Amplifier for driving headphones

Hello Pavel47,

This is a very general question making it difficult to respond. 

It is not often you find a Class-D amplifier driving a headphone due to the RFI that would be produced with a headphone cable. Then the inductance is somewhat unpredictable which can be an issue with Class-D amps. Due to this usually you will find Class-G amps used for headphones because they will reference to ground. This takes care of the issue if someone uses the headphone output to go to another device's input. There are some capless output designs where there is a reference to the common mode voltage not ground. This is fine for headphones where the amp is located in the headphone itself. There are also some Class-D amps for headphones (we have some) but they are meant to be located right at the driver in the earcup not in the device where a cable would be used. 

We have a lot of codecs some with DSP and other with only converters that have outputs for driving headphones. 

In addition, there are some linear amps we produce that were designed when there was a war of cell phone makers to make the best sounding headphone output of their phones. So we have a dual opamp that is meant for driving headphones. 

The short answer to your question is that most smartphones and tablets use some sort of codec with a headphone amp built in. 

What are you designing and how many units do you think you will produce? What are the design constraints? 

Dave T

Hello Dave,

Thanks. Headphone interface it's just one of the options required.

Ideal device I'm looking for is as follows (full duplex):

• input1 - micro, output1 - I2S
• input2 - I2S, output2 - headphones (speaker)

Hello Pavel47,

Look at the ADAU1787. This part will do what you listed. It has two serial ports and a headphone output. It also has a fast DSP core and a SigmaDSP core so it is quite a capable part. 

The ADAU1372 is just a codec and can drive headphones but it only has one serial port. 

The ADAU1772 is the same basic codec as the 1372 but it has a fast DSP that can do some filtering and limiting. 

The ADAU1777 is the same basic part as the 1772 but it can process faster for lower latency and is a BGA footprint. 

There are other codecs from our Maxim portfolio that might be suitable. Your design constraints will help to narrow this down. 

What are your constraints? 

Power? dissipation for the entire part and output for the headphones?

Size?

Footprint? 

Voltage rails available?

Then there are the signal processing needs that you have detailed a little. 

Dave T

Hi Dave,

Thanks ! Indeed ADAU1787 is what I'm looking for.

Is there a device that manifests even more integration, i.e.

• input1 - micro, output1 - S/PDIF
• input2 - S/PDIF, output2 - headphones (speaker)

If not, does exist a device that do the following:

• input1: I2S, output1 - S/PDIF
• input2: S/PDIF, output2 - I2S

I've found at TI two quite old devices that do separate functions (only TX or only RX): DIT4096 and DIR9001

Hello Pavel47,

SPDIF is getting more difficult to find because there is less call for it in the industry where there is a high volume of sales. So this means that chip makers do not design it into new parts. We have the ADAU1452 that has SPDIF in and out along with the sample rate converters to properly interface it into a system but the part is much larger, more power hungry and does not have any converters. So you would need to use a codec of some sort. 

Dave T

Thanks Dave,

If not S/PDIF, does exist other solutions to transmit/receive audio over single line ?

Hello Pavel47,

How far do you need to run the line? 

PDM is close to a single line, it is a clock and data and can run a little further but not too much more than a meter? I am not exactly sure how far but a far distance is relative to the technology being used. 

We have the A2B network standard that can go far on a twisted pair but again, this is relative and it is much more complicated to design in. SPDIF/AES3 was good for this sort of thing. You can use a PWM signal but it has its own set of issues. 

There is still good old analog. It still works well for the right situation.

Dave T

Hi Dave,

DaveThib said:

How far do you need to run the line? 

In my application, this is not an issue because the medium is not coaxial, nor optical cable. What matters is the standard that allows audio to be transmitted on a single line.

Hello Pavel,

Impedance and length is everything even with a single line standard. Even with SPDIF you need to pay attention to the transmission line details. Length will determine capacitance which will effect the signal. Even with good old analog, length and impedance is still important. I recall years ago measuring the roll off of the top end with a single ended wire going about 100 ft from one room in a house to another. This was with high quality cable! differential low impedance would have been needed. If your medium is not coaxial or optical then what is it? 

Dave T