AK4490 DAC Upgrade

[Stuart Polansky]

Well, at least we hope it's an upgrade. Several years ago, while recovering from shoulder surgery and using whatever was on the shelf, I cobbled up a DAC using the DIYINHK AK4490 board, XMOS I2S USB converter and a sample rate board. No driver needed for Linux or Apple machines. Sounded pretty good, several club members have heard it.

Since the initial build, and with the use of two hands, made a couple of minor upgrades. Changing the original AD827JN chip, which many feared was fake because of the cheap board price, to an LME49720 chip from Tom P. resulted in no audible (to me) improvement. Later, the on-board el-cheapo RCAs were removed and okay quality jacks mounted on an external plate. At the same time, a Corcom line filter was added and things tidied up a bit. Put the "chassis" which is a 1/2" thick piece of plastic on spikes. Noticeable improvement in sound.

Fast forward a few years and I'm gonna try a tube output stage for the DAC. Not sure if it will be an improvement, but figured I'd give the op-amp (DC coupled) version a fair shake by improving components to the extent practical, given the tight board. To that end, the AC terminal strips were removed (for physical space), diodes replaced with Shottkys. Original 3-pin regulators: LM317, LM337, LT1085-3.3 and LT1085-5 were each replaced. The LT1085s were assumed to be fakes again, but if they were real, it was a wasted effort. The LM317 was replaced with an LM1085, and the LM337 with a LT1033. Caps were replaced with real film and foil (2.2nF) and metalized (1nF) polypropylene caps, about twice the physical size of the originals. Resistors will be 1/4 watt PRPs once they arrive from Parts Connexion. The output jacks will be upgraded again and the wires feeding them will be Cardas 17.5. Also, the regulator output electrolytics will get film bypasses.

The whole mess will go into a little plastic instrument case with a 1/2" plastic plate under the DAC and I2S boards.

Hopefully, after the holidays, this will be complete. If any sonic improvement is heard here, I'll be anxious to find a Covid-Fear-Free home to listen with others and compare to a known good DAC.

Will post updates as I toddle along!

Stuart

Attachments

[Stuart Polansky]

New components

Attachments

[tomp]

Not surprised there was no sonic difference when replacing the Op Amps. Both are way better than what is needed for audio work.

[David McGown]

Tom, Stuart,

In my personal experience, the LME49720 is just OK, there are better sounding op-amps. The TO-99 metal can version of the same op-amp is much better sounding, I replaced the 8pin DIP with a metal can version on a Browndog adapter in a Topping D50S in my music studio, and it had a more relaxed, tonally balanced sound with better detail and more depth of image. I suspect the shielding of the metal can helps significantly. I have also had good luck with the Muses 8820 (bipolar) and Muses 8920 (J-fet input), both of which are now available pretty cheap at Mouser. I used these to replace LM49720 in a Marchand electronic crossover. The well healed can try the Muse02 and Muse03, but they are pretty expensive in comparison, one can get discrete opamps like the Burson or Sparkos for not much more. It does get crazy, like chasing after 6SN7s.

I think the AK4490 is a voltage output DAC, so the op-amp is a unity gain differential to single-ended converter, with an LP filter network. Should be able to do that in tubes pretty easily.

David

[SoundMods]

Just a note to say that operational amplifiers can be designed and built with vacuum tubes. Yes -- tubes take up a lot of space -- but OH the sound!

Here is but one example:

Attachments

[David McGown]

Just a note to say that operational amplifiers can be designed and built with vacuum tubes. Yes -- tubes take up a lot of space -- but OH the sound!

Here is but one example:

Something is not right about that circuit as drawn. Perhaps I do not understand it, but there seems to be missing plate resistors. Also, as drawn, the grids are tied together, probably a graphics error.

Broskie has a Unbalancer circuit. Note that both plates of the input section are loaded. The foll

http://www.tubecad.com/2011/03/blog0203.htm

User's guide:

http://glass-ware.com/User_Guides/Unbal ... %20-PS.pdf

This might do what is needed, though it is not unity gain.

David

David

[SoundMods]

Just a note to say that operational amplifiers can be designed and built with vacuum tubes. Yes -- tubes take up a lot of space -- but OH the sound!

Here is but one example:

Something is not right about that circuit as drawn. Perhaps I do not understand it, but there seems to be missing plate resistors. Also, as drawn, the grids are tied together, probably a graphics error.

David

You'll notice that it is an op-amp with complimentary +/- 300-vdc power supplies. Different rules.

[David McGown]

Walt,

Still, unless the inputs are cap coupled to the preceding stage, the +/- grid inputs need to be at 0VDC. The cathode will be at a slightly positive voltage, with the 220K resistor to the -300V supply (resulting in 1.33mA current draw for both sections). So, in the first section, the plate for the - input is tied directly to +300V plate supply, and the + input has a 220K plate resistor. Lets assume that the grids tied together is an error, since the cathodes ARE tied together. So how does the circuit work from a DC perspective and keep the inputs at 0V? My first thought was that the schematic is in error. The circuit would be in DC balance IF there is a similar 220K plate resistor for the (-) input side, which means that both sections would have 150V drop through the plate resistors, 150V across the tube, the input grids at 0V. Both sections would have the same operating point (given matched sections). However, I did check with other schematics online, and am chastised that the schematic you posted IS correct for the Philbrick K2-W. I am stumped on what is going on.

David

[FerdinandII]

What most of us see as "Diff Amps", in this circuit are actually "Current Mirrors".
Pretty common in the op amp world, and you're right, they just don't look right in a tube circuit....

Walt,

Still, unless the inputs are cap coupled to the preceding stage, the +/- grid inputs need to be at 0VDC. The cathode will be at a slightly positive voltage, with the 220K resistor to the -300V supply (resulting in 1.33mA current draw for both sections). So, in the first section, the plate for the - input is tied directly to +300V plate supply, and the + input has a 220K plate resistor. Lets assume that the grids tied together is an error, since the cathodes ARE tied together. So how does the circuit work from a DC perspective and keep the inputs at 0V? My first thought was that the schematic is in error. The circuit would be in DC balance IF there is a similar 220K plate resistor for the (-) input side, which means that both sections would have 150V drop through the plate resistors, 150V across the tube, the input grids at 0V. Both sections would have the same operating point (given matched sections). However, I did check with other schematics online, and am chastised that the schematic you posted IS correct for the Philbrick K2-W. I am stumped on what is going on.

David

[SoundMods]

Walt,

Still, unless the inputs are cap coupled to the preceding stage, the +/- grid inputs need to be at 0VDC. The cathode will be at a slightly positive voltage, with the 220K resistor to the -300V supply (resulting in 1.33mA current draw for both sections). So, in the first section, the plate for the - input is tied directly to +300V plate supply, and the + input has a 220K plate resistor. Lets assume that the grids tied together is an error, since the cathodes ARE tied together. So how does the circuit work from a DC perspective and keep the inputs at 0V? My first thought was that the schematic is in error. The circuit would be in DC balance IF there is a similar 220K plate resistor for the (-) input side, which means that both sections would have 150V drop through the plate resistors, 150V across the tube, the input grids at 0V. Both sections would have the same operating point (given matched sections). However, I did check with other schematics online, and am chastised that the schematic you posted IS correct for the Philbrick K2-W. I am stumped on what is going on.

David

Any problem -- including electronics design -- can be dealt with at least six different ways -- chances are that all of them will work.