USB DAC Design questions

[ppl]

Quote:

Originally Posted by DaKi][er While you can get away with it in headphone amps, single supply virtual grounds should be avoided, they add far more problems than they fix Really, there is 2 ways you could design- Either design it to be portable, where you would want it to run off USB power or as this was set out to be, a standalone dac to run off mains with its own transformer, psu and all that with no halfway could be portable could be fixed standalone, it should be either one of the other as designing for both just makes compromises. Have pads for the AC from the transformer/s and put everything else onboard. I’m a very strong believer that the psu should be built in as part of the design and leave very little choice I don’t have any personal interest in building one of these as I already did one for myself and I am not much of a fan for designing projects like these in groups as there are always far too many people wanting to put in their own input and in the end always throws it off course. I like to design by myself, for myself and everyone else is an afterthought. What you guy's need to do is work out exactly where you want to end up and start listing the requirements, with everyone having different ideas that will be fun

I disagree that a quality DAC can not be made using a virtual ground providing the Virtual ground is done correctly and using a fair amount of output current. I like the LT-1210 set up as a low bandwidth unity gain VF for a virtual ground. Using this configuration I measured the noise floor between the supply rail and the virtual ground and obtained -135 Db. Using this DAC chip http://www.cirrus.com/en/pubs/whiteP...s43122wp-1.pdf
And the DIR-1703 you’re using, so I would not abruptly discount the virtual ground system and portable use.

I have just found this thread and find this quite interesting. Regarding a DAC for the PPA AOS once made one http://www.aoselectronics.com/pace.html so you might give it a look over, this DAC is no longer available however.

Someone posted about using a pre regulator ahead of the TPS-79101 regs that are only good to 6 volts max even for the adjust type have this been investigated more

 

[00940]

The power supply section is not completly fixed yet. Only the final regulation is onboard (using reg101 for 5v supplies and tps79333 for 3.3v supplies). Since we are using only one supply for digital/analog, I must say a virtual ground makes some sense.

ppl : there will be a serious imbalance in between the positive and negative supply. Would you advice to put a load on the negative side to balance the draws ?

 

[ppl]

with the LT1210 no with the BUF-634 stacked in a pair of two I get balanced supply with ref o the virtual ground of no more than 300 mV and this is quite ok for a Split rail supply regardless of how it is derived. With the LT-1210 the balance is within 20 Mv. Both with the Digital supply pulling 92 am from the pos rail to the Virtual ground. Also I like the over sampling idea I also am working on that however a different chip that allows bypassing so as to conserve battery life when over sampling is not required. It appears to my measurements that beyond ICQ conditions that the current drain of the digital circuits are in direct proportion to the data rate. http://www.cirrus.com/en/pubs/proBulletin/CS8421_PB.pdf
http://www.cirrus.com/en/pubs/rdData...RD5381_RD1.pdf
BTW I am still waiting to hear how long you could run that TPS regulator on 9 volts with out falure.
.

 

[00940]

We will have preregulation at 5.5V for the digital section, with a simple lm317.
The dac will be wall powered, inefficiency is no big deal.

 

[ppl]

Is this Dac to be euro-card size? If so I assume an external AC supply is the thinking? If this is true I would give serious consideration to some isolation of the analog stage or OP-AMP’s used on the DAC’s output. If an I/V type is used then power supply integrity is of more importance than it is with a voltage output DAC. At the Minimum an R/C network should be inserted in the supply rails powering the op amps. A simple 10 ohms series resistor with a 0.1uF cap is the absolute min however a good 10uF of so of quality Electrolytic is technically more suitable hear as a large value bypassed with a low value film cap. Is a better solution. A PPA style JFET TLE and caps are what I would do if using a virtuial ground supply however if AC only operation is considered then my favor is twards a normal split rail supply a Jung type should be great hear, as the Virtual ground is great in a portable unit if a true split rail supply is available then it should be used, remember it only is (1) more wire and the cost of the power supply parts go way down as the LT-1210 is not a low cost part.

