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

Hi Branko,


My current setup is Volumio > Amanero Combo 384 > ReceiverPi > Fifopi > Buffalo III-SE Pro (sync mode). The setup works really well without any dropouts or signal lock issues.

You do have to keep u.fl cables from FiFopi to DAC as short as possible so all would have to be in the same enclosure.


Hi,

maybe I am to late for the party:

I tried to read as much as possible. But the questions keep coming :)

At the moment I use a Buffalo III-SE Pro (ESS 9038) DAC with Mercury I/V stage form twisted pear.

I feed it with Roon to their BeagleBoneBlack(with botic driver) -->Hermes--> Chorus combo with quite some problems. (When it is working it sounds great)

I am quite happy with the DAC and would like to optimize the "streamer"-side. So I would like to go the Roon --> RPi4-->FifoPi--> ReclockPi --> ??? way.

My main questions are: what is the best way to connect the FifoPi to the Buffalo for I2S.

Is there an option for an extra Streamer case or has it be in the same case as the DAC?

Thank you

Branko
 

TNT

Member
Joined 2003
Paid Member
The old spdif intreface i/o board.... from the pdf...

- I2S input to DIT:
J6: PH 2.0 mm 7-pin connector, configured as


Is it really J6? Isnt it J8 that would be i2s input? Typo in the PDF in the text but OK in the picture?

How do I use the board as a i2s -> toslink converter? Do I ned MLCK or is it enough with the 3 i2s signals?

//
 

Attachments

  • j8.jpg
    j8.jpg
    406.1 KB · Views: 399
StationPi

I am finally implementing StationPi, to act as a platform for a Rpi/FifoPi/ReclockPi digital platform. At first I missed the note that FifoPi still needs it's own clean power supply.
Am I correct that the ideal is a 5v PS for the Rpi, another for FIFO dirty side, a very good 5v for Fifo clean, and a separate 3.3v supply for ReclockPi?
 
I am finally implementing StationPi, to act as a platform for a Rpi/FifoPi/ReclockPi digital platform. At first I missed the note that FifoPi still needs it's own clean power supply.
Am I correct that the ideal is a 5v PS for the Rpi, another for FIFO dirty side, a very good 5v for Fifo clean, and a separate 3.3v supply for ReclockPi?

Hi,

in my setup for the STATIONPI, i use 5v for the rpi, 5v for the fifopi dirty side, 3.3 v for the fifo clean side, 3.3v for the reclockpi

regards
 
Last edited:
Thanks Tubo. I had missed the need to power the FIFO clean side.
It was weird in that everything worked but no lock or clock select. I checked every voltage and realized there was no power on the clean side. Checked every trace when it occurred to read the directions. Duh!
As usual a good design. I'll use same power supply with separate regs for Rpi and FIFO dirty side. Then separate PS with individual regs for clean side and ReclockPi. Should really isolate clean and dirty.
 
Member
Joined 2018
Paid Member
Remember there are R58 and R59 on the bottom of FifoPi to be jumpered to hard bypass the onboard LDOs to feed 3v3 clean side. Apparently the 'best' way if using a very good 3v3 psu!
This powers the clocks and obviously one can then separate the power pins to individual clocks and feed them separately.
I'm sure you all already know this!
 
Ian,

Is there an estimate when the sinepi will be available?

Good question!


The SinePi is almost down. I re-design the PCB for 7 times until I'm happy with the result. Too many jobs to get the best performance.

I'm trying to make it available soon. The only thing I'm concerned about is the cost for such a low quantity production. But no worry, I'll figure it out.

Good weekend.
Ian
 
Disabled Account
Joined 2020
Disabled Account
Joined 2020
I am finally implementing StationPi, to act as a platform for a Rpi/FifoPi/ReclockPi digital platform. At first I missed the note that FifoPi still needs it's own clean power supply.
Am I correct that the ideal is a 5v PS for the Rpi, another for FIFO dirty side, a very good 5v for Fifo clean, and a separate 3.3v supply for ReclockPi?

I think the initial discussion took place here:

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

+- the next 10 posts.

I may start resurrecting the StationPi project in a few months as the optional capacitors are starting to arrive @ Mouser end of August. It was about a 4 month wait to get it back in stock. So if you plan to use the tiny optional capacitors with the StationPi, it may be good to pick up off Mouser because of supply shortages.
 
Last edited:
Thanks for the reference to the PS @A123.
I am a fan of the old Blackgate caps around digital. I still have a good supply from when they were cheap. I used the optional cap spot to put in some BG F and BG N. I got everything up and running today on the bench. Just finished listening to a couple of tracks using headphones on a commercial amp. Even with an NDK SDA clock and Denon amp it sounds very promising. I'll swap it into the audio room with proper clock and amps tomorrow and get a good listen.
 
Disabled Account
Joined 2020
Blackgate works. Well played. I will consider that when it's time to tinker down the road.

The well-mastered (Gold Discs, Target, CBS, etc) redbook PCM is just fantastic with the ReClockPi and a low latency realtime OS. A well-mastered track is like adding a ReClockPi on top of a ReClockPi. The Audiophile tight layered bass that only a low latency system can provide teases the senses when paired with Ian's hardware. It's just a super quick tight bass that's impossible without low latency. Addicting. That was the main reason I wanted to switch to 12V, but now that low latency OS is available for RPi I was happy to stick with RPi.

I have no idea how the StationPi and 5 MHz clock will scale when I implement EOY. Exciting times.

Good luck with the swap and let us know how it works out...
 
Last edited:
Disabled Account
Joined 2019
worths a try if you have no Black Gates

@A123,


If non having Black Gate that are NLA or hard to find. I found sometimes Cornell DUblier FCA Acrylic smd caps were a good substitute - while not as good/ different- sometimes for decoupling purpose VS a BG Std. I know IanCanada used such tech underneath the FiFOPI with the Rubycon equivalent, but I have no experience with it. The Panasonic SMD equivalent technology is not sounding as good from a subjective hearing point of view on power supply lines. Anyway this little FCA cap can be found in 0805 size and with a main Panasonic 47 uF SEP near the load or better APSA100ETD470MFA5S gives a good combo enough for digital, than can be heard in the final sounding result often.
 
Last edited:
It is never a good idea to use one type of cap for everything. Each component has it’s own sound signature due to the construction and used materials. Using the same caps everywhere will push this sound signature to the front.

Black gate is fantastic if you use the right types on the right spot. I still have a small box of them waiting for the final stage of some of my designs (which can take some time....)
And paper in oil for is the perfect choice for power supply in tube stuff (western electric if you can find them) together with the unbeatable BG WKZ.

But for available electrolytic caps my choice is Elna Silmic for analogue and oscon for digital.
These are both very good but do not try to use them in the wrong place, oscon in analogue is not good, just as silmic in digital.
And then we have the ultracapacitors, unlocked to use safely with the nice boards of Ian. These are best in ps of digital and hf (like clocks) places.

Film caps are a separate (and much more elaborate) story.

Just my 2 cents......