DAC shopping thoughts

[firedog]

You can certainly consider this OT if you would like, but I would hesitate to spend serious money on a DAC these days. With A+ or HQP offering much of the performance of expensive DACs without high hardware costs, I personally don't think it's worth it.

 

The next DAC I'm particularly interested in seeing is Miska's DSC-2 whenever he's able to get around to it. There are probably sources to have that built, if you're not up for DIY, that would be well below $2000.

 

Or you could look at something that will do a nice job being fed by oversampling software, like the T+A for $4000, the iDSD Pro for $1000, or even the revised micro-iDSD for $500.

 

 

Sent from my iPhone using Computer Audiophile

 

Intersting take.

 

The T+A interests me, but the dealer here doesn't keep them in stock (no audition) and was sort of a jerk about it when I enquired. Basically his response was, "Go ahead and order one, it's your problem if you're not happy afterwards". So not too interested in working with someone like that, and the price is high enough that it makes me pretty risk averse.

For comparison, I'm going to look at the Simaudio 280 and 380, and the dealer doesn't stock the 380. But he told me he'd order it for me at about a 20% discount if I order it unheard from him (and I can audition it when I travel abroad this winter). That's at least someone who understands something about customer service and who I'd feel comfortable working with.

 

I'm definitely not into DIY, if you ever find someone reputable who could build Miska's DAC I'd look into it.

 

The other 2 DACs you mention don't seem to me to be an upgrade over what I have. I like my DAC (especially now with the LPS powered mRendu), but it has a what I would call a little of a "pro" type sound, and I'm looking for something with what I'd call a more "natural" sound, and an improvement with aspects like leading edge, cymbal trails, etc.

 

We will see if I find that in the DACs I'm listening to over the next few months...

[Jay-dub]

Do you think the Chord MoJo digital volume control is good enough to run it directly into a power amp?

Could you point me to anywhere this might have already been discussed!

Thanks

I would trust the results of Miska's measurements better than any user's subjective impressions. He recommends feeding the Mojo with high-rate PCM and disabling its built-in volume control. I have no knowledge whether he would consider it a good solution to apply volume control in software before feeding it to the Mojo, which would also make it possible to hook up the Mojo directly to a power amp; to me it seems like a good way of doing things.

 

http://www.computeraudiophile.com/f8-general-forum/upsampling-anything-other-your-digital-analogue-converter-s-internal-conversion-rate-30240/index3.html#post594624

[ctsooner]

Heard the Chord DACs this weekend. The Codex should be in your listening auditions oif you are looking at the DACs being mentioned. Forget specs and measurements and go listen. The Mojo is excellent in its price range. Like all chords though, it lacks bass punch. Personally you can do much better for a home system with a 2 k price range. For 600 and under its in the top choices for me though.

 

 

Sent from my iPad using Computer Audiophile

[ecwl]

I would trust the results of Miska's measurements better than any user's subjective impressions. He recommends feeding the Mojo with high-rate PCM and disabling its built-in volume control. I have no knowledge whether he would consider it a good solution to apply volume control in software before feeding it to the Mojo, which would also make it possible to hook up the Mojo directly to a power amp; to me it seems like a good way of doing things.

 

http://www.computeraudiophile.com/f8-general-forum/upsampling-anything-other-your-digital-analogue-converter-s-internal-conversion-rate-30240/index3.html#post594624

I think Miska's HQPlayer is a very unique situation because it is theoretically possible that HQPlayer has more computational power than Chord Mojo or maybe even Chord DAVE's FPGA so the upsampling filter to 32-bit 768kHz would sound better than the upsampling WTA filter of Chord with its own FPGA doing volume control. But nowhere did I see franz159 using HQPlayer or has a super powerful computer to run HQPlayer at 1 million taps. Miska says it's better to use HQPlayer. Rob Watts thinks the Chord FPGA is better. Who knows... Obviously, hooking up an ultra-powerful computer to a DAC causes other problems, unless you're going to get a microRendu and send the upsampled data through the microRendu. Now we have increased the complexity of the system to a point where a $600 Mojo becomes a $600 Mojo + $400 LPS-1 + $400 microRendu + $2000 CPU+GPU+Desktop+HQPlayer and if you want Tidal, it's another $500 for Roon... My take is always whatever works for people is what works. As long as we are happy with our setup, it's all good.

[Jay-dub]

possible that HQPlayer has more computational power than Chord Mojo or maybe even Chord DAVE's FPGA so the upsampling filter to 32-bit 768kHz would sound better than the upsampling WTA filter of Chord with its own FPGA doing volume control. But nowhere did I see franz159 using HQPlayer or has a super powerful computer to run HQPlayer at 1 million taps.

