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

Member
Joined 2007
Paid Member
Maybe adjust a delay in VLC player ?

Thanks Eldam, this does work very well with VLC Player.

Still there is an issue with other sources, online TV for example.

So I thought I might cover this as follows:

I would use 2 pieces of Twisted pear OTTO II (1:2 - two way i2s switch, one after the USB bord to be able to send the I2S signal either to IANs board or directly to the DAC and the second switch prior to the DAC as 2:1 switch).
In my case, the DAC has a relay switch anyway, so I only need one OTTO. The other option were to put a second OTTO between FIFO board and Clockboard, which would give the opportunity to use the pcm board also in "video mode".

This should bypass FIFO in case of playing videos.

Anything negative to expect?


Ernst
 
FIFO or Clock broken ?

Hello Ian,

I need help please t debug a problem.

After three times listening sessions, My Fifo 2 or clock 2 is (are) not working anymore !

All was good, before I switched again the Front end to my AYA 2 DAC the forth times ! Aya & your Front end are not on the same plug so are switched on one after the other in no particular order !

all the leds (fifo, isolator, clock) are green but there is a pop at the switching ON, the red light at the input of the Fifo immediately swap from red to green, the same on the clock board !

The source is a WaveBoard I/O : I rechecked in direct with the same uf-l cable : it works !

But now when I reconnect the front end : fifo + isolator + clock : the streamer doesn't see anymore the front end like it did the first three times !

All is powered with serious TS7A boards with 3.29 V for the clock board and 4.99 V for the Fifo (isolator are feeded at the two sides by each of the two boards and the light are green) !

Is there case of broken FIFO, how to check if the FIFO is broken or the Clock board (I checked twice the connections, changed the uf-lcable by new ones, the dac and the Wave I/O are working fine, I surmise an electric choc when I switched On the front End after the DAC or vice-versa the fourth times !)

Is there an order for switching on the front end ? And above all how to solve the situation ?! Is it so fragile ?

Help please, I was appreciating what I listened to the first three times :)

regards

Eldam
 
It works again !

despite new uf-l plugs : it seems if one of the uf-l plugs doesn't see by contact a signal , the input red led of the Fifo doesn't turn red, either the clock board side. So moving the uf-l cables on their plugs help (I tied several times before it works ! I surmised tired uf-l plugs on the pcbs )

But if you turn the gulty ufl- cable on the good plug : the clock board green leds (all the leds) flash and the after coming back stable (green without flashing), the streamer see again the I2S signal. It was stable green before despite all the I2S signal were seen !

Conclusion : only one uf-l cable or plug with a bad contact suffice to hide the front end to the soft streamer.

False alert ! changing all the ufl cable was not enough, uf-l soldered plugs (pcbs ones) are fragile after several connections.

:) I like a lot what I listen to : far better than the v1 Dam DAC I solded (never tried the other filter or the other layouts).

And I didn't pluged the I2StoPCM yet !
 
Last edited:
Hello Ian,

I am thoroughly enjoying the fruits of your labour with the latest clock and fifo boards working hard in my system - thank you!

Exciting news about the multi-channel / DSD FIFO. I am interested to try DSD. Are you able to give us anymore information about it? I am particularly interested to know:

- Whether it will be possible to re-use the existing, latest clock board?
- Will we need new clocks for higher rate DSD playback?
- Will it be possible to use in stereo as opposed to multi-channel?
- If a new clock board is needed, would it be possible to retain the dual mono outputs on the new board?

All the best to you,
Crom
 
Hello Ian,

I am thoroughly enjoying the fruits of your labour with the latest clock and fifo boards working hard in my system - thank you!

Exciting news about the multi-channel / DSD FIFO. I am interested to try DSD. Are you able to give us anymore information about it? I am particularly interested to know:

- Whether it will be possible to re-use the existing, latest clock board?
- Will we need new clocks for higher rate DSD playback?
- Will it be possible to use in stereo as opposed to multi-channel?
- If a new clock board is needed, would it be possible to retain the dual mono outputs on the new board?

All the best to you,
Crom

Hi Crom,

I'm glad that you are happy with my FIFO KIT in your system:).

Thank you for your interest in my current multi-channel FIFO project.
Just let me answer your questions:

- Whether it will be possible to re-use the existing, latest clock board?
:New multi-channel FIFO has 8 DSD /14 I2S channels with 7 DATA and one LR/DATA signals. So it need a new clock board to achieve those new features. However XOs on the old clock board can be re-used for sure.

- Will we need new clocks for higher rate DSD playback?
:Any XO works for old clock board will work for the new one, as well as higher rate DSD playback. 45/49 or 90/98 are recommended as long as your DAC MCLK works with those frequencies.

- Will it be possible to use in stereo as opposed to multi-channel?
multi-channel FIFO works for stereo playback. You just need picking up only two channels in this case. However re-mapping stereo into multi-channel can also be taken into consideration as a new feature.

- If a new clock board is needed, would it be possible to retain the dual mono outputs on the new board?
:It is possible. The only thing is that more u.fl sockets will be mounted on board for dual mono applications, since we are all happy with the high quality coaxial cables for DSD/I2S signals.

At current stage, I'm still open to suggestions. Please let me know what do you want to see if I can achieve.

Regards,
Ian
 
Hi Crom,

I'm glad that you are happy with my FIFO KIT in your system:).

