Well take a closer look at the alternative filter - then read up what Julian Dunn had to say about equiripple filters and time smearing. That'll make you more of a DAC expert instantly 
I have only a little experience of listening to PCM1792 - its sound I think would depend to a large degree on what you adopt for I/V. I have one in my Asus soundcard and modded the onboard I/V with promising results. In short, don't use what's on the TI datasheet, either in opamps or component values.
I have only a little experience of listening to PCM1792 - its sound I think would depend to a large degree on what you adopt for I/V. I have one in my Asus soundcard and modded the onboard I/V with promising results. In short, don't use what's on the TI datasheet, either in opamps or component values.
Hello Patrick
I am interested in this ESS9022 DAC and have a couple of questions that I hope you might be able to answer.
I want to replace the SAA7323 Bitstream DAC in my CDP and have thought about using a DAC with I2S output from the SAA7310 Decoder / SAA7220 digital filter in the player. I can access the Word Select, Serial Bit Data and Serial Bit Clock from the 7310. However, the 7220 is an oversampling filter and looking at the 2 page ESS9022 data sheet available to download, I see that the there is an oversampling filter on boars the chip.
My questions are whether the ESS9022 DAC is suitable to use as I have suggested and, if the ESS9022 is suitable, would I need to bypass/disconnect the TDA7220? The clock is through the DAC and then distributed the the decoder and filter. Would I need to transfer the clock insertion?
My Rotel is modified as much as I can, with low noise regulators for each chip (including transport servos/drivers), Tent clock, new discrete FET op-amps (based on the N. Pass DIY op amps). I think the bitstream DAC is the weak link.
Any help would be appreciated.
Thanks,
Joe
The ES9022 has built in oversampling and can accept clock rate up to 50MHz.
The table inside the datasheets tells you what data rate and resolution it can support.
It can also accept synchronised and asycnronised data (e.g. when using 50MHz XO).
So it is rather flexible.
I do not know the SAA7220 in depth and how it is configured in your CDP.
Thus I do not feel comfortable at all giving you any advice.
I can only hope that someone who knows the chip can help you further.
Patrick
The table inside the datasheets tells you what data rate and resolution it can support.
It can also accept synchronised and asycnronised data (e.g. when using 50MHz XO).
So it is rather flexible.
I do not know the SAA7220 in depth and how it is configured in your CDP.
Thus I do not feel comfortable at all giving you any advice.
I can only hope that someone who knows the chip can help you further.
Patrick
Hi Joe - I'm fairly familiar with the SAA7220 but I'm rather confused. The SAA7323 was designed as a replacement to the SAA7220/TDA1541 combination and as such contains the digital filter function previously implemented by the SAA7220. So I can't figure out why any player would have both chips.
Do you have the service manual or would you indicate what the model number is?I hope my enquiry is not too far off topic.
Hello Patrick,
I can respect that. Thanks.
Hello abraxalito
The player is a Rotel RCD965BX. The chips are shown on the schematics as:
U111 SAA7220;
U106 SAA7310;
U112 SAA7323.
The I2S connections, as best I can understand from the schematic and data sheets are connected as follows:
I2S word select 44.1kHz (WSAB): 7220 pin 1 and 7310 pin 4 to 7323 pin 30 (WSI)
I2S serial clock (CLAB): 7220 pin 2 and 7310 pin 3 to 7323 pin 31 (CLI)
I2S serial data (DAAB): 7220 pin 1 and 7310 pin 2 to 7310 pin 32 (DAI)
The system clock is 11.2896mHz connected to pins 25 and 24 of the 7323 DAC and then distributed to the rest of the board. I had thought to bypass the DAC completely, but it supplies the system clock.
I recall seeing a discussion where the SAA7220 was bypassed and the I2S feed taken directly from SAA7310 to a TDA1541. The DAC then operated as NOS.
However, the associated hardware and PSUs took up a lot of room in the player's case. I don't have that much room left with an extra 3 transformer/PSUs and a Clock PSU taking up most available space. I do have access to a very clean 7V supply to, potentially, feed the ES9022.