It's not going to take anywhere near a million taps to outperform the Mojo. Miska recommends software upsampling to 16x because the Mojo uses linear interpolation from 8x on up, which is strictly speaking a two-tap filter, but computationally even simpler, because it can be implemented with no multiply operations at all! Even if you just use iTunes to upsample to 768 kHz you'll get most of the benefit.

[ecwl]

The other 2 DACs you mention don't seem to me to be an upgrade over what I have. I like my DAC (especially now with the LPS powered mRendu), but it has a what I would call a little of a "pro" type sound, and I'm looking for something with what I'd call a more "natural" sound, and an improvement with aspects like leading edge, cymbal trails, etc.

What Jud is referring to is this: All DACs are actually made of two parts, one is the upsampling/digital filter part and the other is the true DAC part. So our CD 16-bit 44.1kHz get upsampled, filtered and converted to say 2.8MHz 5-bit in a dCS DAC, 5.6MHz 1-bit in a PS Audio Directstream DAC, to sort of 8-bit 10MHz for your Mytek Sabre DAC chip or to 352kHz 24-bit? in a Schiit Yggradasil. And then the DAC portion, be it DSD, multibit SDM DAC chip or R2R component takes over.

Lately, there has been more emphasis on the digital filter/upsampling. Chord's Rob Watts and HQPlayer's Miska has been saying that they feel the lack of computing resources to perform the upsampling is a major reason why DACs are underperforming sonically. Others are designing alternating digital filters that still doesn't require a lot of computing resources (such as the apodising filters in Meridien and lots of other DAC designs now).

So what Jud is saying is that Miska's HQPlayer's philosophy is to not do the upsampling within the DAC unit itself but let the computer run HQPlayer to do the upsampling from 44.1kHz 16-bit to 1xDSD/2xDSD/4xDSD or even 8xDSD and then you can just send the data out to a very simple DSD DAC at a very low cost and get much better sound than buying an expensive DAC with its internal upsampling/digital filter and DAC component which may be more compromised. So buying an iDSD Pro or micro-iDSD is not a compromise or a downgrade because you would upgrade your computing side with HQPlayer and maybe a new desktop with powerful GPU+CPU to upsample to 8xDSD to feed the iDSD. The purpose is to mimick what you would get if you have say a Chord DACs with its more powerful FPGA for upsampling.

[ecwl]

It's not going to take anywhere near a million taps to outperform the Mojo. Miska recommends software upsampling to 16x because the Mojo uses linear interpolation from 8x on up, which is strictly speaking a two-tap filter, but computationally even simpler, because it can be implemented with no multiply operations at all! Even if you just use iTunes to upsample to 768 kHz you'll get most of the benefit.

I think you misunderstood what Rob Watts and Miska are saying. So Mojo uses linear interpolation from 16x up (not 8x because that's Hugo), but that means it uses about 28000 taps (who really knows since Chord doesn't specify) using the WTA filter to upsample from 44.1kHz to 704kHz. And then because Chord Mojo runs at 104MHz, it then uses linear interpolation FIR filter to upsample from 704kHz to 104Mhz. Now, you can program HQPlayer to use many more taps to upsample/filter from 44.1kHz to 704kHz. However, we should think of these filtering options as exponential functions that we learnt in first-year university calculus classes. For example, there are multiple exponential functions that can converge into e or pi. But some functions would converge faster and others would converge slower onto the true value of e or pi. So what exponential function you use will determine what order polynomial you'll need to compute to in order to get a good enough approximation of e or pi. So what Rob Watts is claiming is that the WTA filter is better than linear FIR filters when using the same tap length. I think in the past he said 256 taps running his WTA filter sounds better than 2048 or even 4096 taps running the standard FIR filters. Obviously, Miska also has his own filters that I'm sure is great. But we are really comparing say Chord Hugo's 28000 taps WTA filter with Miska's filter running with how ever many taps the CPU+GPU can do. that's why no, iTunes upsampling to 768kHz is not going to sound better than a Chord Mojo's internal upsampling.

Another aspect, which is why Rob Watts designed Chord DAVE the way he did was that there is a limit in terms of how much data Miska can send via USB to the DAC. So even though computationally HQPlayer can theoretically do a better real-time upsampling/filter to 256x, there is no way for HQPlayer to send via USB into a DAC 11.2MHz @ 32-bit, whereas Chord DAVE can take the 44.1kHz 16-bit data via USB and then use the 164000 taps running the WTA filter to do the 256x conversion to 11.2MHz 32-bit internally.

