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

[qusp]

Originally Posted by roger57

I thought this was the " Asynchronous I2S FIFO project, an ultimate weapon to fight the jitter" forum...?

someone mentioned using SMPS with this 'asynchronous i2s fifo', I explained my reasoning why not, but where I thought it was not only appropriate, but superior. i've contributed quite a lot of on topic conversation and support on this i2s fifo project over the last 18 months or so. I feel zero guilt for the short diversion.

was your post on topic?

[Adrculda]

Back to my SMPS argument :P
It would be lighter and more convenient / compact rather then a toroid / battery combo no?

Benefits, would hardly outweigh the use of either if properly setup and filtered in my opinion.

Since no more than 2 or 3 amps is required and voltage is low...
It would be a cheaper or at par ( including an Audiophile filter section on the output ) but a lot lighter and simpler to put in slim case.

[qusp]

you are free to pursue it for curiosity, I see no benefit and only extra problems to solve, a linear supply to power it is pretty small and dwarfed by the rest of the build. an isolator board was developed for this design with part of its mission being isolating the clock from the USB ground and its SMPS, why would you think people would be interested in putting one back in?

I find it exceedingly doubtful you can produce a quieter SMPS supply, so it can only be noisier. the fifo takes only ~150mA, so it really doesnt need that much space for anything, a tiny transformer, a tiny regulator and 2 x A123 battery cells is nothing. the only thing that would motivate me is higher performance and it just isnt there. lighter? having my dac 250gm lighter? I dont often carry it with me.

it doesnt have to be a battery/reg combo, if you are happy with the sort of performance SMPS gives then you wouldnt need the hassle of batteries, tiny linear regs powering the onboard regs would suffice.

[iancanada]

Originally Posted by Adrculda

Back to my SMPS argument :P
It would be lighter and more convenient / compact rather then a toroid / battery combo no?

Benefits, would hardly outweigh the use of either if properly setup and filtered in my opinion.

Since no more than 2 or 3 amps is required and voltage is low...
It would be a cheaper or at par ( including an Audiophile filter section on the output ) but a lot lighter and simpler to put in slim case.

Hi Adrculda,

Here is the noise density plot of TPS7A4700. But the noise performance is still not as good as a decent battery pack. The whole goal of the FIFO is to reduce the clock jitter. It gains neither from the high efficiency nor from the compact size of a power supply. With the isolator board, the power consumption is just around 100mA for the clock board and half of isolator, the only thing helps is the low noise performance of a power supply.

Theoretically a switching power supply is not good at low noise application comparing with linear. The high frequency high power switching noise is very hard to be decoupled, even using multi-stages CLC filter, because there is no ideal capacitor and inductor in the real world. The EMI noise introducing from ground into DAC may even more difficult to deal with.

But I might be wrong. Please let me know if your can really find a switching power supply with competitive noise performance.



Regards,

Ian

[crobbins5421]

If using isolator board with FIFO and single clock board, to power the clock board with a single 3.2V LiFePO4 battery one must remove L12 and U5 (with 0R resistor between pins 1 and 5). This will allow me to connect at J12 correct?

[iancanada]

Originally Posted by crobbins5421

If using isolator board with FIFO and single clock board, to power the clock board with a single 3.2V LiFePO4 battery one must remove L12 and U5 (with 0R resistor between pins 1 and 5). This will allow me to connect at J12 correct?

isolator powered from J9 on the clock board side. It works good with dual xo clock board. To use a single clock board, you have to figure out flying power wire to power the isolator clock side. Please refer to the schematic of the isolator board.

For single xo clock board, yes, if your use 3.2V battery, you have to remove U5 and short the LDO footprint with a 0 ohm jumper.

Ian

[Fridrik]

Should I expect a 'sonic improvement' if I use the isolator board in this case?

[iancanada]

Originally Posted by Fridrik

Should I expect a 'sonic improvement' if I use the isolator board in this case?

Theoretically yes, because ground loop is removed.

Ian

[t27049215]

Hi,
I am new to this kit, and after reading this thread for a while, some things are still not clear to me, hope somebody could help. Thanks in advance.

What is the delay time since input?
There is a table in the manual: "FIFO half full delay time". Is this the answer?

If there is some delay, then there will be lip sync problem when watching videos. I'm concerned of this.
Is it there a "low latency mode" switch to toggle so the entire buffering process could be bypassed?
I'm not demanding, just asking if this is feasible and easy to implement.

[qusp]

toggle entire buffering board? sure, you can put a suitable switch at the input and bypass the entire fifo.....

of course there is a delay, the delay is based on the samplerate, it is how the fifo works, it cannot be bypassed and still have the buffer in circuit. the only thing you can do is delay your video by the same amount.

this is for high quality audio, if you want to make it work with video i'm afraid the ball is in your court.

it would be really good if people would use the 'search this thread' function, the number of times we have answered this question is getting quite large.