As I think about this, in my rather non-technical way, if the I2S connections as outlined are suitable for the ES9022, I could, bypass the 7220 filter and leave the 7323 DAC in place. This would leave the system clock to untouched and remove the 4 X over sampling of the 7220. I could then remove power to the analog output section and the ES9022 could sit on a little board near the 7310. Or is that too easy?
Cheers,
Joe
Hello Patrick,
I can respect that. Thanks.
Hi Joe - I'm fairly familiar with the SAA7220 but I'm rather confused. The SAA7323 was designed as a replacement to the SAA7220/TDA1541 combination and as such contains the digital filter function previously implemented by the SAA7220. So I can't figure out why any player would have both chips.
Hello abraxalito
The player is a Rotel RCD965BX. The chips are shown on the schematics as:
U111 SAA7220;
U106 SAA7310;
U112 SAA7323.
The I2S connections, as best I can understand from the schematic and data sheets are connected as follows:
I2S word select 44.1kHz (WSAB): 7220 pin 1 and 7310 pin 4 to 7323 pin 30 (WSI)
I2S serial clock (CLAB): 7220 pin 2 and 7310 pin 3 to 7323 pin 31 (CLI)
I2S serial data (DAAB): 7220 pin 1 and 7310 pin 2 to 7310 pin 32 (DAI)
The system clock is 11.2896mHz connected to pins 25 and 24 of the 7323 DAC and then distributed to the rest of the board. I had thought to bypass the DAC completely, but it supplies the system clock.
I recall seeing a discussion where the SAA7220 was bypassed and the I2S feed taken directly from SAA7310 to a TDA1541. The DAC then operated as NOS.
However, the associated hardware and PSUs took up a lot of room in the player's case. I don't have that much room left with an extra 3 transformer/PSUs and a Clock PSU taking up most available space. I do have access to a very clean 7V supply to, potentially, feed the ES9022.
As I think about this, in my rather non-technical way, if the I2S connections as outlined are suitable for the ES9022, I could, bypass the 7220 filter and leave the 7323 DAC in place. This would leave the system clock to untouched and remove the 4 X over sampling of the 7220. I could then remove power to the analog output section and the ES9022 could sit on a little board near the 7310. Or is that too easy?
Cheers,
Joe
Last edited:
> The system clock is 11.2896mHz connected to pins 25 and 24 of the 7323 DAC and then distributed to the rest of the board.
> I had thought to bypass the DAC completely, but it supplies the system clock.
That's easy.
You use ES9022 with 45.1584MHz XO (examples already shown in this thread).
Use 2 FFs to divide it down by 4 to supply the rest of the player.
Patrick
> I had thought to bypass the DAC completely, but it supplies the system clock.
That's easy.
You use ES9022 with 45.1584MHz XO (examples already shown in this thread).
Use 2 FFs to divide it down by 4 to supply the rest of the player.
Patrick
OK I found the service manual and now my curiosity is assuaged
The 7220 is only being used to create the SPDIF output - so if you don't need that I'd remove that chip completely. Its a current hog and creates a fair amount of noise on the PSU. Then you can use the I2S feed to that chip to your ESS DAC. Bear in mind that there are some other functions which the 7220 (SAA7323 in your case) performs which won't be done to the I2S stream to your DAC - such as muting/interpolating over errors. Thus you might well experience clicks and the like instead of smooth fades. Functions like track seek, search and pause will be affected I would guess.Thank you Patrick and abraxalito.
Yes the SAA7220 draws enough current to need substantial heatsink on the LM317 regulator I have given it. On the other hand, I am not sure I want to have the clicks and pops that could come with its removal, so I will buy some ES9022s, get a datasheet and look at a complete DAC. Maybe take the SPDIF from the 7220 (pin 14).
Thanks again for you help which has now given me a clear direction.