This is why I think Miska spends a lot more time optimizing DSD performance in HQPlayer because there are now great USB receiver that can handle 22.4MHz 1-bit DSD data.

I personally think that all DAC designs involve technical challenges and compromises. And obviously all designers are going to say their DAC is the best. It is much more interesting to see how they approach the design and we are truly lucky that we have people like Miska, Rob Watts or John Swenson who would openly comment on their approaches.

[firedog]

What Jud is referring to is this: All DACs are actually made of two parts, one is the upsampling/digital filter part and the other is the true DAC part. So our CD 16-bit 44.1kHz get upsampled, filtered and converted to say 2.8MHz 5-bit in a dCS DAC, 5.6MHz 1-bit in a PS Audio Directstream DAC, to sort of 8-bit 10MHz for your Mytek Sabre DAC chip or to 352kHz 24-bit? in a Schiit Yggradasil. And then the DAC portion, be it DSD, multibit SDM DAC chip or R2R component takes over.

Lately, there has been more emphasis on the digital filter/upsampling. Chord's Rob Watts and HQPlayer's Miska has been saying that they feel the lack of computing resources to perform the upsampling is a major reason why DACs are underperforming sonically. Others are designing alternating digital filters that still doesn't require a lot of computing resources (such as the apodising filters in Meridien and lots of other DAC designs now).

So what Jud is saying is that Miska's HQPlayer's philosophy is to not do the upsampling within the DAC unit itself but let the computer run HQPlayer to do the upsampling from 44.1kHz 16-bit to 1xDSD/2xDSD/4xDSD or even 8xDSD and then you can just send the data out to a very simple DSD DAC at a very low cost and get much better sound than buying an expensive DAC with its internal upsampling/digital filter and DAC component which may be more compromised. So buying an iDSD Pro or micro-iDSD is not a compromise or a downgrade because you would upgrade your computing side with HQPlayer and maybe a new desktop with powerful GPU+CPU to upsample to 8xDSD to feed the iDSD. The purpose is to mimick what you would get if you have say a Chord DACs with its more powerful FPGA for upsampling.

 

If you'd looked at my signature you'd have seen I'm an HQP user, and saved yourself a lot of writing.

[ecwl]

If you'd looked at my signature you'd have seen I'm an HQP user, and saved yourself a lot of writing.

 

Haha. Good point. I think there are HQPlayer users who don't actually upsample and merely use the HQPlayer to send music to their microRendu. Of course, they could also just do that using Roon. Anyway, you never mentioned how you're upsampling into your Mytek. If you're already upsampling to 2xDSD (DSD128), I think Mytek feeds the DSD128 signal through the ESS Sabre chip still. Since it sounds like you're happy with it, you really owe it to yourself to find a true pure DSD DAC at the highest sampling rate (DSD512). I think you'll be surprised by the improvement off HQPlayer. I'm guessing the T+A DAC 8 DSD fits the bill?

[udis]

For what it's worth. I upgraded my Mytek to Vega and that was a significant jump. Recently, i upgraded my Vega to Bricasti M1SE, and that was a massive jump. So went from $1.5k - $3.5K - $10k. Can't comment on the law of diminishing returns with DACs because i haven't listened to all DACs at various price points. However, for me, all upgrades were well worth it. From my auditioning, going from a $5k DAC to a $10K provided a big delta in terms of overall sound quality in my system.

[Jud]

If you'd looked at my signature you'd have seen I'm an HQP user, and saved yourself a lot of writing.

 

Yep, interesting that you are getting the same quality of sound from the Mytek regardless of which HQP filter and modulator combo you use.

 

The only potential change from what you are hearing offered by the DACs I mentioned would be any possible advantage of DSD256 or DSD512.

 

 

Sent from my iPhone using Computer Audiophile

[ecwl]

Yep, interesting that you are getting the same quality of sound from the Mytek regardless of which HQP filter and modulator combo you use.

 

The only potential change from what you are hearing offered by the DACs I mentioned would be any possible advantage of DSD256 or DSD512.

Or maybe it's not that surprising if the DSD128 is going through the ESS Sabre chip in the Mytek and getting some sort of processing to the signal... I think the iFi/AMR designer talked about this issue as most multi-bit SDM DAC chips that offer DSD playback are not true native DSD.

The DACs you recommended are true native DSD or essentially true native DSD through the DAC chip.

 

 

Sent from my iPhone using Computer Audiophile

[esmit]

The T+A cannot be fed DSD512 by the microRendu (yet?), no suitable Linux driver. Buy the Mojo, if you like it order the Hugo TT ;-) and sell or reuse the Mojo elsewhere.

