DAC-Preamp having these specifications?

Rather than asking for feedback about specific products, I thought I instead would ask for your recommendations of a DAC-Preamp product(s) having a desired set of following specifications:

Output Impedance: < 300 Ohms
Balanced outputs (XLR)
Unbalanced outputs (RCA)
2 or more analog inputs (RCA or XLR)
2 or more digital inputs (coaxial + optical or USB)
DAC section preferably uses discrete components over op-amps
DAC signal processing capabilities should the usual fare though need not include every conceivable format
Price: ~$2,500 or less

Mytek Manhattan - But I think they are currently pretty much out of stock. Not sure what the output stages are, but they sound great. :) 
The Manhattan II’s price of $6k knocks this out of contention. The $2200 Brooklyn DAC+ includes one less analog input than I would prefer, but has the desired output impedance (75 Ohms). Not sure what their mix of op-amps in the unit might be. 
The Benchmark DAC3 is another contender that appears to fit most of the specifications list. I understand that some ding it for having too many op-amps in the unit that adversely affects its sound qualities compared with units having discrete components. Audioengr told me the op-amps contribute to the unit’s internal jitter in the DAC section being too high.
I have used Bel Canto DACs the last 5 years.
I currently have the DAC 3.7. It is a hybrid design with SS and Class D, and is incredibly accurate (they use TOTL clocks), musical, fast, with top to bottom balance as well as gravitas in the low end.

Can be had for $2500. has an outboard PSU- it is a "virtual battery supply", that has such low distortion that it mimics an actual battery )(batteries have less distortion than a power supply attached to the wall constantly).

this is a $5000-6000 DAC that is priced at $2500 due to Bel Canto not really advertising much and using more word of mouth to sell products.

In fact, this is a discontinued model b/c it was so good in combination with the Bel Canto Renderer that is was detracting from their sales of their new line of "Black" and "Black EX" products, which run over $15,000. I hope I am making my point here ....

only drawbacks: one Analog input, BUT you can connect a high end splitter for 2-4 inputs (eg. from Goldpoint, $300-400) http://www.goldpt.com/index.html ) or a cheap one from Schiit.

But, the ADC in this DAC is very high end, and like no other I have experienced. it will take a signal, eg from my tuner, convert it to digital, then use their propriety ADC to convert to 24/192 hi res audio. And it sounds amazing.

The DAC is always in Standby mode, using 0 watts!
When being used, the two units (Audio and power) only use a paltry 18 watts. So you save money not only on using less power, but use less AC during the summer to keep the room cool!

best of luck, and I have no affiliation with Bel Canto, just a fan trying to get the word out,


Check our the measurements of the Benchmark DAC 3.

If there is any internal jitter it has not shown up in any of the measurements - so likely very small.

The J-Test is perfect. Discrete components aren’t necessarily better - usually you can achieve a lower noise floor but at -160 dbfs the noise floor is pretty low on the DAC 3...

@shadorne yes, I reviewed JA’s measurements of the DAC3 this morning to see whether he measured the jitter. I know it’s low. 
DAC technology is a moving target. The ESS Pro chip set in the Benchmark product is already obsolete, though I’m sure it’s great as implemented.


Though it’s not a stated product specification, it would be desirable to have some commitment to upgradable pathways for the product.
The ESS 9038 is multi-channel for HT. Higher performance when you average 4 channels per side to get stereo. I think there are some products that use it for stereo. The implementation is just as important as the chip.
now we are digging a little deeper into the OPs choices: DACs

IMO, I dont like the Sabre DACs, I think they sound too sweet and syrupy, like a sonic signal being over processed a 1000 times. Doesn't even sound real to me. I prefer AKM and Burr-Brown chips myself.