 

[00940]

The DAC will fit on an eurocard (it could actually be a bit smaller, we'll see). The aim was to provide for regulation onboard. So only DC would have to be provided. The interest of the rail splitter would be to allow for a wider range of PS to be used.

Yes, we use a dedicated I/V (OPA1632), since the chosen DAC is the PCM1798. However, even a voltage out DAC needs a good, clean analog supply to feed the opamps inside the DAC (or its analog circuitry).

The LT1210 is expensive but wouldn't the LT1206 be sufficient ? It's already 4$ cheaper, for 250ma (minimum) output current. And do we actually need so much current ? If we put a constant negative load on the splitter, of roughly the value of the idling digital section, we probably need way less, more like 50ma, to balance the splitter under all conditions. A pair of paralled njm4556 would give 4X70ma peak, for less than 1$.

We will certainly use RC networks. And all power pins on board are bypassed by a small ceramic cap (0805), another bigger ceramic or tantalum cap (1206) for the digital chips and a small electrolytic (around 47uf). And each power pin in the digital section is fed through a ferrite.

In the end, the ideal solution would be to use three supplies. A digital (+) and two analog (+/-). Since this solution involved at least three transformers (or the expensive three outputs elpac WM113), it was ruled out for cost considerations (we're aiming for 200$ without case). Thus the interaction in between the analog and digital supply will happen no matter what, rail splitter or not.

 

[ble0t]

Well, it's been a while, but 00940 and myself have been hard at work and we've finally come up with a preliminary design. First, here are the links to the schematic as well as shots of the board layout:

WARNING: The file sizes aren't that big (.gif format), but I kept the resolution very high to keep the finer details easy to see.

Schematic

Top Layer

Bottom Layer

Both Layers w/Top Silk

We have attempted to design this to be as flexible as possible for anyone who wants to build it.

There is the option of either USB (PCM2707) or SPDIF (CS8416) input which can either be selected via a soft-button switch (i.e. momentary switch instead of a toggle) or set specificaly via a jumper. There is even the option to input I2S directly and use the other input as an additional digital in (i.e. use the board inside of a CD player).

The intended signal path is the same as what had been discused earlier:
PCM2707 / CS8416 --> AD1896 --> PCM179* --> OPA1632/OPA134 based I/V stage

The AD1896 and PCM179* will be clocked via a low jitter Crystek oscillator.

The I/V stage contains pads in various places to allow for the use of whatever I/V and/or filter stage is desired.

The power supply was designed for a 15V DC input (most likely from a wallwart) that is then regulated to 11V using an LM317 and then split to +/-5.5V using an LT1206 for virtual ground. Each digital chip has its own regulator(s) that are fed from the +5.5V supply.

Finally, we do have some space in the lower right portion of the board and we would like input as to what we should use the space for. Our thoughts at this point in time are either a balanced output section or perhaps a small preamp/line driving section. Any and all input/suggestions/comments are welcome.

EDIT: I almost forgot to mention the PIC12F509...I am currently working with someone on getting code for this. I have used a DIP version of the PIC to make things much easier as far as programming is concerned. My thought at this point in time is that I could offer to program the PIC for anyone who would like to build this DAC (as long as the number of people was reasonable).

 

[SnoopyRocks]

It's nice to know that progress is being made.

At a cursory glance, it looks ok.

I am not a fan of the power supply arrangement and the virtual ground in particular. It just seems like more trouble than it's worth for a project like this. This linear regulated +12@0.3A,-12@0.3A,+5@0.86A DC supply uses a 5 pin DIN connector and is only $9.69 at Jameco. Yes, less than $10. It seems ideal for this application and would decrease the board size, parts count and cost, while making it easier to build and troubleshoot.

