Another review of the DS DAC

Audio Components DACs
61

One of the key design goals of the DS was to create a “DAC” which did not contain any sort of “DAC” per say, but instead it just passively converted a DSD stream to analog using a simple low pass filter. Unfortunately this approach necessarily will have a higher noise floor than will most other approaches.

This is one design goal which the DS DAC shares with some of the Lampazator DACs. The fundamental difference between a DS and (some of the) Lamps is that the Lamp essentially just takes an input DSD stream, and it low-pass filters it (turning it into analog) and then it buffers/amplifies this output using tube output stage.

The DS on the other hand takes all input signals, up-samples the input to 20x DSD and then passively low-pass filters the output, thereby transforming it into analog. But instead of using a tube output stage like the Lamp, the DS Sr uses a 100% passive low-pass filter stage with incorporates an output transformer as the buffer, and as a key part of its filter network.

This is largely a simplified version of same the basic design which Ted’s legendary prototype DAC used (although the firmware on the DS is much more refined now). This is the DAC which he originally pitched to PS Audio. And it is, by all accounts, supposed to be the very best sounding DAC that anyone who has heard it has ever heard.

Part of the magic of this design is that it performs a 100% passive conversion of the digital signal to analog. This is, in fact, what the “Direct Stream” in the name is all about. To not convert to analog this way is to have a DAC which isn’t a Direct Stream DAC. Unfortunately the price for this particular very deliberate engineering design decision is always going to be a higher noise floor.

But unless you are running this DAC straight into your power amp (which I do not recommend) , what does this even matter? This DAC is still just as quiet as was an older CD player containing an 18 bit DAC. And nobody ever complained about those players being too noisy back then, did they? In fact, Amir’s so-called objectivist press would-be peers of time in magazines like Stereo Review were fixated on even the worst CD players of the time having “perfect” sound. To them and that the newer generation of players which incorporated 18 bit DACs were completely unnecessary from their point of view because even the worst players measured “good enough” to them. So which way is it? Does any DAC with less than a theoretical 24 bit noise floor now suck, or was the 16 bit noise floor of a early CD players more than adequate as was claimed as the time? This is a fundamental contradiction which his work would seem to have with the claims which were made by many of the members of yesterday;s objectivist community.

Noise floor does not equal resolution. This is one of the areas where Amir and many of todays objectivist go wrong. Higher resolution is a requirement to have a lower noise floor, but the opposite is not necessarily true as signals can still exist well below the noise floor.

And the DS is still much quieter than any type of analog source, even when using noise reduction. And yet I don’t hear very many audiophiles claiming that 15 ips R2R sounds bad, or that a really quiet LP is way too noisy. Although perhaps Amir very well might maintain such a position? But since I don’t waste my time by closely following his work, I wouldn’t actually know.

6 Likes
62

I know we’ve had these conversations many times before but I still don’t understand how “most” (not all) measurements, especially those of a DAC, are all that relevant to an end user. I acknowledge they can be critically important to the engineer\designer of the product as they are trying to determine how all these numbers and graphs will affect the final sound signature of the piece. I get it with amp preamp and speaker matching but a Dac? I recently had a Brooklyn Dac. It was very nice and I’m sure it measured very well - better than the DS I’m sure. Problem was it gave me a headache and the staging was very much in your face. When I auditioned the DS there was no comparison, an effortless, liquid, non fatiguing and transparent flow to the music. Bettered the Brooklyn in every way. I just don’t see how these “I told you so” measurement assaults provide much, if any, value to those with highly resolving systems.

2 Likes
63

This was exactly my experience. I had a Mytek Brooklyn+ which is a ‘studio reference’ DAC. Front row seating, amazing clarity, hurt my ears after 20 minutes. I’m not exaggerating, it honestly hurt my ears due to the harsh highs. I sent it back.

1 Like
64

Yes and I’m sure it measures better than the DS. My guess is much better.

65

Regarding the review in question. . . .

i-don-t-care-i-love-it-8.png600×700 13.4 KB

8 Likes
66

tashnishedears has stated my feelings and intent well. I can lower the noise in the audio band and more and more aggressively filter the resultant ultrasonic noise: but everyone that has listened to that likes the sound less (tho I certainty expect that some would like the filtered top better.) If someone doesn’t like the DS’s sound then the DS isn’t for them, no problem, there are a lot of fine DACs out there (and TTs, etc.)

I could skip the transformer, but then the DS would integrate well into fewer systems and have more analog noise. Sure there are a lot of better transformers out there, but they cost too much for a product at this price point. I’m reasonably happy with the performance of the transformers we use. When I started the project I thought that transformers were crazy for this application and did a lot of work that avoided them, but the very first time I simply passively filtered a DSD stream with resistors, caps and a transformer I knew I was on the right track and I had to learn and changed my mind about transformers.

I’m not going to waste my time reading and rebutting the other stuff at his site, I’ve been there before and see how he treats people and how he learns. In any real conversation each party must expect to possibly change their mind. I know that I’m not going to change my mind based on anything amirm says and I know that he’s not going to change his mind based on anything I say. There’s no end in sight in such a circumstance and I don’t need to waste my time making him happy.

I wrote what I thought about his review earlier and knew full well that someone would point him to it. I have in the past privately emailed some more detailed rational for the choices I’ve made to some members of this forum, but some of those private emails were immediately posted on other forums so I have stopped answering technical questions to most people in email.

As anyone who has been here for any length of time knows I’m happy to answer questions that are asked in good faith and I think many of the above questions deserve a good answer, on the other hand I’ve answered most of the questions on asked on this thread before elsewhere (even when I knew that some of the posters were trolls.)

Some people thrive in a contentious environment and some people like a spirited debate, but I don’t enjoy either now that I’m no longer a teenager (no denigration or judgement intended about others who are different than I am) so I’m not going to read any more of this thread.

