Comments on my cs8416 pcm1730 dac project

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi

I would like to get some comments on my schematics. I have tried to follow the directions in the datasheets, but it would be great to if someone spotted potential mistakes before I start making the pcb. The cs8416 is configured in hardware mode and using I2S encoding for transfer to the pcm1730. I have tried it before, but I used too thin traces, so I guess the board had lots of opens and did not work. I think I'll use smd for the 47k resistors this time, because there are so many!

Should I consider adding a src4192 between the two to get better jitter performance?

How does 1730 sound compared to pcm1792/3?).

Is my power on reset too simple?

I have used the recommended decoupling - is that enough?
What are the most critical power nodes? I guess it would be VA on cs8416 and the 5V on the output of pcm1730.

If I decided to add a usb interface too, how could I mux in the signal?

I think I'll use Lars Nielsen's design for current-voltage conversion:
http://www.diyhifi.dk/DA_A1.pdf
When this works, the next step in my project is to make the line stage using tubes. :D


Thanks,

Christian
 

Attachments

  • dac.pdf
    13.1 KB · Views: 441
Hey, I'm also working on a design using CS8416, which is a work in progress. I have a few questions. Why do you use the 47K resistors on pins 10-13? I didn't see that in the data sheet. Also, why do you pull pin 15 to gnd? Isn't it better to have the emphasis match on? Finally, I've been told to put 100 ohm resistors on pins 26-28, and don't you need a 47K pulldown resistor on SDOUT to put the receiver in HW mode (p. 44 data sheet)? Of course, I'm learning as well, so I beg you not to take my word as law. I just wanted to point these things out, so maybe someone can tell us who is correct. Good luck!
 

Attachments

  • ezdac_schem.pdf
    75.2 KB · Views: 364
Thanks for your comments.
I was not sure about the 100 resistors, I guess it depends on what the load is. I know I need to connect sdout->47k->ground, but I guess I was too tired last night. :cannotbe:
I don't understand your point with pin 15, because you connect it just like I do. You might be talking about pin 16, in that case I'll have to dig into the datasheet again.
Maybe I have misread the information in section 13.2 regarding pull up/down resistors, to me it just sounded like it needed to be done for all selections. But I guess you are right because pin 10-13 are not "start-up" option pins.

I have attached a version with a corrected sdout.
 

Attachments

  • dacr2.pdf
    13.2 KB · Views: 282
I actually don't really know what the de-emphasis filter does and if it should be on or off. It sounds to me like it doesn't really matter because it is has something to do with selecting datatype and I'll only use it for audio.

I think you are right about the rxsel and txsel, so I have removed the resistors. I read that TX(pin 20) has it's own internal resistor, so should it just be connected directly to ground without a 47k resistor?

Any ideas on how to mux in another i2s signal? I guess it's a bad idea to use a hardware switch.

I was thinking about making the IV circuit on another pcb, is that possible, or are the current signals from the dac too sensitive to travel a few centimeters to another board? That way I can test different IV's later.
 
I think you can make a separate PCB for the I/V. Look, for example, at the RAKK DAC, which uses that approach. I'm thinking about going with a balanced I/V on the same board using either AD8065 or AD8610 (which are pin compatible). It's hard to decide which op amp, because there are so many, but I think if the design is simple enough, I will be able to try several different ones, as long as they are pin compatible. I don't want to have to make different boards.
 
Promiteus makes a nice board w/power supply for the I/V output stage based on the Pass Labs DI. I bought a set without a dac design fleshed out with the intention of using a dac chip with differential outputs to drive balanced interconnects. I am new to this digital stuff and have yet to put together a pcb but was thinking of the same chipset with an iso150 digital coupler between chips for isolation. It seems like a easy addition but have not heard of any disadvantages. I will also post a couple of links to messages about input transformers for the digital input. Again, this seems like an easy addition and people seem to think that it helps clean up the signal. I am not in a position to start any new projects right now but have been reading everything on dacs that i can get my hands on and would be interested in following this conversation.
 
