[n_maher]

Quote:

Originally Posted by Alf Guys, first of all I would like to apologize for abusing your patience. I have not had enough time to spend the design. It is VERY time consuming exercise and requires a great deal of concentration, which unfortunately I could not provide in the past months.

Alf, don't sweat it man - First you've only got so many hours in a day to get stuff done and if you're anyting like me DIY usually comes at the end of the day which means some days it doens't happen at all. Second, after reading your list of options I can tell you there's probably not going to be any consensus with that many variables. My opinion would be to go with the design that offers the simplest approach which still giving builders options that they can add for themselves. You can please everyone all the time

.

Nate

 

[00940]

Quote:

Originally Posted by Nisbeth Hey, as someone who does technical marketing for a living I take offense at that Knowing how to make the maximum out of the parts and using that knowledge to measure impressive specs (that sometimes borders on cheating..) is referred to as "specmanship" among the engineers where I work. Some consider this an artform

Indeed. Is it said anywhere that the measurements were taken on the evaluation board ?

 

[steinchen]

Quote:

Originally Posted by Alf I need your feedback on the following prospective changes: ...

1) remove TPS6734

2) if it doesn't hurt just leave it in and make the part optional

3) I don't need the headphone driver. If you're going to implement it I'd advocate for MAX4410 and against AD8397 since AD8397 is a very finicky chip

4) Fix the grounds

 

[Francis_Vaughan]

Be careful when you worry about the grounds. The idea of somehow improving things by separating the DAC proper analog grounds from the digital ground with a resistor is widely discredited, and if you read back to my initial commentary on the layout you can see the reasons why. "Fixing the grounds" by cutting the ground-plane and installing a resistor between them will seriously damage the performance of the DAC.

I will repeat the main point - a DAC is not an audio frequency device. Nor is it a simple digital device. The usual ideas of grounds in audio simply do not work. The only way of thinking about these systems is as a complex RF device which has significant switching energy that must be controlled. This cannot be done with a split ground-plane. At any moment current will be flowing through the wafer from any one of the +ve inputs and essentially flowing out via any one of the ground pins. If you cannot reason about the return ground path for every single possible switching current you will create a system which will induce significant energy into other parts of the wafer - energy that is very likely signal correlated and capable of seriously degrading audio performance. Splitting ground-planes will create such a system.

 

[Alf]

Quote:

Originally Posted by 00940 Indeed. Is it said anywhere that the measurements were taken on the evaluation board ?

Good point! It must be one of those "specmanship" tricks. I don’t feel that bad anymore.

 

[Alf]

Assuming that a headphone driver is a go, which chip should we use? My only concern about MAX4410 is its availability in the UK (and presumably Europe). Otherwise it seems to be great. There are some other good chips around. LM4908, TPA152, MAX9722 – to name a few. Has anyone tried these chips? Which one is considered to be the best nowadays?


There is one other thing that bothers me lately. If these headphone drivers are as good as their specs say they are, why people build CMOY and other entry-level headamps?

 

[MASantos]

The headphone driver would discard the need for and headphone amplifier right? If so, I don't really agree with this implementation since it will not allow the use of other amps such as mints or pimetas. If it is possible have this driver as an option and still be able to have a "line out" signal then it might be useful in some cases.
Adding the the headphone driver will change this project from a DAC to a all in one DACplus amplifier board, which was not the original goal.
But if both options are possible that's a plus!!

Manuel

 

[Nisbeth]

Quote:

Originally Posted by Alf There is one other thing that bothers me lately. If these headphone drivers are as good as their specs say they are, why people build CMOY and other entry-level headamps?

AFAIK noone's really tested these drivers in depth, so very few are able to offer qualified comments on SQ. Also, a CMoy or other small amp is much easier to layout and assemble than something which involves SMD's. After you get a few CMoy's running, most people "graduate" to building something more complex but a MAX4410 is not really a good way to practice soldering

/U.

 

[dviswa]

Quote:

Originally Posted by Alf What do you think?

Quote:

Originally Posted by Alf 1) I would like to remove TPS6734. It does not do any good. RMAA show much better figures without it. In USB-powered configuration, REGs will be connected to USB directly. I tried this and it works very well.

I did not use this bus powered option, so no comments.

Quote:

Originally Posted by Alf 2) LM317 might be useful but NOT required for the project. I am thinking about removing it as well. If someone needs it, Tangent’s TREAD would be an obvious remedy.