Joe
Yes the SAA7220 draws enough current to need substantial heatsink on the LM317 regulator I have given it. On the other hand, I am not sure I want to have the clicks and pops that could come with its removal, so I will buy some ES9022s, get a datasheet and look at a complete DAC. Maybe take the SPDIF from the 7220 (pin 14).
Thanks again for you help which has now given me a clear direction.
Joe
The clicks and pops don't arise from the removal of the 7220 they come from taking the 'raw' I2S stream (as opposed to the processed one, done in both the 7220 and 7323) and feeding that into any DAC.
Certainly making use of the SPDIF output gets around many of the problems you've mentioned even though it brings its own separate set of challenges. Good luck!
The story of an ESS9022 dac
My digital front end for several years has been a cMP2 computer (please see Computer Asylum) which had a juli@ sound card fitted.
Initially I used the SPDIF output to a Buffalo 32s DAC.....
After a year or so I used the I2S output from the juli@ to a Buffalo II DAC / IVY 3 changing to Buffalo II / Onetics transformers....
Then I decided to try a QLS QA550 Card Player (just out of curiosity) initally going via I2S to the Buffalo II.....
It was at this point that I noticed the talk about the ESS9022 (got there eventually!) and I thought I would give it a go - however it was so small that even with the adaptor boards I could not really do anything (mid sixties / failing eyesight etc)....
Enter EUVL, who on a quid pro quo basis made me a balanced dac using the ESS9022 (lucky me!)...
SO I modded the QA550 until I killed it and put the 9022 to one side ....
A few months later I read about all these USB dacs/devices (I know too much time on my hands) in particular the exau2i (USB to I2S converter) being used by some cMP2 / juli@ people...
So goodbye juli@, hello exau2i, feeding in to my Buffalo II / Onetics - quite an improvement but.....
Where did I put that ESS9022 dac? Found and fitted on the outputs of the exau2i - now that is what I call synergy!!!
I cannot believe that such tiny dac chip can produce so much music, it is one of the least digital / most analogue sound I have heard in my system.....
Hmmmmm - thats nice
So thanks to jkeny for posing the question and many, many thanks to EUVL for helping me out.
I know I have cheated (the dac was not DIY it was DI-EUVL) but for people thinking about the ESS9022 (and I assume the 9023) all I can say is give it a go.
As a matter of interest the DAC is two boards each approx 1.75 inches by 1.5 inches.
My digital front end for several years has been a cMP2 computer (please see Computer Asylum) which had a juli@ sound card fitted.
Initially I used the SPDIF output to a Buffalo 32s DAC.....
After a year or so I used the I2S output from the juli@ to a Buffalo II DAC / IVY 3 changing to Buffalo II / Onetics transformers....
Then I decided to try a QLS QA550 Card Player (just out of curiosity) initally going via I2S to the Buffalo II.....
It was at this point that I noticed the talk about the ESS9022 (got there eventually!
) and I thought I would give it a go - however it was so small that even with the adaptor boards I could not really do anything (mid sixties / failing eyesight etc)....Enter EUVL, who on a quid pro quo basis made me a balanced dac using the ESS9022 (lucky me!)...
SO I modded the QA550 until I killed it and put the 9022 to one side ....
A few months later I read about all these USB dacs/devices (I know too much time on my hands) in particular the exau2i (USB to I2S converter) being used by some cMP2 / juli@ people...
So goodbye juli@, hello exau2i, feeding in to my Buffalo II / Onetics - quite an improvement but.....
Where did I put that ESS9022 dac? Found and fitted on the outputs of the exau2i - now that is what I call synergy!!!
I cannot believe that such tiny dac chip can produce so much music, it is one of the least digital / most analogue sound I have heard in my system.....
Hmmmmm - thats nice
So thanks to jkeny for posing the question and many, many thanks to EUVL for helping me out.
I know I have cheated (the dac was not DIY it was DI-EUVL) but for people thinking about the ESS9022 (and I assume the 9023) all I can say is give it a go.
As a matter of interest the DAC is two boards each approx 1.75 inches by 1.5 inches.
