Perhaps a issue aswered ad infinitum out there...
I have a low cost Bluetooth to IIS board and I want to use with a dynosaur DAC since is fun to play with.
In that case I want to match the IIS board with a YM3433 SRC+AD1860 DAC.
A lot of info is out there, for example: http://www.pavouk.org/hw/modulardac/en_ad1865i2s.html
If I understood well, I need to delay n-1 bits of desired working. In the YM3433 case, I need then to delay 15 bits (since it is 16 bits).
The normal Bluetooth until 4.2 is maximum 16 bits, so is ok to play with.
Since all pages I fond are using NOS DAC, with separate L and R DAC's or data inputs, I deduce the YM will not needs the additional 32 bits delay, since it is channel serial like the I2S (L and R inputs at same pin). This will save parts.
Second, not much related... the YM3433 seems to be a SRC with a unrelated output clock. Not being a digital expert, I only deduce if the output is clocked to a different clock, the incoming jitter will be supressed at some level, perhaps? I'm sure about the ASRC making it, but not about this odd YM DF/SRC.
Will be fun to mix past and almost present if it work. And, I feel this will result better sound than the el-cheapo Bluetooth boards having internal DAC (I hope).
In the past I made a assembly of same board + a ES9023. This have a ASRC and worked flawlessy.
Thanks in advance for any hints/tips/advices. Remebereing, a ultraHi-End here is not the question (I know that will not sound like a direct feed DAC or a high-class Bluetooth solution), is more about the cool factor and the sound hope listed above. And I want to add this to a semi-portable thing I'm will make soon.
I have a low cost Bluetooth to IIS board and I want to use with a dynosaur DAC since is fun to play with.
In that case I want to match the IIS board with a YM3433 SRC+AD1860 DAC.
A lot of info is out there, for example: http://www.pavouk.org/hw/modulardac/en_ad1865i2s.html
If I understood well, I need to delay n-1 bits of desired working. In the YM3433 case, I need then to delay 15 bits (since it is 16 bits).
The normal Bluetooth until 4.2 is maximum 16 bits, so is ok to play with.
Since all pages I fond are using NOS DAC, with separate L and R DAC's or data inputs, I deduce the YM will not needs the additional 32 bits delay, since it is channel serial like the I2S (L and R inputs at same pin). This will save parts.
Second, not much related... the YM3433 seems to be a SRC with a unrelated output clock. Not being a digital expert, I only deduce if the output is clocked to a different clock, the incoming jitter will be supressed at some level, perhaps? I'm sure about the ASRC making it, but not about this odd YM DF/SRC.
Will be fun to mix past and almost present if it work. And, I feel this will result better sound than the el-cheapo Bluetooth boards having internal DAC (I hope).
In the past I made a assembly of same board + a ES9023. This have a ASRC and worked flawlessy.
Thanks in advance for any hints/tips/advices. Remebereing, a ultraHi-End here is not the question (I know that will not sound like a direct feed DAC or a high-class Bluetooth solution), is more about the cool factor and the sound hope listed above. And I want to add this to a semi-portable thing I'm will make soon.
Looking at its datasheet,
https://www.alldatasheet.com/datasheet-pdf/view/92436/ETC/YM3433B.html , it derives the output word clock from the input bit clock. The input bit clock must be a multiple of 16 times the input word clock, and that multiple must be in the range from 32 to 192. The output bit clock consists of bursts of an unrelated clock, as shown in the Serial output timing figure on page 6 of the datasheet.
All in all, it is meant for DACs that synchronize their output to the word clock and for sources with a reasonably low jitter bit clock. The sample rate converter does not suppress jitter.
Thanks for the info.
Just yesterday I got some 74HCT164 and test it someday next week to see how it behaves.
Just yesterday I got some 74HCT164 and test it someday next week to see how it behaves.
The AD1865/AD1860 needs a falling edge on its latch input at the end of the LSB being transmitted. They require 18 bits. So if the incoming signal is I2S it must be 24 or 32 bit I2S (not 16 bit), and you'll need to delay 5 or 13 bits respectively from I2S signal (6 or 14 bits for left-justified I2S).
With a right justified signal you simply need the left latch falling edge to correspond to the appropriate edge of the word select and the right latch falling edge to the other - in other words root the word select direct to one of the latches and inverted to the other. Again the signal must be at least 18 bits per word.
You could connect 16 bit I2S to the AD1860/1865, but a more complex circuit would be needed that fakes two LSBs as zero and delays 3 bit times for each of left and right.