Thank you for your interest in my current multi-channel FIFO project.
Just let me answer your questions:

- Whether it will be possible to re-use the existing, latest clock board?
:New multi-channel FIFO has 8 DSD /14 I2S channels with 7 DATA and one LR/DATA signals. So it need a new clock board to achieve those new features. However XOs on the old clock board can be re-used for sure.

- Will we need new clocks for higher rate DSD playback?
:Any XO works for old clock board will work for the new one, as well as higher rate DSD playback. 45/49 or 90/98 are recommended as long as your DAC MCLK works with those frequencies.

- Will it be possible to use in stereo as opposed to multi-channel?
multi-channel FIFO works for stereo playback. You just need picking up only two channels in this case. However re-mapping stereo into multi-channel can also be taken into consideration as a new feature.

- If a new clock board is needed, would it be possible to retain the dual mono outputs on the new board?
:It is possible. The only thing is that more u.fl sockets will be mounted on board for dual mono applications, since we are all happy with the high quality coaxial cables for DSD/I2S signals.

At current stage, I'm still open to suggestions. Please let me know what do you want to see if I can achieve.

Regards,
Ian

Thanks Ian, the ability to play stereo DSD as well as stereo PCM is my prime motivator here.

I definitely would like to retain the dual MCLCK outputs from the current board.

I would also like dual mono outputs as per the current clock board.

It would be useful to be able to off-board the LEDs so that the could be mounted on a front fascia and used to identify DSD/PCM and the rate.

The SMA connectors used for the MCLCK connectors are much better than the u.fl (in terms of reliability after a few insertions, ease of deployment and the ability to use better cable) so I would like to see more use of the larger connectors - potentially as a replacement for all the u.fl?

Many thanks and my name is on the order list as soon as it's available!
All the best
Crom
 

TNT

Member
Joined 2003
Paid Member
FIFO shall be the last thing before the DAC using shortest possible u.fl cabling. No isolator between DAC and FIFO. Isolator before FIFO. Power feed the things on each side of the isolator from sources that don't share ground i.e. from 2 different transformers/rectifiers/LDOs.

//
 
Last edited:
I2S RECLOCKER BIT PERFECT TEST

Hi Ian,
I am also working on I2S reclocker.I want to check following tests.I have some doubts.

1. what is bit perfect test? How to do it?

2. In the thread you have mentioned default mclk=256*Fs. I understood that 256=4*64bit (Left and right channel width).But I didnt get that , how 4 comes here.

3.how to minimize FIFO latency?


Thank You.
 
NDK DuCULoN status update

I think those who are interested in this thread may enthusiastically seek new information on better oscillators.

A Japanese oscillator manufacturer, Nihon Dempa Kogyo Co., Ltd. announced a new OCXO product, DuCULoN (Dual Crystal Ultra Low Noise), developed for digital audio application last week.
http://www.ndk.com/en/news/2015/1190895_e.pdf

Though its size is very large, 45 x 45 mm, the OCXO has an amazing low phase noise performance.
(Please look at a P/N graph shown in my attachment image.)

Bunpei had a chance to borrow their prototype units and to applied them to Chiaki's ES9018 Dual Mono DAC with SDTrans384.
Bunpei realized how strongly a sound quality of digital audio is controlled by clock performance, especially by phase noise feature. The best music in the system was obtained only and simply by replacing oscillators.

However, a list price for DIY users was not announced yet.
They say the product is under a law enforcement to inhibit a possible military use in some notorious countries.

NDK is ready to accept an order for "DuCULoN" OCXO products from foreign countries after their preceding domestic launch in a Japanese market.
(Six DIY users other than me in Japan already purchased the products in spite of high prices explained below and all of them are satisfied with resulting performances!)
Attached is a formal document of NDK OCXO “DuCULoN”.

You can request a quote of the products via Chip1Stop web shop;
Chip One Stop - Shopping site of electronic components and semiconductors.
(This link is set for country "USA". Please "change" a country setting if your country is "not USA".)

Please key-in one of the following part numbers into a search field box in the web page. (You can request a quote for more than two parts at a time by repeating the same operations.)

Bipolar: (This is my top recommendation.)
NH47M47LA-45.1584M-NSA3647C
NH47M47LA-49.152M-NSA3647C

HCMOS:
NH47M47LA-45.1584M-NSA3647A
NH47M47LA-49.152M-NSA3647A

Sine:
NH47M47LA-45.1584M-NSA3647B
NH47M47LA-49.152M-NSA3647B

First, I must tell about their pricing.
Their offering price for one piece may be between 1,550 USD and 1,650 USD or somewhat expensive than that! This price might be beyond the range available to ordinary DIY users.
However, I think resulting performances of the OCXOs are extremely superior. I’d like to recommend the use of the OCXO strongly as master clock generators for Ian's FIFO. You will surely recognize how rich music is stored in regular CD quality sources and how excellently your original DAC system plays them.

Bunpei
 

Attachments

  • DuCULoN_e_0106.pdf
    908.1 KB · Views: 185

TNT

Member
Joined 2003
Paid Member
its funny that in text they don't mention the close in jitter indicating that the don't really get the difference between running a high stability transceiver/receiver (classical osc. application) and the application in a DAC for audio. Like it would matter if the pn is -175 or -155 at 10kHz... figures look of course impressive. Still, very nice performance at e.g, 10 Hz etc.

//