Ultimate Directstream


#1

I am posting my ideas for the future hardware revision of the Directstream DAC.

A couple of my ideas made it into the P10, so I am hopeful.

a. Max out the parts quality.

b. Three separate chassis for power, analog and digital.

c. Four Toslink jacks.

d. Volume knob and source selection knob.


#2

I agree that an ultimate DirectStream should have the best parts we can get (tho the current DS has pretty good parts almost everywhere.)

External power supply boxes aren’t necessarily optimal. We sure don’t want to share any power supply components between the analog and digital sections. For example if we connect the digital box to the analog box only with an optical cable we don’t want to cause ground loops with the power supply connections between a power supply box, the digital box and the analog box.

Four TOSLink jacks? Why not four S/PDIF, AES/EBU, I2S or whatever inputs? I agree that optical connections can be less noisy than copper ones, but I think that careful ground and signal isolation in boxes can work well. TOSLink also doesn’t support DSD, double rate DSD, etc.

I do like volume knobs. Also I’d probably put in auto source selection (like the DS Jr.)


#3

Any way to get a next generation FPGA that lets you double the sampling speed like Huron?


#4

The new Xilinx Artix 7 FPGAs do about 1.5 x as much per cell and have about 1.5 x as many cells per dollar than the Spartan 6’s in the DirectStream. So there’s about 2 x as many resources per buck. Doubling the sample rate again will still take some work since the Artix line isn’t twice as fast as the Spartan’s, but it’s almost certainly possible. On the other hand doubling the sample rate again won’t get the same amount of improvement than the first doubling. I have some ideas on how to use the extra horsepower for better sound in a few other ways: I’ll have to see if they pan out.


#5

Are we at the stage where we need another DS yet?

I thought Ted still had some ideas to try out before maxing out the potential of the current hardware. At least that was what was posted before.


#6

We definitely are not at the limits of the current hardware. I was just taking the first question literally and answering it. I know that the next release will further improve the sound quality and I’m pretty sure I can make a noticeable improvement after that too.


#7

Great news there is more improvement to be had with the existing hardware! Not looking forward to upgrading any part of my stereo at the moment, but will when the time comes. Couple more years of top-of-the-line sound out of the DS sounds great to me. The DS has been a GREAT value in this crazy hobby.


#8

Thanks @tedsmith for confirming more improvements to current hardware in the next release(s).

@amsco15 - am in full agreement. By the way, do you really live in Muddy Waters’ house?


#9
mbauer said

I am posting my ideas for the future hardware revision of the Directstream DAC.

b. Three separate chassis for power, analog and digital.

d. Volume knob and source selection knob.


b comment, In addition to what Ted said, 3 separate encloses adds considerably to material and shipping costs.

d comment, a maxed out DAC would not have a volume control as a great pre amp would be a better option if you want the best sound possible.


#10
yacheah said

@amsco15 - am in full agreement. By the way, do you really live in Muddy Waters’ house?


No, I live a couple suburbs west. His house was recently for sale and I would be very happy living in that house. It reminds me to be humble in all achievement. He was a monster influence on the culture and he considered Westmont paradise. Westmont is nice but his home couldn’t be more humble. It’s right across the street from a very busy rail line and is small. However, I’ve visited his house (a facsimile) in Mississippi and his previous home in Chicago’s Hyde Park neighborhood (you can still buy that house - tough neighborhood, not by the University of Chicago) and Westmont, by comparison, is paradise. It’s just not Malibu where Dylan lives and Muddy should have had enough money to live anywhere he wanted.

A few asides, I have talked to several people that knew Muddy. He was a big deal here in Chicago and lots of people are still alive that worked with him. Like with all people, the reviews are mixed. He did make some real money in his life. It’s easy to say he was taken advantage of. Maybe, however, wine, women and song took a lot of his money. The flip side to that is Dylan, and others, lived the same life and still had/have real wealth left over.

Westmont does celebrate its claim to fame. There is a very small museum in the city center with several artifacts and pictures. Some pictures exist of Eric Clapton at his house in the yard. I’ve lived in the Western Chicago suburbs, with the exception of a long 20 year plus stint in the Air Force, my whole life and never really knew where Muddy’s house was. I was listening to the Muddy Waters Chess set and saw in the accompanying book that the Rolling Stones sent a telegram to his house in Westmont, on Muddy’s birthday (60th?), asking him to attend their concert at the Chicago Stadium (mid 70s?). A limo would be sent. I thought that was really cool. Plugged in the address on my GPS and took a look.

Sorry for the ramble. I like telling the story.


#11
Ted Smith said

We definitely are not at the limits of the current hardware. I was just taking the first question literally and answering it. I know that the next release will further improve the sound quality and I’m pretty sure I can make a noticeable improvement after that too.


We love when you say such things smiley-music005_gif

#12
Ted Smith said

I agree that an ultimate DirectStream should have the best parts we can get (tho the current DS has pretty good parts almost everywhere.)

External power supply boxes aren’t necessarily optimal. We sure don’t want to share any power supply components between the analog and digital sections. For example if we connect the digital box to the analog box only with an optical cable we don’t want to cause ground loops with the power supply connections between a power supply box, the digital box and the analog box.

