Experience with this DIY DAC ?

Hi Sheldon

I've just uploaded the schematic for someone else to see:

CS4398 DAC2 192KHz.pdf

BTW interesting site you have.


Thanks Rich, I appreciate the schematic. I'll post mine as soon as I finish my power supply portion of the schematic capture.

Thanks for the site comment. My site is getting old and long in the tooth. I should add some new content, but that involves actually doing something interesting. :D


Sheldon
 
Looking at the schematic, I see a few issues. Obviously the DAC works, so I can only assume the schematic doesn't reflect the as-built configuration, or the CS8421 defaults.

Here's some note about things that I'm not sure of (they differ from my design based on the data sheets):

In the CS8416 portion, I set TXSEL0 and TXSEL1 to high so that the RXP3 input is put out on the TX pin. Since there is no input there, that pin should not be a noise source. Also the audio pin is open, and depending on the output format desired. (I ground it for left justified data when combined with U)

In the CS8421 portion, the schematic appears to be incomplete. In the schematic SAIF and SAOF pins are open. I haven't found in the data sheet what input and output modes that represents, but I defined the modes in my design:

SAIF: 1.96K to GND - left justified data
SAOF: 63.4K to GND - Left justified 24 bit data

Those are just my data buss decisions, you could easily use other modes, but those match my CS8416 and CS4398 outputs and inputs.

Sheldon
 
IF anybody is interested, here's my DAC schematic. It's not complete, but has the bulk of the features I want. I've still tuning values and considering how to implement an input transformer with the CS8416 multiplexer. The output is a pair of Lundahl transformers.
Sheldon

I looked at the schematic and found a few errors.

The CS8421 needs +2.5V power (VD), so R112 should be 220R and the clock oscillator powered from VCL, not VCD.
If you are netlisting, the wiring around C109 is missing a dot and ground; C61 needs power.
U5 (24.576MHz clock module) are notorious for polluting it's VDD. I've seen 10R or ferrite beads on the power to it, then 0.1/100uF decoupling. R10-12, R19-23 around 33-47R works well.

Otherwise it looks good :)
 
For what it's worth, the Chinese dac boards are set for standard I2S. I'm feeding the 8421 directly from an SAA7210 and CM4/19, a notable improvement over SPDIF.

I've done the same previously. Fed I2S from the SAA7210 on a Meridian transport (CDM2) into the filter on an external DAC.

Its works very well removing the change to spdif and back again but I understand its not very good over long distances. I only tried up to about 25cm.
 
prairiemystic -

Thanks for looking over my schematic, and catching my blunders. I wasn't sure if I would need data line resistors (the evaluation board schematics don't have them). I figured I'd need at least a jumper for layout anyway, so I added zero ohm resistors.

Regarding I2S:

Using i2S as an input is an interesting idea, and a good reason to do the inter-chip communication in I2S mode. I'll have to read more into it and what the transmission limits are. I'm using this DAC with an Airport Express, so the TOSLink and SPDIF are my main focus. In the last few years, I've been streaming all my music (lossless at home, low compression MP3 for the phone), and the user interface of iTunes (and the phone remote app) and the airport expresses is hard to beat.

Sheldon
 
To all,

This DAC comes in two version, with upsampling board 8421 and without upsampling board - closed jumper.

Can anyone explain differences in sound between these two version ?

Mine is w/o upsampling board. The ebay seller offer me the 8241 board for USD35 and the new version with PCM1798 chip for USD45.

Please let me know if this is could be consider a worthwhile upgrade.....???
 
After some debate I ended up using mine without upsampling board. Some would probably disagree with that.
For my needs the stripped down version(no USB, no upsampling and even no toslink) is fine.
The PCM1798 board looks interesting but I believe(correct me if I'm wrong) that it will not work well with 1:1 output trafos - if that's what you're using.
I think the PCM1798 chip is generally more natural sounding than the CS4398 and I've noticed a few manufacturers are using it in their high end DACs. Personally I would have given it a try but, for now, I don't want to have to buy new trafos and parts again.
 
After some debate I ended up using mine without upsampling board. Some would probably disagree with that.
For my needs the stripped down version(no USB, no upsampling and even no toslink) is fine.
The PCM1798 board looks interesting but I believe(correct me if I'm wrong) that it will not work well with 1:1 output trafos - if that's what you're using.
I think the PCM1798 chip is generally more natural sounding than the CS4398 and I've noticed a few manufacturers are using it in their high end DACs. Personally I would have given it a try but, for now, I don't want to have to buy new trafos and parts again.

