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

To achieve higher performance and better sound quality, I finally decided to design my own FIFO for raspberry Pi based on my third generation FIFO technology.

I'd like FifoPi to have the following features and specifications:

1. Up to 768KHz full range I2S support;

2. DSD64,128,256,512,1024;

3. built-in 768KHz DoP decoder to enable RaspberryPi native DSD play back up to DSD256;

4. Isolator board will be included;

5. Clock board (isolated) will also be integrated, works with interchangeable XOs through standard sockets or adapters;

6. Supports full rang of XO frequencies from 11.2894 to 98.3040MHz;

7. Has two DC inputs for both isolated clean power and RaspberryPi power;

8. Has an additional non-isolated GPIO connector to make DAC controller working in isolated mode;

9. Works with good 5V linear power through on-board low noise LDO or pure direct 3.3V power from ulatr-capacitor or LifePO4 battery power supply;

10 Works with all RPi DAC HAT and external high-end DAC at synchronized master clock mode to remove jitter;

11. Optional external display panel with adjustable delay time;

12. DIY friendly with options and flexibility.


FifoPi_1
by Ian, on Flickr

Ian



Great news Ian!
 
amazing! all your goodies roll into one credit card size board.

what low noise ldo you used? what is the size of fifo?

only one master clock output for dac? does it
does it has second master clock output for beaglebone black and amanero usb in slave clock control mode like mcfifo? bbb and amanero can do dsd512 output, adaptor pcb board can be make so we are not restricted to just rpi

Thanks Minivan,

It has LT3042 ultra low noise LDO on board. But I'm highly recommended using 3.3V ultra capacitor power supply or 3.3V LifePO4 power supply directly as clean power to lower the jitter level even more.

It has a secondary mclk u.fl output at bottom side of PCB. There are two mclk outputs in total.

It works with your DSD music either from DSD direct or from DoP.

Regards,
Ian
 
To achieve higher performance and better sound quality, I finally decided to design my own FIFO for raspberry Pi based on my third generation FIFO technology.

I'd like FifoPi to have the following features and specifications:

1. Up to 768KHz full range I2S support;

2. DSD64,128,256,512,1024;

3. built-in 768KHz DoP decoder to enable RaspberryPi native DSD play back up to DSD256;

4. Isolator board will be included;

5. Clock board (isolated) will also be integrated, works with interchangeable XOs through standard sockets or adapters;

6. Supports full rang of XO frequencies from 11.2894 to 98.3040MHz;

7. Has two DC inputs for both isolated clean power and RaspberryPi power;

8. Has an additional non-isolated GPIO connector to make DAC controller working in isolated mode;

9. Works with good 5V linear power through on-board low noise LDO or pure direct 3.3V power from ulatr-capacitor or LifePO4 battery power supply;

10 Works with all RPi DAC HAT and external high-end DAC at synchronized master clock mode to remove jitter;

11. Optional external display panel with adjustable delay time;

12. DIY friendly with options and flexibility.


FifoPi_1
by Ian, on Flickr

Ian

Hi Ian - very interested in this project. I was waiting for the Allo Katana, but you have my attention :D

Keep us posted.

Thanks!

Whit
 
To achieve higher performance and better sound quality, I finally decided to design my own FIFO for raspberry Pi based on my third generation FIFO technology.

I'd like FifoPi to have the following features and specifications:

1. Up to 768KHz full range I2S support;

2. DSD64,128,256,512,1024;

3. built-in 768KHz DoP decoder to enable RaspberryPi native DSD play back up to DSD256;

4. Isolator board will be included;

5. Clock board (isolated) will also be integrated, works with interchangeable XOs through standard sockets or adapters;

6. Supports full rang of XO frequencies from 11.2894 to 98.3040MHz;

7. Has two DC inputs for both isolated clean power and RaspberryPi power;

8. Has an additional non-isolated GPIO connector to make DAC controller working in isolated mode;

9. Works with good 5V linear power through on-board low noise LDO or pure direct 3.3V power from ulatr-capacitor or LifePO4 battery power supply;

10 Works with all RPi DAC HAT and external high-end DAC at synchronized master clock mode to remove jitter;

11. Optional external display panel with adjustable delay time;

12. DIY friendly with options and flexibility.


FifoPi_1
by Ian, on Flickr

Ian
can this card work with the PCM1704?
It is likely that the "Universal I2S-PCM driver board " is required. Quite right?
 
Last edited:
To achieve higher performance and better sound quality, I finally decided to design my own FIFO for raspberry Pi based on my third generation FIFO technology.

