Asynchronous I2S FIFO project, an ultimate weapon to fight the jitter

[TNT]

So, where is your gain?

 

[Zen Mod]

re-read my posts ;

using I/V stage , you "just" re-locate I/V resistor away from DAC chip output , to output of I/V stage

so - if you have - with 4 paralleled Sabre outputs 15mApp at disposal , you're forwarding that same current swing to output of I/V stage

then , choose what you'll make with it - 2Vrms , or 30Vrms ....... matter of choice and construction

 

[TNT]

So how is one location better than the other?

 

[Zen Mod]

you're in Moulin Rouge ;

is it better sitting and observing from 10m from stage ............ or 60m from stage , using binoculars ?

take binoculars as optical amplifying device

in second case , you're letting image to fade , then you're amplifying it

edit : take Pass D1 as example ; there are several threads , explaining it

 

[TNT]

I/V is voltage gain.

 

[Zen Mod]

anyone said different ?

 

[TNT]

case 2: DAC - I/V stage without active voltage gain - buffer , to have 1Vrms per leg

 

[Zen Mod]

I explained that later - with info about re-locating place of I/V resistor

so , there is no active part , contributing to either voltage or current gain , entire output voltage is developed with exact amount of DAC's output current , across one resistor (per leg , of course )

 

[TNT]

perpetum mobile

 

[Zen Mod]

as I said - look at D1 output stage

 

[TNT]

I feel I have a competence gap.

 

[clivem]

Ian,

If I want to "tap" 3.3V from the FIFO board to drive the isolator stacked on top of my Amanero, (which is driving the FIFO), where is the best place to get it from? Is one of the TP(oints) on the FIFO board downstream of the 3.3V regulator?

 

[AR2]

you're in Moulin Rouge ;

Yes, I am... Go on...

 

[TNT]

 

[hochopeper]

Clive,

if the isolator on the amanero is only sending i2s to the FIFO the FIFO inputs are 5V tolerant I'm pretty sure, so you can send 5V back to the isolator ... there is a 5V output that you can get to ... look up the manual. Unless you're sending i2s via the SPDIF input board, in that case you might need 3.3V I'm not sure the DIX9211 takes 5V inputs.

Cheers,
Chris

 

[znarfomat]

The i2s output voltage at the WaveIO isolator is 2,6V and the i2s voltage after the dualXO board is 1,6v. Does this mean anything!?! Might this be the problem for my distorting DAC?!? I didn´t try the PCM board yet!

hi apoo, did you look at the signals with a scope? if so, 1,6v is certainly not enough. otherwise it's meaningless.
if you have the PCM board, I think you should use that to bypass Doede's flipflop circuit. 1794 can accept several formats and PCM board can most certainly make one of those.
you can also run the PCM board in half speed mode, so you get 512fs rather than 1024fs (at 44.1 kHz).

 

[ryanj]

Hi Everyone,

Good to see this project getting the attention it deserves in the audio world, its a great product.

I've been enjoying it in my system for over a year or so now with no need or feel that i should improve on the sound quality anymore.

But i remember Ian saying battery power over any type of regulator will give great results. So I installed a simple 6 V lantern battery to power the whole FIFO.

To put it simply - on a few recordings, mainly live acoustic, it sounded like a different recording that i was listening to, it was that much better.

The music became much more coherent, so now i want to try running the XO directly off a lifepo4 cell.

To do this ill need to populate my isolator board and install it, and my battery management board. So i guess i have a lot of work to do.

Does anyone have parts for these boards leftover? Im not overly keen on paying the premium for only a few parts online. If anyone can help that would be great.

Anyone else using the FIFO with the TDA1541A in NOS?

Ryan

 

[Ken Newton]

I explained that later - with info about re-locating place of I/V resistor

so , there is no active part , contributing to either voltage or current gain , entire output voltage is developed with exact amount of DAC's output current , across one resistor (per leg , of course )

Zen Mod,

Your general question has at least two fundamental axis of topology. One axis, is passive I/V versus active I/V. The other axis is feedback versus non-feedback. These axis lead to the following four basic configurations.

1) ACTIVE I/V with FEEDBACK:

Many of us who have experimented with DAC design and implementation have concluded that feedback (voltage or current) based active I/V circuits typically implemented with IC op-amps objectively measures the best, yet subjectively sounds the worst. It optimally provides the D/A chip's output with the lowest impedance virtual ground. Although, it should be mentioned that simulations usually show this virtual ground to become inductive at high frequencies due to the decreasing open-loop gain of the circuit. In general the problems with feedback circuits attempting to process very wide-band signals are well documented. Band-limiting the D/A chip's output signal current helps here, but that is not necessarily easy to do without increasing the virtual ground impedance.


2) ACTIVE I/V without FEEDBACK:

This is exemplified by the D1, Zen I/V, Jocko Homo I/V, and the AD844, among others, and commonly classified as current-conveyors. These are easy to identify because they typically feature a grounded-base/gate input stage. As would be expected, THD is objectively higher than with their feedback based equivalent circuits above, as is the virtual ground impedance. However, the lack of feedback being applied to the wide-band input signals seems to to produce a subjectively more natural (or, less annoying) sound. In addition, these open-loop circuits can be made to generate very low THD too if enough bias current can be run through them.

One implementation reservation that I personally harbor regarding this type circuits is that the full A.C. signal current swing of the D/A chip must be linearly handled by the supply regulators. The amplitude of the signal current can be quite significant, reaching tens of milliamps peak, depending on the particular D/A chip and whether that chip has been paralleled, which is often done to lower D/A chip noise. This concern can be eliminated if the current-conveyor features active constant current-source connections to the supply rails. This can be easily incorporated via the inclusion of a folded-cascode element to the circuit.


3) PASSIVE I/V with FEEDBACK:

A resistor connecting the D/A chip's current output to ground, followed by a feedback based voltage amplifier of medium to high gain. Among the advantages over the active non-feedback current-conveyor I/V circuits is that the D/A outputs can be presented with a constant, non-inductive load impedance. The main disadvantage probably is the relatively significant value of that load impedance, which I've seen as high as several hundred ohms on some such DACs. Typically, higher the load impedance seen by the D/A chip outputs, the higher the THD. Even so, I've subjectively obtained very good sound this way.

One of my diy DACs utilizes a PCM1794A + 75 ohm resistor passive I/V + simple first order low-pass ultrasonic filter. The resulting 200mV pk. audio signal is then A.C. coupled to my power amp (yes, without any preamp nor volume control in between), which features an op-amp feedback based input stage. Simply glorious.


4) PASSIVE I/V without FEEDBACK:

Resistor connected to ground, as above, followed by an open-loop medium to high voltage gain stage. To my knowledge, most commercial DACs featuring a triode analog stage are based on this topology. It features the same advantages and disadvantages as above, except for having higher THD, which is particularly evident if the voltage gain stage is solid-state. I've not yet experimented with this topology, but would definitely go this way if I wanted to incorporate a triode analog stage.

 

[Zen Mod]

so , per your classification , I was talking about type 2) ACTIVE I/V without FEEDBACK , as my choice for Sabre

in fact , one path is buffers -xformer , as AR2 tried , another path is 2)

we'll hear what's more mellow for ears

 

[Greg Stewart]

TPS7A4700 Board Input Caps..

Ian (or anyone else),

Have you tried the Panasonic SP Polymer caps instead of the Tants there? Owen and Hochopeper both mentioned them, with Owen suggesting he preferred them to the Tants for his up-coming shunt reg board.

I'm building up the bare reg boards I got in one of the last GBs and am considering the Panasonics over the Tants

TIA!

Greg in Mississippi