Analog Devices DIY tube DAC

[ssstromar]

Hello

I may be new on this forum but i'm not new to audio and electronics. I'll keep this short for the first time. I want to build a tube output DAC based on Analog Devices (or maybe BurrBrown) DAC. I decided to go that way becouse I do not want to copy or recreate an existing design based on many times described TDA1541(A) and TDA1543(A). Tube output is a "must be" for me. Anybody can create a bad sounding DAC based on any chip so i asked myself why not try to build a good one with the newest technology available.

Q: Any suggestion on the DAC selection (as mentioned earlier AD or BB maybe Cirrus) with Delta-Sigma OS 24(?) bit.

Thank you in advance,

Jernej

PS: Please don't mind my bad english

[phn]

Originally posted by ssstromar
why not try to build a good one with the newest technology available.

Valves aren't exactly a new technology.

[rfbrw]

Originally posted by ssstromar
Hello

Q: Any suggestion on the DAC selection (as mentioned earlier AD or BB maybe Cirrus) with Delta-Sigma OS 24(?) bit.

Thank you in advance,

Jernej

PS: Please don't mind my bad english

The AD1853 from AD seems to be popular amongst the makers of expensive dacs and though that is not necessarily a good reason to use them, it is as good as any other reason. The AD1852 which has a voltage output would make the use of valves a little easier. You might also want to consider the dual differential example given in the AD1852 datasheet.

[ssstromar]

The basic idea for the project is to use as few as possible components on the analog outputs. We all heard some of very good sounding CDP with directly connected outputs. No analog filters used, no buffers (OP-AMPS).

I would use only simple RC filtering on the output and a cathode coupled diferential pair of maybe ECC83 valves. Valves operate only in theory well above 100kHz or even less so they are natural filters. For the analog section I would definetly use P2P with silver coated OFC (UPC) solid core wire.

AD1852 has voltage outputs so it is the first choice for the project. AD1853 has the same arcitecture but it's a I-DAC so scratch that (or do I ?).

What benefits have mono design compared to single chip conversion for both chanels.

[rfbrw]

Originally posted by ssstromar

What benefits have mono design compared to single chip conversion for both chanels.

A 3dB improvement in the noise figures.

[Vidalgo]

Originally posted by ssstromar
... to use as few as possible components on the analog outputs.
... No analog filters used, no buffers (OP-AMPS).

AD1852 has voltage outputs so it is the first choice for the project. AD1853 has the same arcitecture but it's a I-DAC so scratch that (or do I ?).

Voltage output DAC chip = I/U opamp inside of the chip. Cost-effective solution. Still need an external filtering stage.
Current-output DAC chip = external I/U, with or without opamp. Perfomance solution.

[Holger Barske]

Current-output DAC chip = external I/U, with or without opamp. Perfomance solution.

Absolutely agreed. You could for example use a transformer solution for I/U conversion.

[ssstromar]

A 3dB improvement in the noise figures.

Will consider in further development.

As we know AD1852 and others have diferential outputs which are in 95% used only to be summed up with an OP-AMP output stage. OK some CDPs have balanced outputs as well but many users simply connect them to SE input stages of amplifiers via interconect cables which only propagate positive signal and ground.

So, come to think of it that my idea of diferential pair of valves in theory isn't that bad, DAC outputs will be evenly loaded so can be valve outputs regardless of SE or balanced load.
Or in theory why not use only one DAC diferential output becouse all of benefits which come with diferential pairs are only used to the OPAMP, beyond that is SE.

CS8414 seems to be the most common choice for the input reciever.

[Holger Barske]

CS8414 is out of production, you might go for CS8416.

[rfbrw]

Originally posted by ssstromar


So, come to think of it that my idea of diferential pair of valves in theory isn't that bad, DAC outputs will be evenly loaded so can be valve outputs regardless of SE or balanced load.
Or in theory why not use only one DAC diferential output becouse all of benefits which come with diferential pairs are only used to the OPAMP, beyond that is SE.

By all means use a single chip, but an opamp is not the only way to sum differential outputs. It can be done with valves, fets or transistors and the improvement remains whether or not the following equipment is balanced or not.