Denafrips discrete R2R Multibit

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Member
Joined 2017
Paid Member
I think it's cool. My DAC is also R2R multibit. But my choice is not hybrid one .I prefer monolithic. Famous multibit PCM1704 is already obsolete. But industorial DACs ,for example LTC2642 etc, are availeble.

I think you can get good result with monolithic ICs.
 
I'm a sucker for good R2R Multibit dacs. And this is very cool looking under it's skirt. It just has to sound ?????????
https://www.head-fi.org/f/threads/denafrips-terminator-the-king-of-r2r-dac.851085/

Cheers George

I can confirm the Ares is a very good sounding R2R, and I have the Soekris 1021 V3. IMO, the Ares sound better. I might start a thread on the modding the Ares when I have the time to mod it (not much to tweak really as the power supply is pretty good, but maybe few extra caps here and there):)
 
Last edited:
And pay twice the price so you can wait a year for modders to improve on it? Lol.

Funny. But sadly a bit true. I did post to Seokris asking not to use a noisy 1117 for 3.3.v for both the clock and the fpga. The Ares at least uses pre-regulation and 5 lt1763 after for the digital section and 3 more for the resistor network.

And the new 1541? People pay extra for the linear supplies for digital, the Ares uses an O core tranny and a huge capacitor bank.
The Soekris 1541 has 3 Meanwell SMPS module :eek:
I hope those Meanwell are super low noise as switcher noise are really difficult to get rid of.
 
Member
Joined 2009
Paid Member
Funny. But sadly a bit true. I did post to Seokris asking not to use a noisy 1117 for 3.3.v for both the clock and the fpga. The Ares at least uses pre-regulation and 5 lt1763 after for the digital section and 3 more for the resistor network.

And the new 1541? People pay extra for the linear supplies for digital, the Ares uses an O core tranny and a huge capacitor bank.
The Soekris 1541 has 3 Meanwell SMPS module :eek:
I hope those Meanwell are super low noise as switcher noise are really difficult to get rid of.

First, the Si514 and Si570 clocks have internal regulators, see Silicon Labs AN255 for performance details (The described Si550 use same architecture), so their power supplies are not that critical....

And it's actually easier to filter a little high frequency noise that the huge 50/60 hz magnetic field from a transformer, I have no problems with it at all.

dac1541: High reliability high efficiency switch mode supplies, following by LC filters, following by high speed, ultra low noise, ultra low impedance discrete regulators, a total of six of them. Digital clocks powered by LP5907's.

Most others: Transformers with large magnetic fields, followed by inefficient linear regulators using standard chips, with a shitload of unreliable electrolytic capacitors trying to fix things....

So which one is the smart way of designing and will work like new after ten years of service ?

And btw, you do know that the Ares do not have a reclocking fifo, so the 50 pS jitter from the AK4118 spdif receiver goes right though the DAC ?
 
And it's actually easier to filter a little high frequency noise that the huge 50/60 hz magnetic field from a transformer, I have no problems with it at all.

Filtering out normal-mode noise is fairly easy, enough chokes and ceramic caps ('lytics for damping too) and you're done. Presumably that's what you're referring to. Common-mode noise on the other hand....

Most others: Transformers with large magnetic fields, followed by inefficient linear regulators using standard chips, with a shitload of unreliable electrolytic capacitors trying to fix things....

The Denafrips does have a most impressive capacitor array. I wonder what circuit aspect is so sensitive to PSU noise that they need that many?

So which one is the smart way of designing and will work like new after ten years of service ?

SMPSUs have unreliable electrolytics inside too, probably running at much higher internal temps than the array visible in the Denafrips.
 
Its not the distance in itself that's responsible for adding jitter, its the noise voltage between the grounds of the oscillator and the DAC. Which is increased by distance normally, but is also increased by larger noise currents and is decreased by having lower ground impedance.
 
Thanks Abrax for those inputs.

Not sure I understand ? Do you mean zero current or at least the two active devices should not share/cross the same currents on the Gnd plane because those last having different impedance despite number is very low ?

So maybe Gnd islands tied together more than a plain Gnd plane or cuts in the plain Gnd plain to drive the currents and avoid crossings? Or XO close to the DAC chip area ?

Is such a targett can be achieved by isolator chips, or are they just for noise in voltage domain ?

Btw sorry, totally off topic
 
Last edited:
Member
Joined 2009
Paid Member
Filtering out normal-mode noise is fairly easy, enough chokes and ceramic caps ('lytics for damping too) and you're done. Presumably that's what you're referring to. Common-mode noise on the other hand....

The Denafrips does have a most impressive capacitor array. I wonder what circuit aspect is so sensitive to PSU noise that they need that many?

SMPSUs have unreliable electrolytics inside too, probably running at much higher internal temps than the array visible in the Denafrips.

Yes, the electrolytic capacitors is the first part to fail in a SMPS, can tell you that from experience.... But those new Meanwell SMPS have a MTBF of 1.5M hours, very very impressive. And people tell me that they really are good. Time will tell of course....

But MTBF goes down as part count goes up, that's the issue with all those electrolytic capacitors on the Ares....
 
The Denafrips does have a most impressive capacitor array. I wonder what circuit aspect is so sensitive to PSU noise that they need that many?

SMPSUs have unreliable electrolytics inside too, probably running at much higher internal temps than the array visible in the Denafrips.

Isn't this done to lower the esr to a minuscule amount?

Agree, the single worst component in SMP's are the caps, they s**t themselves, no matter what the quality.

Cheers George
 
Not sure I understand ? Do you mean zero current or at least the two active devices should not share/cross the same currents on the Gnd plane because those last having different impedance despite number is very low ?

Its tricky to visualize - the oscillator has its own power supply, the DAC has its own power supply. Normally both will be regulated with low-noise devices. Both regulated supplies normally come from the same transformer winding which is what leads to noise being introduced in the 0V. One way to solve the problem is to have the oscillator run from its own dedicated winding then join the two 0Vs at the DAC, this ensures no external ground currents can pollute the 0V between the DAC and oscillator. So yeah, isolated supplies are the best solution I think.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.