 

Sent from my D6633 using Computer Audiophile mobile app

[esmit]

But do not forget Mytek Brooklyn or Manhattan II

 

Sent from my D6633 using Computer Audiophile mobile app

[Ralf11]

re: you have the funds to get something pretty pricey, but would that be a mistake on sort of the conceptual level:

 

1. what else would you do with those funds?

- do you have $20,00 speakers?

- room treatments?

- build a new room or remodel one to make it the very best for SQ?

 

2. what would the value of that new DAC be in 2 years?

 

- no matter what I'd wait for the next CES...

[orgel]

The T+A interests me, but the dealer here doesn't keep them in stock (no audition) and was sort of a jerk about it when I enquired. Basically his response was, "Go ahead and order one, it's your problem if you're not happy afterwards".

 

FWIW, if you order the T+A directly from Rutherford Audio,, they offer a 30-day money-back guarantee and will pay for return shipping.

 

I'm in a similar position to yours. After a fair amount of research and plenty of dithering, I'm down to choosing between the T+A and the exaSound e32. I'm currently leaning toward the e32.

 

--David

[R1200CL]

FYI, I'm either getting a Dac with a volume control or replacing the power amp with an integrated.

 

Firedog; should not MAQ be a demand for your new DAC ?

If you are a Tidal subscriber, I'm quite sure MQA is going to happen some time next year. It would be to bad to miss that opportunity, don't you think so ?

 

Have you considered the Holo Spring R2R DAC ? It has got unbelievable good feedback.

 

The challenge is of cause you most likely will not be able to deliver it back, but you do not know that before you have asked. You may be able to bypass the custom issues by calling it a sample.

Do you use 230 AC and live in EU or EØS area ?

 

Volume control is a challenge if not done properly. If I have understood it correctly, done in the digital domain, will affect the SQ negative.

And done properly in the analog domain, is normally the expensive ones that do.

Maybe anyone with better knowledge can comment on this ?

[ctsooner]

Guys I have the Ayre QX5 twenty and an Ayre AX5 Twenty integrated and I have tested the digital volume on that DAC vs using the pre amp section as it has a direct pass through built in. I've done this with my previous DAC the Empirical Audio ODSE/SEnwith all the updates. Neither one, using digital control, came close to using the preamp. In stores I've heard other DACs all the way up to 30k, give the same sound quality vs using a good preamp or running through the pre in the integrated.

 

It's a marketing thing that some companies have used to get you to spend more for the DAC and make you think you can use one less component. In reality it doesn't work yet. Getting much closer but you will lose intimacy and the musicality and I'm not the only one saying this either. We have other threads about this subject on many forums.

 

 

Sent from my iPhone using Computer Audiophile

[ecwl]

Guys I have the Ayre QX5 twenty and an Ayre AX5 Twenty integrated and I have tested the digital volume on that DAC vs using the pre amp section as it has a direct pass through built in. I've done this with my previous DAC the Empirical Audio ODSE/SEnwith all the updates. Neither one, using digital control, came close to using the preamp.

 

I think the problem is two-fold. One is that some preamplifiers would add distortions that are euphonic and some people just prefer the euphonic distortions. But the other issue is related to DAC linearity at low levels. Once you're using digital volume attenuation, you're relying on the low-level linearity of the DAC (assuming great output stages to begin with). Inherent within each type of DAC designs is a limit to their low-level linearity. In general, DSD DACs are worse than R2R DAC which are worse than DAC chip DACs. Obviously, the problem is more obvious the more attenuation you do. This is also why many DACs actually don't have built-in digital volume attenuation and recommends you to buy a preamplifier. Unfortunately, the linearity is not a measurement that one can easily find for most DACs. I generally look at Hi FI News and Record Review's measurements. The first graph they display shows the measurement and the straighter the sloped line, the better the linearity and the less distorting the digital volume attenuation would be. Obviously, the corollary is that if you're using a preamplifier, you're relying on the linearity of the analog volume attenuation...

 

 

 

Sent from my iPad using Computer Audiophile

[Jud]

I think the problem is two-fold. One is that some preamplifiers would add distortions that are euphonic and some people just prefer the euphonic distortions. But the other issue is related to DAC linearity at low levels. Once you're using digital volume attenuation, you're relying on the low-level linearity of the DAC (assuming great output stages to begin with). Inherent within each type of DAC designs is a limit to their low-level linearity. In general, DSD DACs are worse than R2R DAC which are worse than DAC chip DACs. Obviously, the problem is more obvious the more attenuation you do. This is also why many DACs actually don't have built-in digital volume attenuation and recommends you to buy a preamplifier. Unfortunately, the linearity is not a measurement that one can easily find for most DACs. I generally look at Hi FI News and Record Review's measurements. The first graph they display shows the measurement and the straighter the sloped line, the better the linearity and the less distorting the digital volume attenuation would be. Obviously, the corollary is that if you're using a preamplifier, you're relying on the linearity of the analog volume attenuation...

