 has an SNR of 116 dB. In a system with a properly configured gain structure, this 10 dB difference may not be audible. To hear the difference, the SNR of the power amplifier would need to exceed the SNR of the DAC1 by at least 6 dB. The Benchmark AHB2 is the only power amplifier that is quiet enough to resolve this difference. Even with an AHB2 running in the high-power bridged mono mode, the noise difference between the DAC1 and the DAC2 should only be audible if the efficiency of the speakers exceeds about 96 dB (1 W, 1 m). Bottom line: in most properly configured systems, the noise produced by either converter will be completely inaudible.
The DAC2 is 10 dB quieter than the DAC1 and this difference can become audible when the DAC to amplifier gain staging is grossly misconfigured.
At full volume rotation, the DAC2 is 30 to 32 dB quieter than the noise produced by 16-bit digital encoding. If the volume control on the DAC2 is turned down by 20 dB to compensate for a power amplifier that has too much gain, the DAC2 still provides a SNR of 106 dB (126 dB - 20 dB). At this setting, the DAC2 still exceeds 16-bit noise performance by 10 to 13 dB. In fact, the DAC2 can still deliver 16-bit performance when the volume control is set at -30 dB.
The DAC1 has an analog volume control, and it is also able to deliver an SNR of 106 dB at a -20 dB volume setting. So, from a noise standpoint, both products have an equivalent tolerance to a 20 dB gain misconfiguration. But, if the gain structure is misconfigured by more than 20 dB, forcing very low volume control settings, the DAC1 may be audibly noisier than the DAC2. Again, this noise difference only becomes audible when there is way too much amplifier gain.
Remember, this configuration error can be corrected by applying the analog attenuators on the XLR outputs of the DAC1 and DAC2 converters. If the jumpers are set correctly, converter noise will never be audible.
The DAC2 has a 32-bit digital volume control and this gives it near-perfect L/R balance over the entire rotation. No analog gain control can match this precision. The DAC1 has a very good analog volume control with a factory-adjusted mid-point L/R multi-turn trim pot. This trimmer gives the DAC1 excellent L/R balance in the upper half of the volume control range. But, like all analog volume controls, the L/R balance on the DAC1 degrades significantly, below 50% rotation. The reason for this is that the L/R balance of an analog control is determined by the accuracy of the matching between the left and right resistor elements in the potentiometer. The desired logarithmic response of a volume control tends to magnify differences at the low end of the potentiometer. This L/R balance problem was the primary motivation for moving to a digital control in the DAC2.
This plot shows the L/R balance of the DAC1 and DAC2 converters. The red trace shows that the DAC1 L/R balance had an accuracy of +/- 0.1 dB down to about a -15 dB volume setting. Below this range, the imbalance of the DAC1 reached 0.9 dB (a potentially audible error). In contrast, the green trace shows that the DAC2 had perfect L/R balance over its entire control range (as we would expect from a digital control). Below a gain setting of about -15 dB, the L/R balance errors in the DAC1 may become audible to some listeners.
The concept of inter-sample "overs" is a bit hard to grasp. If you are not familiar with the topic see "Audio That Goes to 11" and "Why Audio Goes to 11". These two application notes introduce the topic. PCM digital signals can contain peaks between samples that can exceed the normal limits of the digital system by about 3 dB.
Intersample overs are common in 44.1 kHz and 48 kHz recordings as these sample rates can have substantial audio energy at 1/4 of the sample rate. Intersample overs are much less of a problem with 2X and 4X sample rates because there is very little audio energy at 1/4 of these higher sample rates.
Substantial intersample over problems can be created by MP3 compression. MP3 files must be decoded and converted back to PCM (a necessary part of the playback process), and when this happens, many intersample overs are created. The same holds true for other lossy compression systems.
When these overs occur, the DAC2 has a clear advantage. The DAC2 will properly reproduce inter-sample overs while the DSP in the DAC1 will overload. When the DSP overloads, a low-level burst of noise splatters across the audio spectrum with the bulk of the energy concentrated in the higher frequencies. These short transient bursts occur at every inter-sample over, and they can become audible if the recording has enough of these overs. These bursts tend to create a false percussive brightness in the high end. The DAC1 and DAC2 both have ruler-flat frequency response and distortion is normally well below audibility. These transient bursts are the only distortion product that should be capable of reaching audibility in the DAC1. These transient bursts of distortion have been completely eliminated in the DAC2 (see Figure 2).
The red trace in Figure 2 shows how a +3 dBFS overload clips the DSP in the DAC1. The vertical red lines are distortion products produced by the DSP overload. In contrast, the green trace shows that the DAC2 reproduces a +3 dBFS overload without distortion. Cursor 1 shows that the DAC2 produces the correct output level (3 dB above the level produced by a 0 dBFS sine wave).
This inter-sample over problem is not unique to the DAC1. Virtually all up-sampling D/A converters have an inter-sample overload problem when playing 44.1 kHz and 48 kHz recordings. The DAC2 is one of very few PCM converters that can cleanly and accurately reproduce inter-sample overs that are 3.5 dB above the normal digital clip point of 0 dBFS. This extra DSP headroom is followed by extra analog headroom. Together these keep the DAC2 clean when a recording contains frequent inter-sample overs. We believe this unique feature of the DAC2 accounts for the audible difference that Gary Galo noted in his review when he said "the tonal balance is extremely neutral, without any of the DAC1’s brightness".
MP3 compression seems to greatly increase the occurrence of inter-sample overs. This side-effect of the data compression process seems to magnify the defects in MP3 compression. I find that it is much easier to identify MP3 compression when listening to a DAC1, than when listening to a DAC2. For this reason, we find that MP3 compressed files are more tolerable on a DAC2 than on a DAC1. In either case, we strongly recommend avoiding MP3 compression. Set your computer CD rip options to a lossless format before ripping your CD collection! But, if a few MP3s sneak into your music collection, the DAC2 will make them more tolerable.
The DAC2 has a diagnostic display that shows the measured sample rate and word length of the incoming audio. This display should match the format of the audio being played. If the format does not match, the upstream device is not delivering a bit-transparent digital stream. By default most computers apply processing to the digital audio. This processing includes sample rate conversion (SRC), volume control, mixing, EQ, and other functions that can cause audible changes in the sound of your recordings.
On many occasions we have found that the diagnostic display on the DAC2 identified an audible problem with a computer-based playback system. The DAC2 display allows the detection of configuration problems that can cause audible differences. These configuration problems can be very difficult to correct without access to a diagnostic display.
This paper is focused on the audible differences between the DAC1 and DAC2, but it is important to note that there are many added features on the DAC2. These include 192 kHz USB, DSD, USB to coaxial conversion, optical to coaxial conversion, increased I/O, and a 12 V trigger. It also adds DIM, MUTE, and POLARITY controls.
We have identified two audible differences between the DAC1 and the DAC2:
We have also identified an operational difference than can make an audible difference