I do not need another DAC

[Fran]

But, who ever said I paid attention to those kinds of statements.

I've had an itch for a while (not he kind you can get a cream in the chemist for) to try a different kind of DAC. I have the ultra precise voltage out types like PCM5102 and ESS9018 based ones, and they sound pretty damn good in my opinion. Even more so when you get the digital signal end lined up right. Over the last few years, led by Sensei Nigel, a few of us built a really nice current output PCM1794 DAC. That build not only resulted in a fantastic sounding DAC, equalling and beating many mainstream offerings, we also learned a LOT. The curve was a line pointing straight up.

So the one that seemed obvious to try is the r2r type DACs and there are a good number of them on the market now. Denafrips, soekris and plenty more offer these now. I was turning all this over in my mind while browsing around and listening. I stumbled across a thread on DIYAudio about an old DAC chip, the AD1862. Now something went off in my brain as I had heard of that chip before, but I couldn't remember where. A bit of searching turned up an impressive pedigree - used in some audio note DACs, excellent reputation, and most of all I saw some people I know in the DIY world gushing over the sound from them. They are current output as well, which also generally is a good sign. The guy who started the thread had very kindly provided not only schematics, but also all the files needed to be able to order the PCBs to allow anyone to build the DAC. He also supplied the same files for a very nice power supply to go with it. As an extra bonus, the way the thing is laid out its very configurable.

In short I decided to go for it...... the first step (after reading around a LOT) was to order parts and the PCBs. Of course, this coincided perfectly with a bad run of COVID and snow in Texas - where the chosen supplier of all the components is. This led to a delay of about 3 weeks before the parts arrived.....

[Diapason]

Ooooh!

[Fran]

So how come all these r2r dacs you see have these arrays of tiny SMD resistors




But the AD1862 chip looks just like any other chip?



Well, here's the thing - there are no r2r dac chips made any more - back when the AD1862 and AD1865 were being made, they manufactured all those resistors that you see in the array all into a single chip, with each resistor being made with extreme precision. Naturally this cost quite a bit to do, so once delta sigma dacs came along, most manufacturing moved over to those. The only way of doing r2r now (other than obtaining the old chips) is to make it using lots of individual resistors. This is the system in the Denafrips, soekris etc dacs.

[Sligolad]

Fair Play Fran on getting this started here.

I have also joined this journey and so glad I did as I thought I was at my destination on the last DAC build but after hearing an early incarnation of this DAC put together on a piece of Plywood I was won over and it has been a revelation.

The DAC itself and the Main Power Supply are Slam Dunks and have been well sorted over on diyAudio https://www.diyaudio.com/forums/digital ... os-2r.html

The interesting stuff that Fran has been working on is the output stage options and so far some of the options have been amazing and there should be a flavour for all in the various options tested so far.

The final area where the work by Nige and Fran has taken us is on the input side where all the learnings with SD Players, reclocking, good power supplies and good clocks has really lifted this end product to amazing new levels of listening pleasure. As Fran said the ease of configuration changes makes it easy to tune the final sound to your preference.

I will post an image capturing a rough early outline of costs and configuration options under discussion at the moment when i figure out how to get images on here again.

[Fran]

A word on the output of this DAC.

Most conventional DACs now output a voltage - and many of them have circuitry built in that will drive amps and even headphones directly, and often all from a single 3.3V - that really is amazing. The humble PCM5102 gives really fantastic results in this category. But this DAC doesn't work like that. Instead of putting out a voltage signal that we can probably all understand, this one puts out a current signal instead - but only just 1mA each way which is also tiny. This means that the end design must have some way of converting that tiny current signal into a ~2V standard output level. This is done by passing the current signal across a resistor - and ohms law means that when a current flows through a resistor you will get a voltage. The correct size of resistor allows you to scale the voltage.

But of course its not as simple as that. The DAC chip has very limited ability to "drive" the resistor and whatever might come after that resistor. What is needed here is a current to voltage output stage (IV stage) and the best of them take the current, and insulate the DAC from the demands of the IV stage so it has no load to drive, and then convert that to voltage.

As Sligolad has mentioned the band are back together, and experiments are currently underway to test out some different output stages - from solid state to valve based to allow each to pick their own preference. I can say that based on what I've heard so far, this DAC has unparalleled performance. Simply some of the best I've heard from a digital source.

We might talk about the front end/source next....

[hudo]

Well, here's the thing - there are no r2r dac chips made any more - back when the AD1862 and AD1865 were being made, they manufactured all those resistors that you see in the array all into a single chip, with each resistor being made with extreme precision.

My DAC has a r2r chip thats still manufacturing - AD5791BRUZ - but Schiit was writing on headfi forum (i think) how difficult was to use that chip. Is it possible thats the only r2r IC made today!?
How will you do upsampling, dithering and all that heavy lifting required before the signal hits the DAC, or is it going to be NOS? Just thinking - many brands are bragging about how advanced proprietary algorithm they have for upsampling (Chord, others with FPGA) - what about building NOS DAC but using HQPlayer for upsampling since it has the most advanced algos inside?

Exciting project guys, hope I'll be able to borrow the masterpiece for a few days when its done:)

[Fran]

Hudo - it'll be NOS - with a filter at the end to remove ultrasonic hash. It'd be great to meet up at some stage to have a listen. I think we are all missing that at the moment.

I suppose in reverse order its time now to mention the source that feeds the DAC. In essence, this can be any source that spits out the common DAC language of i2s - only 4 lines are needed - bit clock, word clock, data and ground. Among our little group the main sources tend to be one of the various Xmos based USB cards, and most prefer to use a particular model of SD card player. These are so simple, and there is something in the SD card protocols that makes them sonically excellent. Typically all these are bought from the likes of Aliexpress for 30-50e each. we will be investigating methods to also have multiple sources to allow switching between 2 or 3 units maybe in time.

So that describes the 3 main sections - the i2s source, the dac itself and then the output stage. The last bit is the power supplies needed - but they will vary according to the output stage.

[hudo]

SD Card, which means copying files to the card, playing songs, then copying new files, playing, rinse and repeat ... ugh lots of work, almost like vinyl :) Just joking, but since i2s is also an alternative, thats how we can use streamers and Tidal/Qobuz I guess!
I got pi2aes RPi HAT (https://www.pi2design.com/pi2aes.html), which has a bunch of outputs, including I2S, and measures really well (https://www.youtube.com/watch?v=Jvu_doQfAI0), would be nice to test it with this DAC.
We can also compare it against Schiit Yggi DAC, which is also AD based r2r implementation.
Keep us updated with build progress:)

[Fran]

Will do. One of the real advantages here are the options presented - pretty much any i2s source will work. I think there are some devices that will put i2s over hdmi as well - not sure of the pinout etc but its an option for sure. It just takes bck, lrck and data (and ground) lines.

So the current to voltage output stage has a big influence on the final sound of the unit and here there are a number of them we plan on testing out. The stock DAC board uses a pair of opamps to do the job - and there are certainly changes to be got by swapping these around. Pretty any single opamp can be tried, and I tried several I had on hand. Of those the AD797 and op27 were my favourites, and these gave a bit more detail and deeper bass. Its not a huge difference, but worth it for the price of a couple of opamps. I didn't have any of the fancy discrete types to try, but conceivably there might be an improvement in trying some.

However, all advice points towards having a discrete output stage. The intention is to try some solid state ones, and then to try some valve ones as well. Already we have tried a lampizator one and the results are impressive......

[Sligolad]

Updated costs...had a mistake in the Power Supply components costs and corrected downwards!