Right, I have been looking about on the forums and the web now for a kit or design to build myself a new external DAC, but it seams all of the ones I have found are VERY simple, or too expensive for me to buy in kit form. So I as thinking after my exams finish in a couple of weeks, I would start tyring to design my own device. Job, and girlfriend permitting of course.
Anyway, I have aspirations of building a fairly high end device using a diferential aray of two DAC outputs per channel (two seperate stereo DACs with one side of each diriving half of one chanels output), differential buffers, and then both balanced and unbalenced line outputs. Also, I would like to have the option of runnig it at higer that CD data rates (at least 24bit, 96kHz) form the inputs.
Now, as I am fairly new to all this, and eveyone here seems to have a lot more experience than me, I was wondering if you would be able to point me in the right direction for which DACs to use in the design. At the moment, I have been looking at the TI PCM1704 (24bit, 96kHz), or the crystal cs43122 (24bit, 192kHz) models. If anyone could tell me subjectivly which of these would be best for my project, or if I have passed by any better units then that would be very helpful.
Also, one area I have not been able to find a lot of information on yet, is the input decoding chips that would compliment the above DACs. I would want the unit to ba able to run from both optical and Coaxial (BNC) inputs, idealy being able to automaticaly switch between the two depending on whihch was in use. So, if someone could suggest a chip which would act as the input buffer/decoder for this, that would be helpful. Finaly, if using the DAC chips mentioned above (or any other ones that are suggested) would I need to include any sort of digital filter or upsampling unit to the design.
Sorry about putting so many questions into the one place, but I would like to get a good idea of what I will be looking at in terms of cost and complexity before I start wadding through all of the data sheets on these devices.
Anyway, I have aspirations of building a fairly high end device using a diferential aray of two DAC outputs per channel (two seperate stereo DACs with one side of each diriving half of one chanels output), differential buffers, and then both balanced and unbalenced line outputs. Also, I would like to have the option of runnig it at higer that CD data rates (at least 24bit, 96kHz) form the inputs.
Now, as I am fairly new to all this, and eveyone here seems to have a lot more experience than me, I was wondering if you would be able to point me in the right direction for which DACs to use in the design. At the moment, I have been looking at the TI PCM1704 (24bit, 96kHz), or the crystal cs43122 (24bit, 192kHz) models. If anyone could tell me subjectivly which of these would be best for my project, or if I have passed by any better units then that would be very helpful.
Also, one area I have not been able to find a lot of information on yet, is the input decoding chips that would compliment the above DACs. I would want the unit to ba able to run from both optical and Coaxial (BNC) inputs, idealy being able to automaticaly switch between the two depending on whihch was in use. So, if someone could suggest a chip which would act as the input buffer/decoder for this, that would be helpful. Finaly, if using the DAC chips mentioned above (or any other ones that are suggested) would I need to include any sort of digital filter or upsampling unit to the design.
Sorry about putting so many questions into the one place, but I would like to get a good idea of what I will be looking at in terms of cost and complexity before I start wadding through all of the data sheets on these devices.
Manufacturers have evaluation boards, which usually show a complete reference design, from the coax and optical inputs, through the receiver chip, and all the way to the analog output and filter for each of these. The datasheets for the EVBs include schematics and circuit layouts.
How about starting with one of those, and making any changes for your project? Then you can ask more specific questions if you have to.
Example:
http://www.cirrus.com/en/pubs/rdDatasheet/cs43122eb-1.pdf
There are also typically "white" papers discussing the merits of a given design approach, upsampling methods, etc.
How about starting with one of those, and making any changes for your project? Then you can ask more specific questions if you have to.
Example:
http://www.cirrus.com/en/pubs/rdDatasheet/cs43122eb-1.pdf
There are also typically "white" papers discussing the merits of a given design approach, upsampling methods, etc.
i had already seen the one you pointed out, and I was aware of the reference desigs, whihc would proabbly help guide the project once I have started, but I was looking for any oppinions out there from people as to what works the best, as I have found very little info regarding which items would be better to use than the others. But, thanks anyway.
peranders, thanks for the sujestions, but the TDA1541 is only 16bit isn't it? and I realy would like to have the ability to go to higher resolution and bitdepths with this project (it seems silly not to when I intend on putting so much effort into it, I don't wna't it to become obsolete too quickly)
peranders, thanks for the sujestions, but the TDA1541 is only 16bit isn't it? and I realy would like to have the ability to go to higher resolution and bitdepths with this project (it seems silly not to when I intend on putting so much effort into it, I don't wna't it to become obsolete too quickly)
opinions
I have heard the Analog Devices parts are more difficult to interface than TI and Crystal, but this is not from experience.