I think I'll try the Pass Labs DI I/V stage as well. I have read some very positive reviews, but I still haven't found the schematics. My focus is finishing the digital part, but when that is done, I'm free to experiment with discrete I/V's based on bjt, jfet, mos and tubes. I have a schematics for balanced I/V using opamps, which I will also put in the comparison.

What are you gaining by inserting a digital coupler between the chips (I assume you are talking about receiver->dac)?

IMHO it will make more sense to do re-clocking, if you want to improve jitter performance. I have been playing with the idea of a src4192 between the devices, but it adds a lot of complexity and I would rather get a working board now than the perfect one.

Are you planning to etch your own pcb?
 
I've attached an updated version of my design, which uses AD1896 for reclocking. That seems to be the most popular ASRC, as far as I can tell. From the data sheet, it really isn't that difficult to design a schematic. I'm going to try to prototype this thing within the next couple of months, if time permits. I'll keep you posted on my progress.
 

Attachments

  • ezdac_schem.pdf
    57.1 KB · Views: 320
Looks quite simple.... maybe I should reconsider.
I thought you needed a clock source (oscillator) for reclocking - is that what you have connected to MCK on ad and pcm? What device do you use to make the clock and at what frequency?
Are you oversampling in ad1896? If you hardwire that stuff don't you have a problem when playing dvds (44.1 vs 46)?

EDIT: Oh, I didn't look at the last page. Is U7 a stock component? Is it really necessary with the tps3809? But, I think I'll see if my supplier something similar, because it might be more reliable than something I do with caps and resistors...
 
cviller said:
Looks quite simple.... maybe I should reconsider.
I thought you needed a clock source (oscillator) for reclocking - is that what you have connected to MCK on ad and pcm? What device do you use to make the clock and at what frequency?
Are you oversampling in ad1896? If you hardwire that stuff don't you have a problem when playing dvds (44.1 vs 46)?

On p. 4, the C33xx (Crystek XO) outputs MCK at 24.576 MHz, which is fed into AD1896 and PCM1794. I have AD1896 set up to output 96 kHz (24.576=256*44.1). I should have said that I am not planning on using this DAC with anything other than 44.1 (I don't even have any DVD-Audio or SACD's, so that's not an issue for me.) I am intentionally avoiding using microcontrollers, so that's why everything needs to be done in hardware. It is also why I will not have the PCB made until I get the prototype working. You must read the data sheets to figure out what clock frequencies you will need. A simpler solution, would be just to use the onboard clock, but who wants simpler? I want to emphasize that my schematic should definitely be taken with a grain of salt at this stage. I don't have a working prototype. I have posted it in several forums, and used feedback to get it to the state that it is, but I'm sure some things will change when I start to build it. You're welcome to copy any part of it you want.
 
Do you mean 96k*256=24.576? Wouldn't you get better sound (less roundoff) if you selected 88.2k as output since your source is 44.1k? It probably depends on how good the algorithm in the src is, but I like when numbers match! :devilr:
(BTW sorry for writing 46 before - I know it's 48k on dvds (video), but it was before I got my coffee)

Maybe I should have two clock-sources to use for the different formats. I have some src4192 lying around, so it's almost a crime not to try them. :D
The problem is that I can't find any suppliers in denmark for these exact frequencies, so I might have to order from digikey, but I have no idea how much they charge for shipping and if they even sell in small quantities.
 
cviller said:
Do you mean 96k*256=24.576? Wouldn't you get better sound (less roundoff) if you selected 88.2k as output since your source is 44.1k? It probably depends on how good the algorithm in the src is, but I like when numbers match! :devilr:
(BTW sorry for writing 46 before - I know it's 48k on dvds (video), but it was before I got my coffee)

Maybe I should have two clock-sources to use for the different formats. I have some src4192 lying around, so it's almost a crime not to try them. :D
The problem is that I can't find any suppliers in denmark for these exact frequencies, so I might have to order from digikey, but I have no idea how much they charge for shipping and if they even sell in small quantities.

