TAS5751M: Using 3 parts, I2C and I2S questions for a single Cortex-M controller.

Part Number: TAS5751M

I am designing a system that requires 3 of the TAS5751M parts configured in PBTL configuration so there is about 50W for each channel driving a 2 ohm load from a supply power range of 12 to 24 vdc.

Perhaps there might be a better part for this application.  I don't need the equalizer or some of the other items in this part, however it is so cheap for the power level that I don't mind using it and nulling out any of the signal flow path items that we don't need for the application.   A 50W mono, I2S, I2C, wide variation supply part was my search criteria.

My Questions:

1) Is there a "best" or "recommended" method to drive the I2C on these such that I can give them 3 unique IDs?

I know the ADR/FAULT pin can help config. two parts, but what about 3 parts?  Ideas/suggestions being the Cortex-M has one I2C channel.

I don't want to have to put one of the amps. on bit banging pins from the CPU if at all possible.

I can spare a CPU pin to maybe hold one or two parts in reset while initializing them to unique ID's?

2) My application will have either 12vdc or 24vdc battery power.  I have already swept through this range with the TPA3116D2 part and found that it appears to handle this with out a noticeable change in output volume levels.  Great!  However is this ok to assume the parts can simply handle that variation in supply voltage and maintain output level?   I was thinking to have my Cortex-M monitor the supply and make any adjustments via. I2C to the ampl. chips to maintain output level if I have to do it that way?  Any ideas/thoughts on this?

3) Does TI have any sample I2C initialization source code (in C) to setup these parts in an embedded design, a library or the like?

4) I am intending to drive all 3 ampl. IC's with one I2S signal that is branched to all three ampl. parts. Will the three ampl. parts be in phase/sync at their outputs or could there be some phase error between the signals one would see at the outputs (speakers).  This is a six speaker acoustic wave warning device for marine use and to have the resonance in the audio in the chamber the drivers must all be in phase.

Thanks for ideas/suggestions/recommendations ahead of time,


  • Hello Marc! Alex can look further into your technical questions, but would any of our simple analog input (TPA31xx or similar) parts work for you as well? Thanks, Jeff
  • Hi Marc,

    1) Yes, this is possible and you are on the right track. The ADR pin on the device allows you to power up two devices with two unique i2c addresses. Additionally, once powered up, additional unique addresses can be set through software by writing to registers 0xF8 and 0xF9.

    Please refer to the I²C Device Address Change Procedure section of the datasheet for these steps:

    1. Write to device address change enable register, 0xF8 with a value of 0xF9A5 A5A5.

    2. Write to device register 0xF9 with a value of 0x0000 00XX, where XX is the new address.

    3. Any writes after that should use the new device address XX.

    Since you only have 2 unique addresses at power up you will need to hold the third device in reset while configuring the second to avoid changing the address on both devices.

    2) Please refer to the Typical Characteristics - Mono PBTL Mode section of the datasheet. It provides THD+N vs. Output Power graphs at 24V, 18V, and 12V. At 12V and 2ohms you can only get up to 30W at 10%THD+N. I compared this to the same graph in the TPA3116 datasheet, which can get between 36-39W under the same conditions. So your distortion with the TAS5751 will be a bit worse than the TAS3116, but if this is not a problem both can hit 50W with greater than 10% THD+N distortion.

    3) I am not aware of any i2c sample code available. However, the TAS5751 EVM comes with software for configuring the device and it is possible to record the i2c commands sent. This may be a good starting point for you.

    4) Since I2S is clocked the devices will be in phase as long as the clocks are synchronized.



  • In reply to Jeff Kohnle:

    Yes, I mention the part I'm currently evaluating (TPA3116D2) because it is easy to use. However I have a CPU which is making the audio from a formula as a wave file which I wish to send digitally to the amp. chip. Of course I could D2A that and send it analog, but why risk interference from the other digital elements by having a 2" trace that is analog in a digital world? The TAS5414C looks amazing too, 4 channels!
  • In reply to Alex Bhandari-Young:

    Thanks for the quick reply Alex.  You covered my questions pretty well.  I am still wondering about question 2.  I am designing for 12vdc operation and if the user connects 24vdc (2 batteries in some applications) I would see that with my controller and drop back the volume or gain such that is performs like the 12vdc supplied system.  I.E. my output power levels would still be in specification with either voltage. I really don't need the 24vdc but need to tolerate it.   A side point, I also have to provide protection if more than 26vdc is applied.  I am thinking that I would monitor this with the controller and not raise the PDN line on over voltage conditions.  Would the device be protected as even though it is not "running" but in a power down state?  What absolute voltage can it withstand in power down state?

