DIY AD1865 NOS DAC for Raspberry Pi

tru168

Member
2016-04-29 6:30 am
Hi all,

Just wish to share my DIY AD1865 NOS DAC for Raspberry Pi. I'm running Moode Audio in both Pi Zero and also Pi 2B. Both sounds great for a small DAC like this .

DAC3.jpg



A video regarding this project:
YouTube

Here's my page regarding this :
DIY NOS AD1865 DAC for Raspberry Pi - Audiophile Diyer
 
Last edited:
From a look at your web page (interesting projects!), it seems you're using passive I/V with the AD1865. Have you experimented with discrete active I/V? I ask mainly because its my understanding that current output R2R DACs (AD5547 is another one, as used by Schiit) run into significant linearity problems when the output's not a virtual short.
 

tru168

Member
2016-04-29 6:30 am
Hi,
Thanks for your comment.
Yes , you are right about the passive I/V stage characteristic in AD1865.

In this raspberry pi DAC project, my first prototype used a 220 ohm resistor as passive I/V, adjust the resistor value still can't make it sound better than the final version. Since the R-2R section output impedance is 1.75K, I'm sure it will increase non-linearity a little if I use passive I/V stage .

I have another SPDIF versions of AD1865 which is one of my experimental board, tried solid state I/V and sound better than passive one.

I finalised this raspberry pi versions with just simply using AD1965's internal Op-Amp and directly output without any coupling capacitor, this will sound better compare to simple passive I/V, and sound quality very near to modified Rudolf Broertjes' solid state I/V stage. I still like the SS I/V though ,just that I need another section of power supply for the SS I/V, and quite cumbersome for Raspberry Pi's application, thats why I end up with the internal Op-Amp.
 
I've gone through passive I/V and active I/V with my DACs (based on TDA1387 so far) and have ended up with step-up transformer as the best sounding solution. Though the particular implementation I've chosen uses many chips in parallel so I doubt its practical to use with AD1865 given that chip's not available at a 'throwaway' price like TDA1387. While TDA1387 is only 16bits its internal noise floor (worst case -86dB) goes down nicely with paralleling and it does have true current source (high impedance) outputs making interfacing that much easier.
 

tru168

Member
2016-04-29 6:30 am
I built paralleled TDA1387 , TDA1543, TDA1545 etc many years ago. I coupled them with DIY type of audio transformer too. Fantastic sound.
I never parallel any AD1865 before because its rare and expensive now. So far I'm happy with it and my SPDIF version will be driving audio trans - vacuum tube - output trans configuration. looks familiar ? ( Audio Note ?!)
 
Last edited:
Audio Note did have a patent on using a trafo in the output stage of a DAC, but it has expired so anyone's free to do that now. I seem to recall seeing a schematic with a 1:15 step-up.


Does the AD1865 sound better than paralleled TDA1387s btw? I've not heard an AD1865, curious to understand how it might be an improvement on what I've currently got.
 

tru168

Member
2016-04-29 6:30 am
Hi,

Yes , I tried several type of IV stage, and all sounds great to me. My favourite IV stage is directly connect the IV out from AD1865 with a 200ohm resistor as shunt resistor, and directly go into 6E5P tube , and then into an audio output transformer.
I ended up with just use the internal opamp in Raspberry Pi version because it sounds best in Pi Hat . I suspect that some IV stage will pick up noises bounced from the ground when use it in raspberry pi. I did try to use very old Raspberry Pi model B with IV stage DAC and the noise problem virtually non exist because inside the old Raspberry Pi Model B using all linear power supply , and others recent raspberry pi using a few internal switching power supply which can generate noises.
 
Hi, friends. Awesome project, Tru168! I hope I get to hear one of these.

Regarding the issue of switching regulator noise from the RPi, I got a nice improvement from an IanCanada IsolatorPi, though some noise still leaks into the I2S. A few people have reported further gains from bypassing the onboard RPi regs and using the LDOVR Mezzanine board. Probably the best option is the new all linear Allo USBridge Sig, but it isn't cheap!