The QA550, or a dsPIC based SD card player, excels when you use synchronous clocking with the ES9022.
That means you have to use a 45.1584MHz XO to drive the 9022 (at close proximity), and then use 2 high speed single gate FFs to divide the frequency down to 11.2896MHz, which is then used to feed the QA550 (with its original XO removed).
In our opinion much more detailed, transparent sound.
Patrick
That means you have to use a 45.1584MHz XO to drive the 9022 (at close proximity), and then use 2 high speed single gate FFs to divide the frequency down to 11.2896MHz, which is then used to feed the QA550 (with its original XO removed).
In our opinion much more detailed, transparent sound.
Patrick
> the qa550 with es9022 beats the exau2i hands down?
That was not what I said.
What I said was, if your material is red book CD, (i.e. 16 bit 44.1kHz), and you are using a QA550 type solution, then you should drive both DAC and SD player with one XO at 45MHz, placed closed to the DAC.
The exau2i has a different application, namely high resolution, high data rate software (>>24bit/192kHz, like 32 bit 384kHz). This the QA550 cannot do, and probably the ES9022 is also too slow. But if your music is not 32 bit 384kHz but rather red book CD, then IMHO the exau2i is a too complicated solution.
Patrick
That was not what I said.
What I said was, if your material is red book CD, (i.e. 16 bit 44.1kHz), and you are using a QA550 type solution, then you should drive both DAC and SD player with one XO at 45MHz, placed closed to the DAC.
The exau2i has a different application, namely high resolution, high data rate software (>>24bit/192kHz, like 32 bit 384kHz). This the QA550 cannot do, and probably the ES9022 is also too slow. But if your music is not 32 bit 384kHz but rather red book CD, then IMHO the exau2i is a too complicated solution.
Patrick
For clarification, Alan's balanced ES9022 has a 50MHz XO.
That means he is using the ES9022 in asynchronous mode, whether with the QA550 or with the exau2i.
And what is interesting in his experience is that there is apparently no sonic advantage in using the high-end Sabre chips with 4x channels in parallel and IV converters at >> 20x the price (for the DAC).
Or, that particular implementation of the DAC was not revealing the true potential of that DAC chip ...
Patrick
That means he is using the ES9022 in asynchronous mode, whether with the QA550 or with the exau2i.
And what is interesting in his experience is that there is apparently no sonic advantage in using the high-end Sabre chips with 4x channels in parallel and IV converters at >> 20x the price (for the DAC).
Or, that particular implementation of the DAC was not revealing the true potential of that DAC chip ...
Patrick
Those that do not take advantage of the programming of the ES9018 chip and use it in Sabre32 mode are better off with a ES9022 with a 49.152 MHz clock with respect to audio quality...
For a period of time I was using a balanced setup with two ES9022 with improved PSU and LDR volume control - this sounded much better than any ES9018 based DAC´s like Buffalo II with I/V etc. (using the Sabre8 compatibility pin mode and programming)...
But when I started to tweak the programming of the ES9018, implemented JFET regulated PSU, 1ps clocks running at integer steps of the word clock like 98.304 MHz for the 48k family of word clocks - then the ES9022 solution was left behind in audio quality.
A new batch of modules ready to go to their new owners.
On the left a standard asynchronous ES9022-50MHz with I2S input.
On the right a WM8804 SPDIF Receiver module, followed by a ES9022-48MHz, with the 1:4 frequency divider which feeds 12MHz to the WM8804 from the same XO as the DAC.
We got tired of etching PCBs, so we have some made for the ES9022.
The next batch of WM8804 will also get factory made PCBs.
Patrick
On the left a standard asynchronous ES9022-50MHz with I2S input.
On the right a WM8804 SPDIF Receiver module, followed by a ES9022-48MHz, with the 1:4 frequency divider which feeds 12MHz to the WM8804 from the same XO as the DAC.
We got tired of etching PCBs, so we have some made for the ES9022.
The next batch of WM8804 will also get factory made PCBs.
Patrick
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.