    I might have to add a MOSFET to the supply line to cut them that way by using my controller to monitor this situation.

    A side note or question, this application requires a high voltage (around 60vac) to the speaker coils.  I am thinking of stepping up the 20vac that I am seeing out of the TPA3116D2 part right now to what i need.  A 2.5:1 transformer would do the trick.  Do you know of any issues driving a transformer load with this part to the final speakers?  I would impedance match the loads, although it will be somewhere in the 2.5 ohms range (two 4.4 ohm driver coils in parallel).  Also there will be 3 sets of these so I most likely will use 3 of the TAS5751M parts in PBTL configuration do drive them all.  The easy thing is I have a fixed sine wave of around 1kHz that is pulsed in patterns as a marine navigation warning device.

    We require approx. 1 amp at the each speaker coil, this will translate to around 2.5 amps from the Class-D amplifier which these parts should be able to handle.  Does this all sound reasonable to your expertise?  (using the TAS5751M instead of the TPA3116D2 I'm playing with now).


    Marc Y.

  • In reply to Marc Yaxley:

    Hi Marc,

    I see, I misunderstood your second question. The TAS5751 is an open loop device. This means there is no feedback from the output to input and it cannot compensate for changes in PVDD to maintain a constant volume. The device you tested, TPA3116, is closed loop. The best solution if you need to maintain the same volume at different supply voltages is to choose a closed loop part.

    For absolute ratings see the absolute max section of the datasheet. The 5751 for example is 30V. The device will not break if this voltage is not exceeded. This voltage should never be exceeded regardless of whether or not the devices in powered down. If you need to tolerate higher voltages some sort of extra protection might need to be added (TVS, load switch, etc), but I am not an expert in those devices so I would point you to post in our circuit protection forum if you need more information.

    Yes you should be able to drive a transformer. To avoid distortion and losses you will want to choose a good quality transformer with low distortion, insertion loss, DCR, etc.


  • In reply to Alex Bhandari-Young:

    Ok, I was wondering about that open/closed loop feature, having more or less glossed over it.  I see now that you mention what it does why I'm seeing what I want to happen in the TPA3116 part when the PVCC changes radically.  This is important so now with that in mind I must look at closed loop parts.  In the same family with the TPA3116 is the TAS5414C which is a 4 channel device.   I can connect two of the four BTL amps into the PBTL configuration for a ton of watts (I only need around 60W for my driver coils) and have way good reliability and less heat perhaps.  I would in this case use two of them as I need 3 channels of 60W to drive my full array of horn drivers.  I would just not use two of the channels on the 2nd device.

    This also frees up the I2C issue as there would be 2 devices.  The only down side is I will have to DAC the signal in my CortexM CPU and pass that as analog to the ampl.  I was trying to only have analog signals within the ampl. itself with the I2S ability of my CPU and some amps.   Not a show stopper though for sure for the higher power and feedback control loop of output levels.

    I think I will add a power MOSFET to the hight side to not supply PVCC if it is over say 28vdc to protect the amplifiers parts.  The CPU and rest of the board will have full regulation of course and can take that voltage.

    Maybe you know of another part for my 3rd channel which is a 2 channel or 1 channel in PBTL and I2C which would fit that need.  The costs are way decent on TI's parts so I don't mind not using one of them fully.

    Looks like this part of the design is pretty narrowed down and I'm on the right track to a total of approx. 150 to 180 watts of reliable audio power.

    Thanks to TI for the assistance and advice, Marc Y.

  • In reply to Marc Yaxley:

    Hi Marc,

    The part you chose looks like a good choice to me. Glad to help!