Sigma delta DACs are not beloved by the diy community, or for that matter by parts of the audiophile crowd, and there is no great art in using them. I suspect that all else being equal (output circuit, layout, power supply, etc., they would sound pretty much the same). Pick one that appeals to you.
R2R based DACs are not really obsolete, even if technology has passed them by. MSB Technology uses them (NOT the stock 16-bit type, but the R2R design) in their top shelf products, the Platinum line. Your design would not be in bad company if you decide to use one.
The 120 db dynamic range of the new sigma delta dacs is 6 orders of magnitude, and hard to realize except using 4-layer professional pcb layout, and even then usually not until the second or third pass. It means signal levels of a few uV, which is why some new DACs are boosting output levels, and departing from the low voltage supply model and 2V rms output. In either case, the dynamic range and high S/N ratio is typically lost as soon as the signal leaves the DAC enclosure, even on the assumption that it survived the output stage (IMO).
Btw, do I understand correctly that you plan to use outputs from two separate DACs for one channel, or did you mean a separate DAC for each channel?
I have heard the Analog Devices parts are more difficult to interface than TI and Crystal, but this is not from experience.
Sigma delta DACs are not beloved by the diy community, or for that matter by parts of the audiophile crowd, and there is no great art in using them. I suspect that all else being equal (output circuit, layout, power supply, etc., they would sound pretty much the same). Pick one that appeals to you.
R2R based DACs are not really obsolete, even if technology has passed them by. MSB Technology uses them (NOT the stock 16-bit type, but the R2R design) in their top shelf products, the Platinum line. Your design would not be in bad company if you decide to use one.
The 120 db dynamic range of the new sigma delta dacs is 6 orders of magnitude, and hard to realize except using 4-layer professional pcb layout, and even then usually not until the second or third pass. It means signal levels of a few uV, which is why some new DACs are boosting output levels, and departing from the low voltage supply model and 2V rms output. In either case, the dynamic range and high S/N ratio is typically lost as soon as the signal leaves the DAC enclosure, even on the assumption that it survived the output stage (IMO).
Btw, do I understand correctly that you plan to use outputs from two separate DACs for one channel, or did you mean a separate DAC for each channel?
Hi PMiczek, in reply to that last question of yours, I was thinking of using one out put from two seperate DAC's for each channel, as that way, I can input the entire data stream into both DAC's and tap off the two right and the two left channels which will come out of them. then, using one inverting and one noninverting buffer, run them into a set of balanced output buffers.
As far as loosing some of the dynamic range, and increasing the S/N ratio goes, I was expecting that owuld occur in part, as I would at best be running a two layer board and I have little practical experience of getting the most from my PCB layouts> But I did want the functionality of some of the newer DACs there, in case I upgrade to SACD, or some other similar format in the future.
Also, do you know anywhere that sells the MSB technologies parts. preferably in europe, as I had not herd of them before nor have I seen them when I have been looking about on RS and Farnells websites (although it would have been ver easy to over look these, as they are not too eas to navigate).
As far as loosing some of the dynamic range, and increasing the S/N ratio goes, I was expecting that owuld occur in part, as I would at best be running a two layer board and I have little practical experience of getting the most from my PCB layouts> But I did want the functionality of some of the newer DACs there, in case I upgrade to SACD, or some other similar format in the future.
Also, do you know anywhere that sells the MSB technologies parts. preferably in europe, as I had not herd of them before nor have I seen them when I have been looking about on RS and Farnells websites (although it would have been ver easy to over look these, as they are not too eas to navigate).
Hi,
There may be an easier way to use the R+/R- and L+/L- outputs on something like the CS43122, but hard to say w/o knowing what you want to accomplish.
MSB makes complete products, not parts. Sorry I did not make it clear, I only used their line as example of how some high-end equipment uses R2R type DACs. See half-way down this page:
http://www.msbtech.com/platinum_line/products/DacII/
All from me for now, time to go burn some steaks!