Four TOSLink jacks? Why not four S/PDIF, AES/EBU, I2S or whatever inputs? I agree that optical connections can be less noisy than copper ones, but I think that careful ground and signal isolation in boxes can work well. TOSLink also doesn’t support DSD, double rate DSD, etc.

I do like volume knobs. Also I’d probably put in auto source selection (like the DS Jr.)


Having several Toslink jacks gives me the opportunity to add several mid-fi sources that do not have careful isolation. No matter how well you design the DS, if you have a copper cable coming from an especially noise source, you have a weak spot.

You can add other inputs also, I don’t have a problem with that, but I want the Toslink.


#13
Ted Smith said On the other hand doubling the sample rate again won't get the same amount of improvement than the first doubling. I have some ideas on how to use the extra horsepower for better sound in a few other ways: I'll have to see if they pan out.
OK, so it may not be feasible in the next year or two, but I'd think eventually (maybe?) we'd have to start thinking about the next generation DS. Or is this truly as good as it can ever get?

#14

No, there are some things we couldn’t do in the DS because of cost: the most obvious one is that I’d like to have used high end Jensen Transformers. Doing better isolation of each input was also considered but would have added a bunch to the cost/price. On the other hand there are newer devices out there that can help and one can always do better the 2nd time.


#15

I certainly understand the OP’s thinking to some extent, but I just have to say Ted - for the first time in, what, 30+ years of owning products with DACs in them, and this is the first one I have no sonic issue with and feel no need to improve upon.happy-048_gif Thanks!


#16

My wish-for feature would be an external clock input for a 10 Mhz master clock. I suspect Ted will give a good technical explanation why an external master clock isn’t necessary in a DirectStream, and I’ve read what Paul has written on the subject (he is luke warm about them). I’m using a master clock on my new DAC and it makes a world of difference over the internal clock. Mind you, the master clock alone is 4 times the price of a DS…in digital, timing is critical. A master clock reference built around a low phase noise Rubidium oscillator in a temperature regulated oven that is frequency stable to 0.05 ppb might be expensive, but if you’re building a SOTA DAC, then having the option of the external clock has merit.


#17
Ted Smith said

No, there are somethings we couldn’t do in the DS because of cost: the most obvious one is that I’d like to have used high end Jensen Transformers. Doing better isolation of each input was also considered but would have added a bunch to the cost/price. On the other hand there are newer devices out there that can help and one can always do better the 2nd time.


Esoteric owners are blowing $40k on a pair of Grandioso D1 mono-DAC, most of those will be sync’d to a $22k Grandioso G1 master clock. Add another $3k for the clock cable that connects them both. If your aiming point is a SOTA product then the gizzards of your DAC should be cost-no-object.

I have the latest Esoteric network DAC sitting side-by-side with a DirectStream Junior. As much as I like the sound of the DSJ, and all those other things to like about the DSJ, sonically there is a night-and-day difference between them. Point being, if you want to do SOTA then you really need to spend the big dollars. I think DSJ was extremely well executed at its price point…I’d like to see what Ted could come up with if budget wasn’t a consideration.


#18

I certainly hope the common sensically priced version of the DS won’t be replaced by a stratospherically priced one.

I wouldn’t be able to afford it.

There are other makes out there for stuff like this. dCS for one.


#19
yacheah said

I certainly hope the common sensically priced version of the DS won’t be replaced by a stratospherically priced one.

I wouldn’t be able to afford it.

There are other makes out there for stuff like this. dCS for one.

I suspect that Ted is capable of producing a state-of-the-art dac at a still reasonable cost. Of course it will be more expensive than the Directstream, but it doesn’t have to be $60k or $100k. Maybe Ted can do it for $10k or $12k.

Ted’s design is very cost efficient. He may be able to make us a $100k dac for $10k.

And you can wait and get a used one.

Ted will probably avoid an external clock. It doesn’t really make sense. Everything has jitter and the further you remove the clock signal from the place of conversion the more you degrade it. Ted will probably just stick with the current approach: better, simpler, cheaper. Maybe upgrade the crystal.


#20

I’d certainly see if there were a better crystal available if I were doing such a product - there wasn’t last I checked, but somewhat ironically people keep using smaller versions, which have more phase noise (not a lot more, but …)

I get a kick out of “Rubidium clocks” in ovens for audio. The temperature doesn’t change in a crystal fast enough to affect the phase noise that we care about for audio (or at worst adding just the insulation from an oven based crystal will do the same thing as a full oven as far as audio is concerned.) And (once again, as far as audio goes) rubidium clocks are just a crystal that’s backed up by a rubidium clock. But the rubidium part is only for long term stability, so it’s a waste in an audio product. Don’t get me wrong some external clocks are better than some other internal clocks, but they would be even better if they took the build quality (and budget) of a rubidium clock, left off the rubidium part and just used the internal crystal (or even a better one.)

And yes, using a 10MHz external clock would definitely be worse than the current internal clock, using either a well done analog PLL to track the 10MHz clock or using a clock rate converter both add much more jitter than a good internal clock. At the very least if the tracking PLL was implemented with a great crystal, it wouldn’t be better being manipulated by tracking something. (Tracking implies changing the frequency, changing the frequency implies more phase noise, at least when the change happens.)