Yes, you're right about the math. As for 88.2 vs 96, it doesn't matter because the filtering doesn't work in the way you're thinking (read the data sheet carefully). I chose 96 because 24.576 is a fairly standard frequency, although I suppose it doesn't matter. The data sheet says that the clock must be greater than 138fs. As for the two clock sources, bad idea to have two crystals on one board. If you want to have multiple frequencies, I think the best way to do it is to have a VCXO (voltage controlled XO), but that will involve much more work on your part.
 
Great! I just read the AD1896 datasheet and I am amazed how much better the guys at AD write datasheets compared to TI. Before this I have only read src4192 and it merely scratches the surface on what is going on.
Do you know if they are using the same principle? The two parts are almost pin compatible and as I said I have some src4192's lying around.
 
cviller said:
Great! I just read the AD1896 datasheet and I am amazed how much better the guys at AD write datasheets compared to TI. Before this I have only read src4192 and it merely scratches the surface on what is going on.
Do you know if they are using the same principle? The two parts are almost pin compatible and as I said I have some src4192's lying around.

It looks that way, at least, the pins are labeled the same. The codes could be different, though. Gotta check...

Update...Yes, they are, under certain conditions. See pg. 28 of the SRC4192 data sheet. It's cheaper, too. Maybe I'll be the one switching.
 
cviller, i can't remember where I have seen this coupler used but it is comperable to using an interstage transformer on a tube amp- reduces noise, common mode among others I believe. I can't remember where I have seen it used- no i can, the evm-1704 eval board from TI. Also lessloss.com

Do a search for "DIY D1 I/V Stage from Pass Dac PCBs for sale" under promitheus' name.

I tried to attach the schem but it is too big. The output stage itself is on many posts.
 
poolorpond said:
cviller, i can't remember where I have seen this coupler used but it is comperable to using an interstage transformer on a tube amp- reduces noise, common mode among others I believe. I can't remember where I have seen it used- no i can, the evm-1704 eval board from TI. Also lessloss.com

I can't seem to find any "couplers" on the evm-1704 schematics. The only thing on the signal part from spdif reciever to filter and dac looks like jumpers (CND2 and CN3). Are you referring to the use of transformers on the input side of the reciever? If that is the case, I think it is a good idea, but I'm going for the cheaper solution with caps.

Do a search for "DIY D1 I/V Stage from Pass Dac PCBs for sale" under promitheus' name.

I tried to attach the schem but it is too big. The output stage itself is on many posts.

Thank you for the effort, but I on exchange in Australia :cheers: and won't see my soldering iron untill I get back home in August, so that's why I haven't really searched for it - it just pops up in random threads and sounds like a good project... ;)
 
I yahooed evm-1704 and the users guide comes up. Figure 1.1 has the block diagram coupling the dig/analog sides of the board. Figure 2-5, 2-6 show the diagram for the iso150. These are available as samples from TI.

Yes, I am not referring to the spdif input transformer, although I would spend the extra time adding that as well. These usually get good reviews when added- as long as the transport can drive them. It is best to put one in the source as well. I got some samples of pulse transformers from somewhere too. This could all be done on the cheap.
 
That's funny. When I searched for evm1704 or evm-1704 on ti.com, I got nothing, so I found the evaluation board for pcm1704 which is not the same board as evm-1704. I still don't see what you can gain from degrading the signal - most of the dac's I have seen here on the forum are without couplers and appears to be working great.
The datasheet of iso150 states that 2ns skew between channels should be expected and more between iso150 devices and you will need 2. Does anyone know if this skew will affect the sound? What about jitter, does it make it better or worse? (I would guess worse)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.