29 Likes
67

You’re saying it can be better? :slight_smile: ModWright are you reading this?

68

Exactly. I am the same as you Lon, all I really care about is the music.

I compared the DS against my Linn Klimax DAC. Linn DAC cost twice the price of the DS - the DS, to my ears, was/is a considerably better sounding dac: it’s as simple as that. Download new firmware - just like changing to a better DS dac for “FREE”. What’s not to like. What’s to discuss.

11 Likes
69

Please re-direct your questions to Ted - I only published a question to Ted along with his answer.
Relaying this info only makes me a messenger, not an expert.

And I published Ted’s answer hoping to bring better understanding of Ted’s implementation to re-assure owners of DSD Sr. that it is a great DAC.

And Ted, I have never posted any private emails on other forums: this is the first time I do such a thing, and I did it only on the PS Audio forum for the reason I stated above - I hope you’re not angry with my initiative as there was absolutely no bad intention on my part, just a wish to calm down people and not get all shaken up by this non-flattering review, so that they can keep enjoying their Direcstream DAC (as I do).

1 Like
70

Ted has admitted this from day one. The Jensen transformers which his prototype DAC used were too expensive for the price point at which this DAC was targeted. So they found some cheaper alternatives which still performed adequately, even if ultimately less well than the Jensens.

However for a modder to replace these transformers is more than just a drop-in operation as these transformers were deliberately chosen to have limited bandwidth. And this transformer bandwidth is an integral part of the overall low-pass filter design. So a replacement of the output transformers also involves a slightly less trivial re-design of the entire output circuit.

71

OK, so I didn’t stop reading this thread entirely.

No I wasn’t referring to your post here. I have no problem with that. I was referring to another situation where someone was cross posting responses between two separate sites.

Re Denafrips: From their specs:

THD+N 0.0010% (= 17 bits)
S/N Ratio -124dB (= 20.5 bits)
Dynamic Range >132dB (= 22 bits)
Stereo Crosstalk -110dB (= 18 bits)

3 Likes
72

Here’s a question about his review -

When he states that S/N is in the 90s, I believe that both Stereophile and Hifi News measured it a to be a bit higher. More like 17 or 18 bits.

Also, with his SINAD chart, is it possible that the lower than standard output voltage give the DSD a disadvantage compared to other dacs? One gets the felling he dings dacs having lower than average output voltage.

73

Great - I felt bad for little while.
Thanks.

1 Like
74

As Ted says, its a free world. These things are important to some people, like me, who prefers to buy things that measure well. The guy who did this test did lay the blame at the output transformer, Ted has quite openly explained that there was a cost issue, which I find slightly odd given what Paul says about using the best components and when I was considering the DSD DAC the UK price was equivalent to $9,000. How much can a transformer cost? $150 tops? The DSD DAC operates in 10X DSD or something. Mine in 24/192 PCM, but takes a completely different approach combining it with amplification, that I read but makes no sense to me at all. I honestly did not understand a word, except the bits about basically zero noise and distortion. So there must be at least 2 ways to skin this cat, probably many more.

Honesty, is any normal person expected to be able to read this?
" Standard systems typically operate on a cascaded circuit: current output convertor, Op Amp current-to-voltage conversion (transconductance), pre-amplification, and then amplification. Each stage in sound reproduction inevitably damages the integrity of the original sound signal. Magic Wire® does the exact opposite: transconductance, anti-aliasing filtering and system total gain all occur synchronically.
Magic Wire® works according to a feedforward model with zero load storage capacity, automatically keeping power at a constant while self-calibrating and compensating for converter dispersion. Magic Wire® massively overshoots manufacturer specifications. The outcome is nothing short of spectacular: The straight wire with gain."

75

I don’t want to guess how he measures, tho I disagree with some of the conclusions they drew JA’s and HiFi News numbers are close to what I measure.

76

The one’s I wanted to use cost about $170 in quantity (depending on which year you checked their prices). But I guess what’s not clear is that all companies have to earn money to pay their employees, etc. and each level of distribution similarly has to make some money. In the case of the DS the factor from cost to MSRP was about 5 or 6 times: Even with a price tag of $150 each that would have added $1500 or $1800 to the price of the DS. Our goal was $4k at the time and we couldn’t meet that number with the cost of the components I chose.

77

The DSD DAC was reviewed in HiFi News by Andrew Everard, the only reviewer I read with any regularity as he is the audio editor of Gramophone, the only magazine I buy (and have done for 30+ years). He’s pretty level headed, reviews a lot of good budget hifi, but was relatively gushing with praise for the sound from the DSD DAC. He reviewed it in 2018 and noted that the noise from the output transformer was much reduced from an earlier review.

“The transformer-coupled output increases THD at low frequencies (though 0.1%/20Hz is much reduced over earlier models). Interesting, THD over the top 20dB of the available dynamic range is fractionally lower (~1dB) via the Memory Player/DAC’s I2S connection (which retains its clock) than if driving the DAC directly via S/PDIF”

78

$170 x 5 or 6 is more like $1,000, but who’s counting? Actually Paul’s counting and his “cost plus” pricing seems to be under constant strain viz. AN3 speakers. The DSD has been under pricing pressure here, it was £6,000 when I was looking in 2014/15, then £5,500 and now £4,800.

79

I have a hard time convincing people that the clock doesn’t matter in the DS and the differences between inputs are mostly due to grounding issues and vary from system to system/setup to setup. Also it obviously wasn’t the transformer which changed but a bug in the software: it’s hard to guess what the cause of a symptom is from one measurement (or a small set of measurements.)

80

$300 x 5 is 1500 and as I said the real numbers are higher.

5 Likes