Again, Celander, Bel Canto DAC 3.7, and you can thank me later.
At least give it a demo ....
@1graber2 the BC 3.7 checks all the boxes except price. It seems $2k higher than what I’m interested in spending. 
Just for grins, I pulled out of storage my CAL Delta transport, Sonic Frontiers Ultra Jitterbug (reclocks transport signal to <40ps jitter) and Theta DS Pro Basic IIIA (interlinked via Canare AES/EBU 110-Ohm digital IC’s) and ran it head to head against my Oppo BDP-83SE as a Redbook system (16-bit, 44kHz). The Theta has Analog Devices processors, akin to the Schiit Yggy (Mike Moffat apparently loves implementing AD processors with robust software). All I can say is that the 1996 setup didn’t embarrass itself against a 2010 Oppo player.
Ok, i was thinking used prices, with range for a used but Exc condition unit from $2400-$3500, which I have seen IME. Many sellers appear to settle around $2500 as a final price, which would fit the budget. 
But you only want new?
celander OP
  DAC-Preamp having these specifications?
Look for one that's direct coupled on it's output buffer, and I would go <100ohms output impedance. And also has these features for the best quality.
Go for one that has settable max analogue gain output, so you can use it’s digital volume control in the top 25% part of it’s range.
This way your sure of not "Bit Stripping". Just like Wadia and Mark Levinson do on their Flagship models.
Today I know M1 Bricasti has it, there must be others as well, these are the ones I would go with as they know about "bit stripping" that's why it’s introduced. As the Wadia attached shows.


Cheers George

There's a store demo Bel Canto DAC 3.5VB for sale on eBay for $1799 (missing remote).  This unit was $5k new.  This is a great opportunity.  BTW, if you are considering a BC DAC you HAVE to get one with the virtual battery supply (VBS).  I got to hear them back to back in my system, and the VBS takes the DAC to a much higher level.  Best of luck. 

Post removed 
@georgehifi My ATC active monitors have a 10-kOhm input impedance, so I need something having an output impedance lower than 1-kOhm.
Yes generally as 1:10 rule I use it, "if" your interconnects are low capacitance and around 1mt long, source to amp.

But your going from dac to speakers with these interconnects, probably around 4mts, so I would err on the side of caution and have the dac’s output impedance 100 or 200ohms.

Cheers George
@georgehifi I thought I deleted my errant post regarding my misread of your suggested “1000-Ohms”. I read your post again and realized you wrote “100Ohms” rather than “1000Ohms”. My bad.

The ATC (analog) preamps have a rated output impedance of 10-Ohms. But I don’t have one of those, opting instead to explore a DAC-preamp. 

Thanks for your advice. 
@georgehifi If a DAC-preamp manufacturer uses only an analog volume pot, then can one assume the digital domain is set for 100% output, thereby ensuring no bit-stripping arises? I’m thinking here of the Benchmark DAC3. 
If a DAC-preamp manufacturer uses only an analog volume pot, then can one assume the digital domain is set for 100% output, thereby ensuring no bit-stripping arises? I’m thinking here of the Benchmark DAC3.
Yes, no "bit stripping" if the volume is done in the analogue domain after the d to a converter. But I haven't seen this done since the old days, maybe the Dac3 is I'll look to see if it is.

Cheers George
Just had a look at what Sterophile makes of the DAC30’s volume control, and it’s a bit vague, but to me point to it being in the digital domain not the analogue as that ES9028PRO d to a converter has volume option.
As they talk about extra resolution if you use it in the top part of it’s range. And benchmark give you the options of 0db or -10db or -20db gain attention structure in the analogue domain output buffer so then it can have the volume control it the top part of the range. (This wording "could" be the reason why some think it has a normal analog volume pot.)
The DAC3 HGC has another feature that, while hardly sexy, addresses a common and underappreciated problem. Is your volume control usually set above or below the halfway point? In my experience, for most systems the answer is "below"—but most systems achieve their best noise performance in the top half of their volume range for both analog and digital volume controls (though for different reasons). Also, analog volume controls exhibit their best channel-matching in that range. The DAC3 offers pads for its balanced outputs—attenuators—that can be set, via internal jumpers, to 0, –10, or –20dB. Benchmark says that when they’re in use, the DAC3 retains its full signal/noise ratio of 128dB, A-weighted. If you’re paying for 24-bit DACs and hi-rez downloads, you’d best get your noise level down to where you can hear at least some of that extra resolution, and that’s harder than you might think

They do what Wadia did by the looks of it.

Quote from Wadia

We strongly recommend that you use your Wadia 521 Decoding Computer connected directly to your power amplifier. Even if you purchased your Wadia 521 Decoding Computer with the intention of connecting it to your preamplifier, we suggest that you try direct connection to your amplifier. Many listeners are surprised by the improvement in performance over even the most expensive preamplifiers.