McCrackers

Does this means you suggest that the PCM1798 should be couple with the on board opamp output stage....???? Really....??? opamp....or you have other i/v circuits works very well with 1798 ????

Others....??? Any thoughts.....

Oh BTW looking at the previous thread 1798 good with a gain circuit after 1:1 traffo.... http://www.diyaudio.com/forums/digital-line-level/137976-experience-diy-dac-143.html#post1963174
 
Last edited:
You need 18-20/1 trafos with a current out dac, totally different, and they are very touchy as far as loading and filtering. It's probably better in the long run but there is a learning curve involved that I myself am not ready to tackle. We would be dealing with very expensive trafos and complicated math. If you're ready to make the jump I'm sure we will support you any way we can.

Personally, I find the upsampling much more fluid and musical, but YMMV.
 
You need 18-20/1 trafos with a current out dac, totally different, and they are very touchy as far as loading and filtering. It's probably better in the long run but there is a learning curve involved that I myself am not ready to tackle. We would be dealing with very expensive trafos and complicated math. If you're ready to make the jump I'm sure we will support you any way we can.

Personally, I find the upsampling much more fluid and musical, but YMMV.

In the vein of helping any way we can:

I used a resistor for I/V and a tube for the 18-20 gain, but that was back in the early to mid 90's with the very good sounding PCM63's. It seems that the resistor I/V is still alive and well (based on other dac designs here and elsewhere).

My experience with Crystal delta-sigma DACs vs ladder DACs in the 90's were that the ladder DACs sounded a lot better to my ears. In fact the PCM63 is still the reference that I compare others to. By the late 90's delta-sigma was really doing a lot better and chips like the Crystal CS4390 were doing every bit as good as the PCM63 in some areas. I stopped fiddling with digital sources at that time and started building electrostatic speakers (speakers are where the real audio system gains are).

The BurrBrown (now TI) DAC that you are referencing also appears to be a delta-sigma DAC (although I haven't researched how the Crystal (now Cirrus) vs BB techniques differ. I'd guess that the actual DAC conversion is very similar, but the analog treatment of it is different.

In my limited and old experience, the real sonic "low hanging fruit" is in the power supplies (delta sigma systems used to be much more sensitive than ladder DAC systems, but that's old data), and in the analog stage. With ladder DACs the analog stage has to have a bit of gain (akin to a MM phono stage), and some filtering. This area can really make or break your signal. With the BB DAC you mentioned, you will have to do something to get about a gain of 20. With the CD4398, that has been done for you for better or worse. I haven't tried the CS4398 yet, but based on the CS4328, CS4329, and CD4390, Crystal was on a path to good sound. In other words, with the CS4398 you have a pretty simple analog portion, with the PCM part, you will have something more elaborate. The PCM part is more of a blank canvas that you can make your masterpiece on.

Sheldon
 
PCM1798 - I'll pass this. Now concentrate on discrete regulator mod which worthwhile....

McCrackers, what's so fuss about the STR ? It just a tiny regulator kit....if so complicated then there's so many regulators module out there that can do the job for a simple +5V supply.....

Good question. I have no experience with this and just followed recommendations from others. I'm now waiting for them to figure out how to get these to work. The question was asked here, on PFM and to the manufacturer himself and despite all the knowledgeable people no one seems to have the answer.
 
Good question. I have no experience with this and just followed recommendations from others. I'm now waiting for them to figure out how to get these to work. The question was asked here, on PFM and to the manufacturer himself and despite all the knowledgeable people no one seems to have the answer.

Found this abundance of diy regulator links. Some might work. Just check out...

Power Supplies and Regulators for Hi-Fi Audio

Just scroll down to Teddy Regulator (Three-terminal shunt and series regulators ) and read the description....
 
A great upgrade for this dac? DIR9001!

Yo, I just received the DIR9001 and mounted on this DAC with the UTC A-20! Great upgrade! Great bass!!! Highly recommended! :D
 

Attachments

  • DIR9001-1.jpg
    DIR9001-1.jpg
    102.1 KB · Views: 446
  • DIR9001-2.jpg
    DIR9001-2.jpg
    98.3 KB · Views: 430