Not sure I agree on this. I am using the DAC with a cheap wall wart. That would not be possible without LM317 and a TREAD is much more costlier, so I would suggest leave it in.

Quote:

Originally Posted by Alf 3) Headphone driver. I am not sure whether we should do it but many people asked for this. We could use either MAX4410 (same as ble0t used in his DAC) or AD8397 (see Tangent’s PINT). Adding a headphone output makes the board bigger and therefore more expensive. The target enclosure still will be Hammond 1455C80x. Any alternative suggestions are welcome.

I do not need this, so as long as this is an optional thing, I think it is OK.

Quote:

Originally Posted by Alf 4) Ground plane. While there may be some improvements to the sound, the RMAA figures are still far from the official datasheet. The only thing in our design that differs from the evaluation board is separate analog and digital grounds. All our revisions of the DAC feature a single ground plane. And all our revisions fail to reach the target figures. I suspect that the single ground plane is the reason for this. While it may be a good idea for bigger boards, our board is just too small. It is not capable to fight digital noise effectively due to its size. The noise penetrates the analog ground and affects the RMAA figures. Fixing the grounds should fix the figures.

The DAC when it was not yet cased up, had a very small amount of hiss, barely observable. After I put it in a Black Hammond case, it got really black

Seriously, this is one real black DAC even at full volume. Fiddling with the grounds I think moves it back into early Alpha, rather than at a Beta stage. I seriously doubt I can hear any further improvements, if any.

Dinesh

 

[Alf]

Quote:

Originally Posted by MASantos The headphone driver would discard the need for and headphone amplifier right? If so, I don't really agree with this implementation since it will not allow the use of other amps such as mints or pimetas.

While you will not need to use an additional amp, there still will be an option to use it if you would like to. I plan to add 2 line outs: before the output capacitors and after.
Quote:

Originally Posted by dviswa Not sure I agree on this. I am using the DAC with a cheap wall wart. That would not be possible without LM317 and a TREAD is much more costlier, so I would suggest leave it in.

You do not need LM317/TREAD to use the DAC with a cheap wall wart as long as it is <=10V. The REGs will do the job.

 

[Alf]

OK, I will not touch the ground.



I have been reading MAX4410 spec and found this:
Quote:

Proper layout and grounding are essential for optimum performance. Connect PGND and SGND together at a single point on the PC board.

Two ground planes connected at a single point. Should it be disregarded? If not, how these grounds should be connected to the DAC ground?

 

[cire]

i sent you a PM regarding this a while ago, but can i get the eagle files for the DAC?

 

[Francis_Vaughan]

Quote:

Two ground planes connected at a single point. Should it be disregarded? If not, how these grounds should be connected to the DAC ground?

No, don't disregard this. We are no longer in DAC land. We are back in audio land. The charge pump runs at 320kHz, which is not RF (not really nowadays anyway.) The suggested seperation of groundplanes makes perfect sense here. It is done to control the imposition of power supply noise into the audio ground of the chip. But the audio ground is most likely a continuation of the DAC groundplane.

In fact keeping the charge pump noise away from the DAC is a good idea too.

 

[Alf]

Major changes:
* Removed the buffer, the switch, and the LM317.
* Added optional LEDs. You can choose which one to use.
* Added output ground capacitor for virtual ground amps.
* Eliminated vias in output traces.
* Improved layout around PCM2702.
* Added 2 ground pads.
* New board size 48.6mm x 50.5mm.


Things to do/discuss:
* Change IC2/IC3 to SOT223-5. This will allow TPS793xx to be used instead of REGs. TPS793xx have higher PSRR and may be better suited for the application. (EDIT: Changed TPS30xx to TPS793xx)
* Increase the size of the USB receptacle mounting holes. I had no problems whatsoever but other people reported the holes are too small.
* Re-introduce the buffer and/or LM317. I am not sure about this. Is it going to be used?


I decided not to add MAX4410 at this time. I think a separate board should be designed.


Board v1.0 Rev 1 - Top

Board v1.0 Rev 1 – Bottom

Board v1.0 Rev 1 – All without bPlace

Schematic v1.0 Rev 1

 

[guzzler]

1) I personally liked the buffered ground, it was one of the most frequently asked questions about the old DAC for portable use.

2) The LM317 wasn't so necessary and very easy to do off board if needed.

3) How's the availability and cost of the TPS30xx cf the REGs? At the end of the day, it's going to make a pretty marginal difference for this board.

4) Could the ground pads be moved directly next to the power inputs, makes life simpler for wiring, and shouldn't obstruct any current routing