PM
There may be an easier way to use the R+/R- and L+/L- outputs on something like the CS43122, but hard to say w/o knowing what you want to accomplish.
MSB makes complete products, not parts. Sorry I did not make it clear, I only used their line as example of how some high-end equipment uses R2R type DACs. See half-way down this page:
http://www.msbtech.com/platinum_line/products/DacII/
All from me for now, time to go burn some steaks!
PM
The PCM1704 and CS43122 are both state-of-the-art parts. I have used both and they sound different but good. You can't take my word for it though. Experimentation is the only way to find the DAC you like the best.
You might want to start with the PCM1704 if you are looking for maximum street cred. Self-styled audiophiles like to make vague negative comments regarding ΔΣ converters, but nobody will ever doubt your choice of PCM1704. The more expensive, the better evidently. As a bonus, you need four of these to make a balanced DAC so you will spend serious money on parts and four-layer circuit boards.
As for the interface chip, I think the CS8412/8414 and CS8420 are the canonical parts. The DIR1703 seems to have some traction but in my opinion it is difficult to use. Avoid the CS8420 if you can't or don't want to program microcontrollers. Analog Devices application support once wrote to me that the AD1896 ASRC was obsolete, but I see they now have it in stock. It may have fewer bugs than the Crystal ASRC.
Ah, after some studying it seems that the AD1896 has been bumped from Rev 0 to Rev. A.
You might want to start with the PCM1704 if you are looking for maximum street cred. Self-styled audiophiles like to make vague negative comments regarding ΔΣ converters, but nobody will ever doubt your choice of PCM1704. The more expensive, the better evidently. As a bonus, you need four of these to make a balanced DAC so you will spend serious money on parts and four-layer circuit boards.
As for the interface chip, I think the CS8412/8414 and CS8420 are the canonical parts. The DIR1703 seems to have some traction but in my opinion it is difficult to use. Avoid the CS8420 if you can't or don't want to program microcontrollers. Analog Devices application support once wrote to me that the AD1896 ASRC was obsolete, but I see they now have it in stock. It may have fewer bugs than the Crystal ASRC.
Ah, after some studying it seems that the AD1896 has been bumped from Rev 0 to Rev. A.
You might consider using the PCM1738 with a CS8414 serving as the input receiver. It has balanced outputs and at around £10-£15 it will cost a fraction of the price of 4 PCM1704's (£20+ each standard. nearer £35+ for for the top grade) and the DF1704 digital filter at another £20 or so.
So long as the SACD player you use has a DSD to PCM converter onboard you will be able to use this dac but as the preferred interface for raw DSD appears to be encrypted IEEE1394 designing a DAC to cater for that option will be an expensive and difficult proposition.
For a look at a commercial design, download the Passlabs D1 service manual at www.passlabs.com .It may prove a bit complex and some of the parts have been discontinued but it should at the very least give you some ideas.
ray.
So long as the SACD player you use has a DSD to PCM converter onboard you will be able to use this dac but as the preferred interface for raw DSD appears to be encrypted IEEE1394 designing a DAC to cater for that option will be an expensive and difficult proposition.
For a look at a commercial design, download the Passlabs D1 service manual at www.passlabs.com .It may prove a bit complex and some of the parts have been discontinued but it should at the very least give you some ideas.
ray.
If you accept that a home-made board will not have the best noise and dynamic range performance possible, I wonder if using one of the Crystal DACs with volume control might not make for a fun project? - especially if you need to program a PIC or something like that to keep the CS8420 in line. You would essentially replace the one function in a preamp that the DAC has any use for, and go from transport to DAC to amp with the fewest interconnects.
(With apologies to MSB, that is essentially what their new Platinum DACII/Preamp does, only for a lot more money)
Maybe find a fellow student to do the digital part....
PM
(back to the grill)
(With apologies to MSB, that is essentially what their new Platinum DACII/Preamp does, only for a lot more money)
Maybe find a fellow student to do the digital part....
PM
(back to the grill)
The PCM1738 does look quite interesting, I have a quick scan of the data sheet and I see that ir would allow me do do away with the DF1704 controler chip, and get diferential output from the one chip instead of using four 1704's which would be much kinder to my wallet.