Optimizing the Output Level Best performance is obtained when operating the Wadia Volume Control near the top of its range. If needed, the maximum output level of your Wadia 521 Decoding Computer can be adjusted to match the overall sensitivity of your system so that the critical listening will take place with the volume control operating in near the top of its range.
The maximum output level of the Wadia 521 Decoding Computer is adjustable by means of a set of internal switches. The Wadia 521 Decoding Computer is factory set to accommodate the most common range of system sensitivity. If you find that your typical volume level during critical listening is below 65 on the volume display, it will be advantageous to use a different setting. To change the output level, consult your dealer.

BTW the DAC30 has a lovely low 10ohm output impedance and will drive anything, just set the gain setting (0,-10, -20db) so you volume is up near full for very loud and you won’t run the risk of "bit stripping" "if" it in the digital domain.

Cheers George
@georgehifi I looked at the owners manual after posting. I think you are correct about the digital domain volume control. Here is what their manual says:

“HGC is Benchmark's unique Hybrid Gain Control system. The DAC3 combines active analog gain control, passive low-impedance attenuators, a 32-bit digital gain control, and a servo-driven volume control.

“All inputs are controlled by the rotary volume control. This volume control moves in response to commands from the remote control. Analog inputs are never converted to digital, and digital inputs never pass through an analog potentiometer. Digital inputs are precisely controlled in the 32-bit DSP system. The DSP system preserves precise L/R balance, and precise stereo imaging, while avoiding any source of noise and distortion.

“Benchmark's unique passive output attenuators provide distortion-free gain reduction without reducing the dynamic range of the converter. The attenuators optimize the gain staging between the DAC3 and the power amplifier. This optimization is absolutely essential for maximizing the dynamic range of the entire playback system.”

The output impedance isn’t uniform as a function of the different passive attenuator pads:
Attenuator:      Output Impedance
0 dB                      60 Ohm
-10 dB                 425 Ohm
-20 dB                 135 Ohm

Hybrid Gain Control system
Yes that "hybrid" word says to me half digital, half analog. The the gain setting (0--10,-20db) is done in the analog and the volume is done by the ess in the digital.

The output impedance isn’t uniform as a function of the different passive attenuator pads:
Attenuator: Output Impedance
0 dB 60 Ohm
-10 dB 425 Ohm
-20 dB 135 Ohm

The output impedance isn't uniform as a product of the gain setting, not the volume control.
That’s because the gain is most probably changed via different feed back resistors of the output buffer which will effect it’s output impedance, not to worry though, they are all low enough for your active ATC speakers.
Just use good quality low capacitance interconnects to them say ( less than <100pF "picofarad" per foot) ask the manufacturer. Keep them short as possible 2-3mt will be fine

Cheers George
BTW: Good on Benchmark for recognizing "bit stripping" even in the ESS-ES9028PRO converter and implementing an analog user changeable gain bock setting like Wadia and ML did. Bricasti did too in the M1 but you change theirs differently. 

Cheers George

The owners manual of the all digital input version, DAC3 DX, makes even more clear the digital volume control of the outputs:

”The MAIN bus drives the XLR outputs and one pair of RCA outputs. The MAIN bus delivers the highest performance because it uses three conversion channels wired in parallel for each XLR connector. The main bus uses 6 of the 8 channels in the ES9028PRO D/A conversion chip. The remaining two channels in the ES9028PRO drive the AUX bus.“

There are essentially 4 ways to get volume control:

1) analog volume control that changes the gain of an active stage - can be controlled digitally

Disadvantage - change in gain characteristics at different volume levels and signal levels can affect dynamics or cause varying compression.  The lower the volume, the higher the distortion.  It usually adds a stage to the signal path as well, adding to the compression and distortion.

2) attenuation with resistors or optical controlled resistors

Disadvantage - large resistances in the signal path add thermal noise and can affect impedance and frequency linearity if used on outputs.  If this attenuation is between stages, the lower the volume, the higher the distortion, the lower the S/N ratio.

3) modify the digital words using DSP

Disadvantage - DSP usually impacts SQ if more than about -9dB of attenuation is required, even with the best DSP software, such as Sonic Studio.  For less than -9dB of attenuation, it works pretty well.  The difference between the quietest tracks and the loudest tracks can easily be 12dB.

4) change the reference voltage that the D/A conversion uses.

Disadvantage - it is difficult to turn the volume down to zero.  It can be quite low, but never zero.  Overall, the best solution of the 4.  Extremely low distortion and it gets lower as the volume is decreased, not higher like the other techniques.  No added resistance or stages to the signal path.

Steve N.