 

[MegaMeee]

Quote:

Originally Posted by SnoopyRocks It's nice to know that progress is being made. At a cursory glance, it looks ok. I am not a fan of the power supply arrangement and the virtual ground in particular. It just seems like more trouble than it's worth for a project like this. This linear regulated +12@0.3A,-12@0.3A,+5@0.86A DC supply uses a 5 pin DIN connector and is only $9.69 at Jameco. Yes, less than $10. It seems ideal for this application and would decrease the board size, parts count and cost, while making it easier to build and troubleshoot.

kind of a problem for us folks in 240v countries

I kinda like the current psu arrangement. The Vin is probably just a tiny bit too high for 12v sla batteries though

. Do you guys have any recommendations for a psu.. What sort of current rating do you think is necessary?

edit: forgot to mention how much I appreciate the effort you guys are putting in

 

[ppl]

Quote:

Originally Posted by SnoopyRocks It's nice to know that progress is being made. At a cursory glance, it looks ok. I am not a fan of the power supply arrangement and the virtual ground in particular. It just seems like more trouble than it's worth for a project like this. This linear regulated +12@0.3A,-12@0.3A,+5@0.86A DC supply uses a 5 pin DIN connector and is only $9.69 at Jameco. Yes, less than $10. It seems ideal for this application and would decrease the board size, parts count and cost, while making it easier to build and troubleshoot.

I tend to agree that if tis DAC is to be AC powered then a true split supply is in order however if battery operation is at all desired then the LT1206 is a good solution however i would put equial capacitence from both pos and Neg rails because as it is will all the capacitence on the Pos rail then instability in the LT1206 may result. have you figured out what operating current your going to Bias the LT1206 at? also why not use the higher current LT1210?

 

[00940]

Ok, here's the reasoning behind the rail splitter.

- there are very few dual (or even triple) outputs wallwarts. And for 220-230VAC inputs, it's even worse. If we design the dac around one of those wallwarts and it is discontinued, it will force people to build a complete PS for it.

- On the other hand, a 15VDC wallwart is easy to find for about anyone.

- Noone forces you to actually use the rail splitter. If you want to connect a multiple taps PS, feel free to do so. We will change the layout to put jumpers on, to allow such modularity.

- I don't know well batteries. The problems is that we have a very small window of operation. The input of the railsplitter must be constantly kept in between 12VDC and 10.4VDC. More than 12VDC and we exceed the rating of the TPS79333 (max 6V input), less than 10.4VDC and the REG101-5 for the analog section of the PCM1798 cannot regulate it anymore.

- The LT1206 is rated at "at least" 250ma, I'm not sure it is really needed to go further with the LT1210. I've to check if they're pin compatible though.

- I will calculate the total current draw of that thing this afternoon.

Update : estimated current draw : around 130mA on the positive line, around 20mA on the negative line. Details :

CS8416 : 5.7 + 5.9 + 2.8 : 14.4ma @3.3V
PCM2707 : 23ma @3.3V
AD1896 : 30ma @3.3V
PCM1798 : 13ma @3.3V, 19ma@5V
OPA1632 : 2X 15ma @11V
OPA134 : 2X 5ma @11V

on the -5.5V : 20ma

on the 5.5V : 14.4+23+30+13+19+15+5 : 119.4ma

 

[ppl]

I would use these regulators from Analog devices thay are good with inputs twice that of the TI part 12 volts vs 6 volts for the TI part. this allows greater power supply voltage flexibility, http://www.analog.com/en/prod/0%2C28...3301%2C00.html

 

[00940]

they are awfully similar to the REG101, aren't they ? Same cost too, compared to the cheaper TPS79333.

 

[00940]

Another option would be to use one more beefy REG102 for the whole digital section, except the "analog" section of the CS8416 which would have its own REG101 regulator to safeguard PLL performances. Then, people using batteries could be a bit more free. The rail splitter should see in between 10.4VDC (a minimum) and 18VDC.

I guess that would also please those who think that there are too many regs onboard

 

[ble0t]

I'll update the PS section to be more modular and thus accomodate anyone who wants to feed the +/- DC voltages in directly.

As for the capacitance existing on the positive rail, I will add some to the negative rail to balance it out a bit. Good catch on that one

As for the other voltage regs, I agree with 00940...the Analog part looks like a competing part to the REG101...and it definitely is more expensive.