Sent from my iPad using Computer Audiophile

 

 

HQPlayer has its own digital volume control, so the DAC doesn't need it.

 

*Note:* If using Linux, there are reports of potential speaker-damaging transients with no additional attenuation in the chain. I believe this is something to do with Linux rather than HQPlayer in particular.

 

ecwl, what is the distinction you are making between "DSD DACs" and "DAC chip DACs"? Examples? Why would "DSD DACs" have inherently worse low level linearity than R2R DACs?

[blownsi]

Firedog; should not MAQ be a demand for your new DAC ?

If you are a Tidal subscriber, I'm quite sure MQA is going to happen some time next year. It would be to bad to miss that opportunity, don't you think so ?

 

Wait... Tidal is going to stream MQA next year? I missed that one, can you elaborate?

[ecwl]

HQPlayer has its own digital volume control, so the DAC doesn't need it.

 

ecwl, what is the distinction you are making between "DSD DACs" and "DAC chip DACs"? Examples? Why would "DSD DACs" have inherently worse low level linearity than R2R DACs?

 

So this has nothing to do with accuracy of the digital attenuation be it done in HQPlayer or within the DAC. The problem with R2R is matching the resistors both for the low level ones but also so that the higher level ones are truly double that of the low level ones. That's why most have multiple R2R chips to average things out. The problem with DSD is different. To get 44kHz 16-bit to DSD64, for simplification (though not technically correct), you can imagine each 44kHz sample becomes 64 1's and 0's. So with null audio signal, you re basically between 1's and 0's like 32 times. Now you can imagine to get a very small signal in, you'd have to switch one of these 1's and 0's but now, you're limited by the switching rate. Obviously, if you go to DSD512, you'll have a lot more leeway as now you have 512 switches per sample to represent the low level signal in the original CD signal. This is absolutely a ridiculous simplification that's not completely representative of what really happens but I think it captures the issue of accurate low-level signal in DSD DACs. If it's not an issue, Emm Labs wouldn't have designed their new DAC to run at DSD1024.

 

 

Sent from my iPad using Computer Audiophile

[ElviaCaprice]

All types. USB. Not in the US and don't have any of those home audition options open to me.

 

If looking for a USB DAC.

 

1. Can operate without the need for 5V vbus (use Intona for better galvanic isolation).

2. Plays all formats

3. Is capable of an effective upgrade in power supply. Best can work with an LPS-1.

 

The older Chord Hugo's fit this billing, but the volume control is manual. Can find one used for around 1K (make sure it doesn't need the 5V vbus for flashing).

[R1200CL]

Wait... Tidal is going to stream MQA next year? I missed that one, can you elaborate?

 

http://www.stereophile.com/content/mqa-and-warner-real-scoop#ktTfzhDHoVrpUC3W.97

 

You may interpret the above interview differently than me. As they states:

 

"When will I be able to stream or download it?

Aye, there's the rub: No one knows, not even Stuart. MQA's focus is on feeding MQA files back into the supply chain, he told me. It's up to the labels to decide whether and when to make them available for download or provide them to streaming services. So far, none of the major labels have announced their distribution plans."

 

Personally I think MQA is for streaming, and I hope record companies don't think different, and would try to sell you the the music one more time, and hold back on streaming availability.

 

We don't have to start another MQA discussion in this thread.

 

But I think people who think about purchasing a new DAC, should take this in to their consideration. Like I guess no one would purchase a DAC without USB interface these days.

 

Which also is an interesting discussion as we now in addition to the MicroRendu also possibly have superior stand alone USB/SPDIF converts, that most likely is better than those inside most DAC present.

Yes, I'm thinking about the Singxer F1.

 

Not to mention i2s interface as a possible requirement for a new DAC. Which I don't know has any benefits or why it exist.

[R1200CL]

If looking for a USB DAC.

 

1. Can operate without the need for 5V vbus (use Intona for better galvanic isolation).

 

I guess if you have the MicroRendu (as Firedog has) you would not need the Intona, right ?

 

As I understand Vbus from the MicroRendu done by the LPS-1 vs separately supply the same 5 Volt direct from a separate LPS-1 will not make any difference.

 

You either have a Vbus DAC, or a DAC with separate PS. Either DC in the range 7 to 12 Volt, or AC.

 

So you probably are saying, do not chose a Vbus DAC ?