[Clutz]

What about the trace lengths between the ASRC and the PCM1798? Can we shift the ASRC to the right to reduce the length of those traces, or is that a non-issue?

Re: Power Supply - I like the 15VDC input setup. I think, like 00940 said, it provides an ease of use and it is capable of being flexible enough to use multiple supplies. I'll probably start out using a 15VDC (a tread, maybe a STEPS?) and then possibly later rig up three psu with three independent psus or something.

 

[00940]

The length of those tracks is a non-issue for me. Actually, having the AD1896 and its noisy decoupling a bit apart of the DAC seems a good thing.

I was however wondering if the I/V doesn't risk offset at the output. Should we add a trimmer to take care of it ? And if yes, how to do it best ? I was thinking of having a buffered voltage divider feeding one of the inputs of the output opamps. It will add like 5$ but it's still cheaper than expensive output caps

 

[Clutz]

Quote:

Originally Posted by 00940 The length of those tracks is a non-issue for me. Actually, having the AD1896 and its noisy decoupling a bit apart of the DAC seems a good thing.

Alright

Quote:

I was however wondering if the I/V doesn't risk offset at the output. Should we add a trimmer to take care of it ? And if yes, how to do it best ? I was thinking of having a buffered voltage divider feeding one of the inputs of the output opamps. It will add like 5$ but it's still cheaper than expensive output caps

I think it seems like a good idea. Good quality output caps are going to cost more than $5 - and keeping caps out of the audio path seems like a good idea too.

 

[ppl]

Quote:

Originally Posted by 00940 The length of those tracks is a non-issue for me. Actually, having the AD1896 and its noisy decoupling a bit apart of the DAC seems a good thing. I was however wondering if the I/V doesn't risk offset at the output. Should we add a trimmer to take care of it ? And if yes, how to do it best ? I was thinking of having a buffered voltage divider feeding one of the inputs of the output opamps. It will add like 5$ but it's still cheaper than expensive output caps

You probaly need to adjust the DC offset however once you find the right resistence a pot is no longer required and all remaining boards can use this predetermond resistor value as thay will all be close to one another. this makes the cost and complexity do way down

 

[00940]

In "theory" there shouldn't be significant offset I think. The differential outputs of the DAC are at 2.5V if my memory's good and this common mode voltage should disappear with the balanced to unbalanced converter. If offset there is, it will be mostly coming from mismatch in I/V resistors and the opamp themselves (an opa134 is specified for 4mV offset iirc). So the offset should be (very?) low but variable for any DAC build. The Belcanto DAC2 (similar topology but with the PCM1738) has no output caps but apparently uses an offset servo with the INA134.

Btw, can someone check my maths ? With a 44.1KHz sampling frequency, with a signal band of 20KHz, with "upsampling" (the AD1896) to 96KHz and 8X oversampling (the pcm1798's digital filter), the first alias should be at 8 x 96 - 20 = 748KHz ?

 

[Clutz]

In "theory" there shouldn't be significant offset I think. The differential outputs of the DAC are at 2.5V if my memory's good and this common mode voltage should disappear with the balanced to unbalanced converter. If offset there is, it will be mostly coming from mismatch in I/V resistors and the opamp themselves (an opa134 is specified for 4mV offset iirc). So the offset should be (very?) low but variable for any DAC build. The Belcanto DAC2 (similar topology but with the PCM1738) has no output caps but apparently uses an offset servo with the INA134.
/QUOTE]

Why not add the DC Servo circuit to the design - but before implementing it, try running the DAC without installing any of the servo's parts and measure the DC offset (i.e. after the opamp I/V, but before the servo). If it's too high, then we can put the required parts in, and if it's not needed then the that circuit just isn't completed- no harm, no foul, right?

 

[ble0t]

After getting some feedback from some of you about the possible desire to use this with batteries or other power sources, we've redesigned the power supply to be a little less picky and also cut down on the component count. Bascially, instead of using individual regulators for each digital chip, we've gone to using a single LT1763 with the already existing passive decoupling. The only thing to maintain its own regulation is the analog supply of the CS8416. Also, I've added some pads to the supply lines for anyone who wants to supply their own +/- voltages. Here are the updated files:

Schematic

Top Layer

Bottom Layer

Both Layers w/Silk

00940 is also working on implementing the servo to minimize offset. If any additional space is left, we will most likely implement an optional balanced output on the board as well. The next update will most likely contain these two additions.

 

[dsavitsk]

Page 16 of the PCM2707 datasheet says this "To avoid back voltage in self-powered operation, the device must not provide power to the pullup resistor on D+ while VBUS of the USB port is inactive."

Are you doing anything about this, or just sugesting that people not power the DAC with the USB cable disconnected, or nothing at all? The 2707 on one of my DACs burned up, and I am not sure of the cause, but I am wondering if this might be it?

-d

 

[Clutz]

When the device is connected to a USB bus, then VBUS is at +5V. Couldn't we use a VBUS to close a relay and power on the rest of the DAC, only when connected to a USB BUS? That way if the DAC was turned on while not plugged into a USB bus, the DAC itself would never be powered?

 

[dsavitsk]

That seems reasonable to me, although you might also just connect the pullup resistor to the usb's power which seems easier. As I say, I am not sure what caused one of mine to die, and while this seems like a possibility, I've not heard of anyone else with this as a problem, and there are some other things I could have done.

 

[Clutz]

Quote:

Originally Posted by dsavitsk That seems reasonable to me, although you might also just connect the pullup resistor to the usb's power which seems easier. As I say, I am not sure what caused one of mine to die, and while this seems like a possibility, I've not hears of anyone else with this as a problem, and there are sme other things I could have done.

It does seem easier to connect VBUS to the pullup. The relay idea seemed like a better idea to me at the time, but now I have no idea why that is, except that I think I'm the type of person who would over engineer things.

Anyhoo.

 

[00940]

The DDDAC for example uses indeed a small system to disconnect the pull-up. I overlooked the issue since the other chips of TI don't need it. We're gonna have a look on that.

 

[Clutz]

Quote:

Originally Posted by dsavitsk That seems reasonable to me, although you might also just connect the pullup resistor to the usb's power which seems easier. As I say, I am not sure what caused one of mine to die, and while this seems like a possibility, I've not heard of anyone else with this as a problem, and there are some other things I could have done.

Page 29, "D+ must not activate (HIGH: 3.3V) while the device is detached from USB".

One problem with this is that D+ is supposed to be pulled up to 3.3V and VBuS is 5.0V, so we'd have to make use of a voltage divider or something. Not that it's a terribly difficult thing to do, I'm just pointing it out for the sake of the design.

So this makes me wonder - what happens if the device is in self-powered mode, we connect D+ through VBUS using a voltage divider, and the usb cord gets yanked? What are the consequences of D+ being pulled high? If this causes damage to the chip, what how long can it safely tolerate this? I don't know how to get the answers to these questions, but it sort of makes me think that perhaps tying VBUS to D+ via a voltage divider and using a relay to turn the whole thing off might not be a bad idea. On the other hand, could quickly turning it off with the relay be damaging?

Edit: I was just looking at the schematic and I noticed that the D+ isn't pulled up at all, or is it me?

 

[00940]

The pullup is included in the emi filter iirc.

 

[Clutz]

Quote:

Originally Posted by 00940 The pullup is included in the emi filter iirc.

I wondered if that was the case. I coudln't find any datasheet for that part.

Coo coo I say.

Cheers,
Brad