I'd like FifoPi to have the following features and specifications:

1. Up to 768KHz full range I2S support;

2. DSD64,128,256,512,1024;

3. built-in 768KHz DoP decoder to enable RaspberryPi native DSD play back up to DSD256;

4. Isolator board will be included;

5. Clock board (isolated) will also be integrated, works with interchangeable XOs through standard sockets or adapters;

6. Supports full rang of XO frequencies from 11.2894 to 98.3040MHz;

7. Has two DC inputs for both isolated clean power and RaspberryPi power;

8. Has an additional non-isolated GPIO connector to make DAC controller working in isolated mode;

9. Works with good 5V linear power through on-board low noise LDO or pure direct 3.3V power from ulatr-capacitor or LifePO4 battery power supply;

10 Works with all RPi DAC HAT and external high-end DAC at synchronized master clock mode to remove jitter;

11. Optional external display panel with adjustable delay time;

12. DIY friendly with options and flexibility.


FifoPi_1
by Ian, on Flickr

Ian

Are symultaneus mode capable? (For tda 1541)
Thank
 
Ian ....Sorry :eek:
Again propose my question:
Can the PCM1704 be managed with the FiFoRpi card?
I think this can be achieved with the Universal I2S-PCM driver board with the FiFoRpi. It's correct?
Furthermore, the answer you gave about the simoultaneus mode with 1541 is unexpected.
Adopting the i2s-PCM driver board would not allow the simoultaneus with the 1541 anyway?
many thanks
 
It has LT3042 ultra low noise LDO on board. But I'm highly recommended using 3.3V ultra capacitor power supply or 3.3V LifePO4 power supply directly as clean power to lower the jitter level even more.

It could be that 1/f noise at low audio frequencies could be one of the biggest problems with RF LDOs and sound quality. In either the LT3045 or LT3042 data sheet, one of them, there is a suggested application circuit that uses an LTC6655 to improve the LF noise issue. Probably better than only increasing Cset. Also, although graphs of noise density at individual frequencies may look very good, we really should be thinking in terms of integrated noise over the frequency band that could affect sound quality for a given type of circuitry.
 
It could be that 1/f noise at low audio frequencies could be one of the biggest problems with RF LDOs and sound quality. In either the LT3045 or LT3042 data sheet, one of them, there is a suggested application circuit that uses an LTC6655 to improve the LF noise issue. Probably better than only increasing Cset. Also, although graphs of noise density at individual frequencies may look very good, we really should be thinking in terms of integrated noise over the frequency band that could affect sound quality for a given type of circuitry.

Yes - that shows up in the LT3042 datasheet. They mention taking Cset as high as 22uF to minimize LF noise. The LTC6655 is a relatively expensive part, and improvements would probably be swamped by poor PCB layput, lack of multi-layer (>2) boards, etc. At these levels, pretty sophisticated measurement techniques would be required to detect differences. I think that is why Iancanada is recommending an ultracap or LifoPO4 supply.

Cheers

Whit
 
Is it possible to feed +5V to the Raspberry thru power connector on RPi I2S cap PCB? I'd be happy not to use that micro-usb, or whatever it is, standard cable.


Hi, my Chinese Dac powered with 12V send power at an DC/DC converter to 5V ant send it at my RPI via RPI Main Connector, the RPI converts this power in 3.3 V (stabilise/filtered) and use that power for the system.


Dario
 
It could be that 1/f noise at low audio frequencies could be one of the biggest problems with RF LDOs and sound quality. In either the LT3045 or LT3042 data sheet, one of them, there is a suggested application circuit that uses an LTC6655 to improve the LF noise issue. Probably better than only increasing Cset. Also, although graphs of noise density at individual frequencies may look very good, we really should be thinking in terms of integrated noise over the frequency band that could affect sound quality for a given type of circuitry.

@Markw4

Totally agree with you. That's why I recommended the direct 3.3V ultra capacitor power supply or 3.3V LifePO4 power supply without any LDO involved.

People was more interested in the noise uV numbers, however the 1/f noise was overlooked.

Regards,
Ian
 
FifoPi with Buffalo-IIIse

Twisted Pear's new DACs have MCLK input. That makes it very easy to run them at sync clock mode to improve sound quality. The connections between FifoPi and Buffalo-IIIse are very straight forward, just four u.fl coaxial cables. Looks very clean.

I also started an all in one LifePO4 power supply project for this kind of applications. Hopefully I can put them working together very soon. I'll have more update then.
Develop ultra capacitor power supply and LiFePO4 battery power supply



FifoPi_BuffaloIIISE
by Ian, on Flickr

Ian
 
FifoPi makes ES9038Q2M DAC HAT working at pure sync clock mode. The MCLK u.fl cable is the only cable that needs to be connected for sound quality. Anything else will be plug and play.

ESS controller can be connected to the non-isolated GPIO of the FifoPi. I'm highly rerecorded doing it this way. With the FifoPi on-board isolator, ESS controller will be 100% isolated from ES9038Q2M DAC in this case. The processor of the controller will never have any galvanic impact to the DAC.

.....

Ian

Interested and following!