However, I am now wishing I had held off asking about this unitl after my exams are over, as I should be revising now, not trying do research the bes setup for a new DAC for me.
One thing I was wondering now though, is if there would be any interest out there in people chipping in on this project, and sharing in the output, as I am quite happy to look through all of the aplication notes (as soon as I have finished my exams on the 6th, well perhapse on the sunday after the hangover
) and come up with what should be a decent circuit design. But I would be greatfull for any help I could get with the PCB layout, as I ahven't realy attempted anything of this magnitude before, so there may be a ratehr large benifit in someone with a little more experience chipping in to cover that part of the design.
But, if people do want to do that, am I looking at a popular feature set (Differential operation at up to 24bit 96kHz opperation), or would people rather see some other features in there too?
Anyway, I should go and do some more work now before I get to bed (I am trying to force mechanics into my head today, which I have no interest in whatsoever, but is supposed to give me a "solid fooundation in engineering principles"), so this should be it from me today (but I'm guessing won't be).
Andrew.
However, I am now wishing I had held off asking about this unitl after my exams are over, as I should be revising now, not trying do research the bes setup for a new DAC for me.
One thing I was wondering now though, is if there would be any interest out there in people chipping in on this project, and sharing in the output, as I am quite happy to look through all of the aplication notes (as soon as I have finished my exams on the 6th, well perhapse on the sunday after the hangover
) and come up with what should be a decent circuit design. But I would be greatfull for any help I could get with the PCB layout, as I ahven't realy attempted anything of this magnitude before, so there may be a ratehr large benifit in someone with a little more experience chipping in to cover that part of the design.But, if people do want to do that, am I looking at a popular feature set (Differential operation at up to 24bit 96kHz opperation), or would people rather see some other features in there too?
Anyway, I should go and do some more work now before I get to bed (I am trying to force mechanics into my head today, which I have no interest in whatsoever, but is supposed to give me a "solid fooundation in engineering principles"), so this should be it from me today (but I'm guessing won't be).
Andrew.
Well, I have been looking about and have downloaded about 10 or so different data sheets on various DAC's and a couple of interface chips from cirrus, but I only managed to get one printed out before I ran out of paper (I HATE reading data sheets on my computer screen 
). But I should get round to looking through these in the next couple of weeks, so I start to make some choices as to which chip I will end up using. But, I think at the moment (for reasons of cost) I will be going for one of the stereo models with differential outputs rather than using two or more descreet devices as I had originaly pplanned, as the though of spending a couple of hundred pound just on the chips was a little scary.
). But I should get round to looking through these in the next couple of weeks, so I start to make some choices as to which chip I will end up using. But, I think at the moment (for reasons of cost) I will be going for one of the stereo models with differential outputs rather than using two or more descreet devices as I had originaly pplanned, as the though of spending a couple of hundred pound just on the chips was a little scary.
peranders said:
For me this was a joke but for other people this is dead serious
I haven't heard this ancient DAC but I doubt that development has stood still since the stone age. Is really 16/44 from an old DAC better that state of the art 96/24 or 192/24 DAC?.......
Per-Anders, if you haven't heard that DAC howcome you're so negative about it ?? Seems pretty blunt to me to comment on something you don't know. I recall you commented a few times on the fact that newer DAC chips are better than TDAxxxx and now you're saying you don't even heard the old part ?!?! It isn't specs only that are important. Nowadays price is the most important spec there is for the manufacturers of audio chips since there are more competitors than before ( and quality has even less weight than before ). I sure know a lot of 96/24 DAC chips that are clearly inferior to some of the old R2R types in terms of musicality etc. But on paper they're very good. So please go ahead and design whatever you can with 10 layer boards and 32 bit chips that are better on paper but don't judge on something you can only assume.
Since I am a self-styled audiophile ( according to some specialists that like to speak in general ) I like to make vague negative comments regarding Delta Sigma converters because I compared them with multibit DAC's. So I do know what I heard and that's more than you do apparantly.
TDA1543 sells for 2 to 3 $, TDA1541A for 6 $. They are around for quite some time now, longer than most state of the art Delta Sigma's anyhow. I wouldn't recommend them for a new design though
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.