Empirical Audio

@audioengr So the DAC3 uses method 2, with the attenuation arising after the last active stage? I’m unclear about this. 

No different than an external passive resistive attenuator. Changes impedance and can cause roll-off. Adds thermal noise. Critical to use low-capacitance interconnects with this type.  I stay away from these myself.

Another type of volume control I failed to mention is by transformer with multiple taps. An autotransformer is usually the design. This is typically not done in a DAC. Never seen it in a DAC. Quite expensive if done right with quality transformers, but this method has very few drawbacks. Only saturation of the transformers is an issue. Mostly the disadvantage is cost.

Steve N.

Empirical Audio

So the DAC3 uses method 2
This (linked) is where I believe the volume is controlled in the digital domain before d/a conversion. The other so called "hybrid" part is just an analogue gain adjusting (0db -10db -20db) on the output buffer for the maximum level volts obtainable if the digital volume in the ESS chip is full up.


Cheers George
I have sent a request to benchmark about this topic.
I have sent a request to benchmark about this topic.
So did I yesterday, nothing back yet.

Cheers George

Make sure you use RCA outputs if you are concerned about noise, Benchmark has pro-grade RCA - it runs at about 18Volts. This is a lot stronger than regular consumer RCA (2 volts) and confers a huge advantage. The SNR advantage is about 13dB.

Benchmark wrote an article about volume control in their older DAC 1. It is dated but the concepts are valid.


The Benchmark DAC 3 combines active analog gain control, passive low-impedance attenuators, a 32-bit digital gain control, and a servo-driven volume control. It basically uses a combination of Steve’s methods in order to minimize deliterious affects of excessive volume attenuation from one approach.

It is important to keep headroom so running signals as high as possible through an audio device and chain helps achieve the lowest noise. Running high signal levels is the way to do it and as Steve explained any attenuation through DSP, resistors or otherwise is going to reduce performance. A precision resistor pad on the output is an excellent way to allow a device to run at the highest signal levels possible and maintain good SNR while still attenuating the signal enough not to overload the next stage.

If you are really serious about noise then you should purchase their HPA-4 and run the Benchmark DAC with no attenuation at reference level (close to full volume) and feed the XLR analog output to the HPA-4 (for volume control prior to power amp). Personally I think this is overkill as the DAC 3 has one of the better volume control systems currently available on most DACs.


@shadorne thanks for the links. Looks like the DAC3 HDR-VC™ system combines an active gain circuit with a low-impedance passive attenuator, where the motorized volume pot controls the active gain circuit and the passive attenuator is located after the active circuitry.

The DAC 3 is more advanced than the HDR-VC of the DAC 1. The DAC 3 has 32 bit DSP volume control also in addition to passive attenuators and active analog gain circuit. So for best results you should feed the DAC 3 the full bit perfect signal (no software volume control). The rule of thumb is keep volume as high (without clipping) as you can to maintain headroom and highest SNR. 
@shadorne yes, I see that. Rory of Benchmark sent a bunch of links of application notes to me. Looks like they’ve also improved the master clock and how jitter is controlled in the DAC of the DAC3 unit.

Yes they have addressed fully the common PLL issue of slave hunting master. In some cases the way that a slave adjusts to a master can actually produce worse jitter - they have fully addressed this. They reduced the clock cycle adjustment rate to below 1Hz which should be completely inaudible (this is the maximum frequency tone you could hear if the slave was in continual adjustment to try to match the master - of course 1 Hz isn’t audible). Mathematically their technique is both solid technically and elegantly simple and runs at 250 GHz with clock adjustments of 4 picosecond.
@celander ... or anyone
Here is your Bel Canto 3.7 DAC with PSU within budget of $2500.

IMO, it is pretty risk- free.
Not just bc of the Exc SQ, but b/c if for some insane reason you don’t like it, you can sell it for the same price you paid, or even make a couple hundred dollars with a resell.

This is is Mint unit that a AV dealer is just trying to offload at less than 1/2 price of the MSRP 

Re: the Benchmark DACs, I almost bought one 5 years ago. The jitter and specs looked great, but when I actually listed to the Benchmanrk, I found it to be dry and sterile sounding. And no no offense to the Benchmark lovers, just IME. 

John Strongzer of Bel Canto has been designing DACs and Class D for over 20 years. 
Again, Bel Canto advertises very little, they don’t spend much of their budget money on advertising (like Benchmark), so that means more R and D $  is going directly into the product.