Thanks for the reminder. I see Pavouk choosen the delay for matching the DAC chip and make the additional delays for acomplishing that (for the other channel).
Based by the frequencies I measured in the Bluetooth module, it have the typical 32 bits frame for each channel (in I2S).
Oh well... With this the bits will be aligned, and several people used it with DAC with success, but I dunno about the YM, if it simply will "ignore" the last "unused" 16 bits of the frame, until the LE changes for the other channel. Only a testing will show if it works...
For now I'm not be over adventurous to send the direct I2S to the AD1860, since I know the Bluetooth works with audio signal only to 44100Hz/16 bits on disponible modes, hence my tentative to use this filter I have laying out in some drawer (I'm not the usual NOS user; only if I can upsample in the PC or cellphone and send it "oversampled" to the DAC ;-).
I hope to have some noise free time this week, to test this.
Based by the frequencies I measured in the Bluetooth module, it have the typical 32 bits frame for each channel (in I2S).
Oh well... With this the bits will be aligned, and several people used it with DAC with success, but I dunno about the YM, if it simply will "ignore" the last "unused" 16 bits of the frame, until the LE changes for the other channel. Only a testing will show if it works...
For now I'm not be over adventurous to send the direct I2S to the AD1860, since I know the Bluetooth works with audio signal only to 44100Hz/16 bits on disponible modes, hence my tentative to use this filter I have laying out in some drawer (I'm not the usual NOS user; only if I can upsample in the PC or cellphone and send it "oversampled" to the DAC ;-).
I hope to have some noise free time this week, to test this.
Hmmm... if it simply "pass" the jitter, not sounds like a great solution for the el-cheapo board. Surely this will have high jitter; it even not have the audio frequencies crystals; only the typical 12MHz from these applications. Even so... if sound quality superates the all-integrated also el-cheapo solution, then it will serve to it's target use.
I tried this board only for fun some months ago, using a PMD100 and PCM63 based DAC: resulted in a terribly muted/closed in, recessive sound, perhaps due to lack of any jitter rejection (for that, for sure a PLL will be need). The sound was worst that the another all-integrated! Measured signal are bad with bluetooth, but are "ok" using a FLAC file in a card direct in the player.
Just take one of the cheap CS8421 ASRC boards, you have a reasonable jitter attenuation and format conversion included. Just needs some 16k 0805 resistors (in parallel) to configure it.
Sorry, I didn't look that up.
Correction: I thought you'll keep the YM3433, this DF is expecting 16bit R-J. Clearly, a YM3434 has much better DF specs, but anyhow, I would definitely keep the DF
Correction: I thought you'll keep the YM3433, this DF is expecting 16bit R-J. Clearly, a YM3434 has much better DF specs, but anyhow, I would definitely keep the DF
Last edited:
This week I got a weak flu and have some time to play with it.
The concept of delaying the bits, even with 32bit frame, worked wonders with the YM3433.
But, the jitter quality of the Bluetooth board are atrocious. No wonder why the PMD100 and PCM63 falied also. These I not measured.
Here some measurements:
Jitter from card player part:
Jitter from Bluettoth (HEEELP):
Another scale:
From one "all-in-one" solution (better):
And, for a reality check, one from the same AD1860+YM3433, but using a old horse called CS8412, via optical link:
So is correct, the YM don't provides any attenuation of jitter. And the AD reproduces the LSB toggle bit ok with the CS interface.
Now, the first board I tried, I made 2 years ago in a hurry using a ES9023 (async mode)(I not measured it due to lack of time - 2 days before travelling). Perhaps there the result are satisfactory, compensating a little this horror, since my friend linked it.
Maybe, if I insist to use this board, some money with the ASRC or a ES DAC will be needed...
The concept of delaying the bits, even with 32bit frame, worked wonders with the YM3433.
But, the jitter quality of the Bluetooth board are atrocious. No wonder why the PMD100 and PCM63 falied also. These I not measured.
Here some measurements:
Jitter from card player part:
Jitter from Bluettoth (HEEELP):
Another scale:
From one "all-in-one" solution (better):
And, for a reality check, one from the same AD1860+YM3433, but using a old horse called CS8412, via optical link:
So is correct, the YM don't provides any attenuation of jitter. And the AD reproduces the LSB toggle bit ok with the CS interface.
Now, the first board I tried, I made 2 years ago in a hurry using a ES9023 (async mode)(I not measured it due to lack of time - 2 days before travelling). Perhaps there the result are satisfactory, compensating a little this horror, since my friend linked it.
Maybe, if I insist to use this board, some money with the ASRC or a ES DAC will be needed...
