Dual Purpose Headphone / Line Out using differential "capless" mode...

Hello Tony,

See http://www.ti.com/litv/pdf/sloa143.

and

http://e2e.ti.com/videos/m/audio/97359.aspx

and

http://www.ti.com/litv/pdf/slyt023 (page 34)

In the DAC32, we have several modes of operation for the headphone amplifiers: differential, pseudo-differential and single-ended.

Differential mode is typically not used with interfacing through a standard 3.5mm jack.

In Pseudo-differential mode, the HPCOM is the virtual ground (capless) and is sometimes connected to a 3.5mm jack for headphone use. This might be suitable for driving headphones and floating loads, but might cause issues if the ground of the receiver is tied to earth ground and your device (if conected to USB) is connected to earth as well. The issue is that HPCOM is typically 1.35V, so this could be pulled by the receiver to ground relative to our DAC32's AVss.

The single-ended (cap) mode is the safest to interface with unknown load types. To avoid ground loops, the best that can be done is to ensure that both AVSS_DAC and DRVSS are routed as close as possible to the ground pin of the jack.

Other techinques may require large PCB areas (isolation, etc.).

If you would prefer a codec with ground sense and direct path, take a look at http://focus.ti.com/docs/prod/folders/print/tlv320aic36.html.

A more cost-effective solution is to use our DAC32 along with a TPA6132A2 http://focus.ti.com/docs/prod/folders/print/tpa6132a2.html in which you could apply the techniques mentioned in the video above.

Regards,

J-

Thanks J, I've been through the material more than a few times over the weekend.

The TLV320AIC26 looks a nice device.  Expensive part just for playback, going to add cost to go multi-layer (trying to avoid BGAs in the design) and would incur more development time.  

I have been looking at using the TPA6132A2 as an amplifier for the DAC32 - purely for the directpath feature.  The datasheet mentions the inputs can be driven differentially, given they are less than +/-1V.  I'm unsure if this limitation would effect the volume control range, or facilitate noise entering into the system.  Using series capacitors was what i was hoping to avoid with the capless operation, but I suspect I might need them for the inputs to the APA.

I've looked at the TPA6130A2 as well.  Specifically to provide the volume control without series caps from the DAC32 (set as fixed output level, below +/-1V).  It's starting to add up in power consumption and board area, and unsure on the potential opportunity for noise to be introduced.

While the TI chipsets are quite impressive, they aren't quite providing an elegant solution for this application.

Do you have any suggestions for this application that may help favour the TI chipsets?

The requirement for having both the USB cable connected and audio playback running is not typical use for the product.

If pseudo-differential mode was used, could the DAC32 be damaged from the ground loop connection?  Or would it result in audio distortion?

Would the addition of a ground break loop resistor reduce the risk of damage and distortion effects?

Hi Tony,

The DAC32 has short circuit protection to limit the current. If the sleeve is connected to HPLCOM and HPRCOM, then might be seen as either DC float if not grounded, or as ground potential (short circuit if the impedance between grounds is small) if they share the same ground. However, this can result in high currents and distortion.

Since in your application the same jack is dual purpose (i.e. headphone and line-out) the safest would be to use single-ended (cap) configuration. It is also the most power-consumption-friendly configuration, since only two of the drivers (L/R) are used. Alternatively, the TPA option would provide best ground loop immunity and compatibility at the same time. Also, it has relatively low bias current when active (e.g. 3.1mA). Using the TPA also allows to add a small low-pass RC filter between the codec (fc ~34kHz, 4.7nF, 1kohm single-ended, 500ohms if connected differentially) to remove out-of-band noise of the sigma delta converter. This allows best noise performance when interfacing with an external unfiltered class-D amplifier through the output jack. The AC-coupling cap between both can be a small value (depending on input impedance selection of TPA).

I will forward this message to the TPA folks in case they have additional comments.

Regards,

J-

The option to reduce the out of band noise for the sigma delta converter is an attractive one!  As is using the TPA to provide the best ground loop immunity and compatibility.

Which leave the volume control.  Specifically, the limitation of driving the inputs to the TPA6132A2 with less than +/-1V.  Will using the DAC32 with the TPA facilitate full volume adjustment?

Hi Tony,

On DAC32, the output gain and DAC mix can be adjusted. The output gain is typically used to set a higher gain value when using a higher common mode voltage. The default for the DAC32 is 1.35VDC. This can be adjusted to 1.8V for higher headroom (register configurable), which is typical for 3.6V supply. The codec+TPA signal chain can be set to a baseline gain and then you can attenuate from there via registers to obtain the desired level.

I cannot seem to find the +/- 1V specification for the TPA that you mention. Is it based on output power? I will let the TPA folks comment on this matter.

Regards,

J-

Hi J, 

I found the reference to the +/-1V on page 14, second last paragraph.

Understand the signal chain and use of gain.  Appreciate your helpful advice :-)

The options based on my current understanding…

1.  DAC32 with DC blocking caps.  Excellent compatibility, lowest power consumption, caps limit audio performance vs cost/size.

2.  DAC32 with pseudo-differential output (capless).  No good if a ground loop exists.  In this case, portable device with USB connected to the PC, while audio connected to an external mains amp.  Unless mains amp has AC coupled inputs.  Otherwise audio great,, lowest cost, less components, low design complexity.

3.  DAC32 with TPA6132A2.  Excellent compatibility, additional low pass filtering practical (best audio?), higher cost, higher power consumption, higher design complexity (PCB).

Hello Tony,

I believe that the 1V restriction in for cases where input capacitors are not used.

"1.  DAC32 with DC blocking caps.  Excellent compatibility, lowest power consumption, caps limit audio performance vs cost/size."

- 47uF is typical for space-constrained applications. The major audio performance limitation is due to the fc of the high pass filter created by the caps. Although 47uF is typically used for portable audio applications, a 100uF (50Hz, 32 ohm) to 200uF (25Hz, 32ohm) ceramic cap per channel yields good results for headphones. 100uF caps come in relatively small sizes (1206, X5R). You could also use a shelf filter to boost the lower end to compensate for a 47uF cap. Beware that this option will reduce the headroom in the digital domain (see http://www.ti.com/litv/pdf/slyt375 for details). You can use this tool to calculate biquad coefficients in case you are interested in going this path: http://focus.ti.com/docs/toolsw/folders/print/coefficient-calc.html.

"2.  DAC32 with pseudo-differential output (capless).  No good if a ground loop exists.  In this case, portable device with USB connected to the PC, while audio connected to an external mains amp.  Unless mains amp has AC coupled inputs.  Otherwise audio great,, lowest cost, less components, low design complexity."

- Capless is mostly suitable for systems where only headphones are used and there is an additional dedicated (cap) line-out jack. It is also good to drive internal speaker amps.

Regards,

J-

Hi J,

I have been over the details with the projects audio developer.  Using DC blocking caps was not a preference, due to the phase shift impact.  Using the TPA and RC filters to reduce the out of band noise was of interest, however there were concerns over layout / physical size and managing appropriate isolation with ground paths - as well as cost.  If only the DAC32 had DirectPath. The interest at present is with pseudo-differentiate outputs and a ground break loop resistor.  Interestingly, one of the primary audio concerns was with minimising clock jitter.

Your advise has been quite valuable J, very appreciated.

The evaluation kit has arrived and I'm keen to put it to good use.

So far for this product, we are using a TI micro, buck regulator, Li-ion Gas-Gauge and Charger, USB ESD protection and of course the DAC32.

TI rocks :-)