• State TI Thinks Resolved
  • Replies 17 replies
  • Answers 9 answers
  • Subscribers 30 subscribers
  • Views 265 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

TPA3255: Using lower Lout than recomended

Zeng.
Prodigy 50 points
Part Number: TPA3255
Other Parts Discussed in Thread: TPA3223, , TPA3220, TPA3221, TPA3245, TPA3223EVM

Tool/software:

Hello

In my specific application I'm forced to use significantly lower output filter inductance than usually recommended: ~1-2uH vs 10uH. 

Output filter capacitor can be increased to maintain the same BW. 

It's clear to me that such a change will increase ripple current, losses and eventually limit/decrease the output power (as peak current limit will be reached at much lower RMS value).

I'm interested to hear if I'm missing more implications and practical recommendations for such design change

Thanks 

  • 0
    TI__Expert 3785 points

    Hi Zeng,

    There are a few concerns with this approach. First as you have noted, the inductance values you have noted are below the datasheet minimum. With an inductance this small, some of the protection circuits in the device might trip resulting in no outputs. Your Q of the filter would also be larger resulting in poorer frequency response. 

    Regards,

    Ramsey

  • 0 in reply to Ramsey Foote
    Prodigy 50 points

    Hi Ramsey

    Thank you for the response. Could you give more details about the protection circuits? 

    Also, I forgot to mention, my load is very much below the spec as well, around 0.5Ohm with some ~10uH.

    Could you also to comment on impacts of those changes on loop performance, PSRR, etc?

  • 0 in reply to Zeng.
    TI__Expert 3785 points

    Hi Zeng,

    We cannot guarantee proper functionality when the datasheet is not adhered to. By going below the datasheet spec in both load and output inductance, you are will run into issues with the OC protection and possibly the DC detection protection. Since the output when switching is a low impedance node, I would doubt that there would be much of an impact to loop performance of the amp itself, but with the low inductance and high ripple you will see artifacts of this on the output. I would recommend to use a lower power device given how much room on the output you will lose to ripple.

    Regards,

    Ramsey

  • 0 in reply to Ramsey Foote
    Prodigy 50 points

    Hi Ramsey

    Thank you so far. I have a slightly different question about TPA3223. In its data sheet on p.33 in Typical application section, it says "TPA3223 can be configured in mono PBTL mode by paralleling the outputs before the LC filter (see ) or after the LC filter." Is it a typo or paralleling before the LC filter actually can be done? Also, on the same page, on Figure 10-6, only one input is used while another one is shorted to ground. Is it how PBTL should be realized?

  • 0 in reply to Zeng.
    TI__Expert 3785 points

    Hi Zeng,

    The TPA3223 only supports post filter PBTL (paralleling outputs after the LC filter). To configure the device in PBTL, IN2_P and IN2_M should be grounded as shown in the figure while IN1_P and IN1_M is your analog input.

    Regards,

    Ramsey

  • 0 in reply to Ramsey Foote
    Prodigy 50 points

    Hi Ramsey

    I'm probably missing something, but when IN2_P and IN2_M are grounded, what will be at OUT2_P and OUT2_M? I guess that they will be also at zero. What's actually paralleled then?

    Also, are you familiar with devices that allow paralleling before the filter?

    Thanks

  • 0 in reply to Zeng.
    TI__Expert 3785 points

    Hi Zeng,

    When the two inputs are grounded, the device recognizes this and puts the device into PBTL mode. OUT2_P and OUT2_M will both be modulating and will be tied together after their filters to make one leg of a differential output. PBTL turns the device from a stereo differential output to a mono differential output. Figure 10-6 shows the schematic for this configuration.

    The TPA3221, TPA3220, TPA3255, TPA3245 are some of the devices that support pre-filter PBTL.

    Regards,

    Ramsey

  • 0 in reply to Ramsey Foote
    Prodigy 50 points

    Hello Ramsey

    We bought TPA3223's EVB (TPA3223EVM) and it seems that it flags overtemperature warning well before reaching the expected output power. In our setup (MONO BTL with 4Ohm load and 41V power supply) the OTW_CLIP LED lit up at 110W into the 4Ohm load, while the spec mentions 200W as the power limit. Do you know if we are missing something? 

    Thanks

  • 0 in reply to Zeng.
    TI__Expert 3785 points

    Hi Zeng,

    A few things are likely going on. The OTW_CLIP pin indicates two things, over temperature warning and clipping. That 200W limit mentioned is at 10% THD which is heavily in the clipping area, so this pin will always be asserted. From the THD vs Pout plot in the datasheet, the clipping will begin right around that 100W mark so I suspect the pin is being asserted due to the clip detect.

    In addition, OTW will not shut down the device. The device can keep getting hotter until the fault pin triggers due to the temperature going to the OTE threshold.

    I would recommend to look at your THD+N and PWM outputs to ensure the clipping is not the error.

    Regards,

    Ramsey

  • 0 in reply to Ramsey Foote
    Prodigy 50 points

    Hi Ramsey

    Indeed it makes sense that some small hardly noticeable clipping starts at 100W. Thanks for the explanation.

    I have another question about TPA3223 application. In our case the load is grounded  at one end. To me it looks reasonable to use a bipolar power supply in such case as I don't want to add a DC block capacitor at the output. In such case a half bridge solution is sufficient. Please advise what do you think is the best way to handle it. 

    Thanks

  • 0 in reply to Zeng.
    TI__Expert 3785 points

    Hi Zeng,

    There would be a few problems with this approach. By using a bipolar supply, your I/O signals will need to be referenced to the negative voltage level. It would also be easy to accidentally damage any ground referenced equipment or accidentally short the negative line. If you don’t want to use a DC blocking cap, you could use a voltage divider (id recommend two 100uF caps to make the divider) on the other leg instead of ground.

    Regards,

    Ramsey

  • 0 in reply to Ramsey Foote
    Prodigy 50 points

    Hi Ramsey

    I'm probably missing something. Where would be ground connection in case of such voltage divider? Shouldn't it be at the middle point (between the capacitors)? If yes, how is that different from using a bipolar power supply (besides DC frequency impedance)? 

  • 0 in reply to Zeng.
    TI__Expert 3785 points

    Hi Zeng,

    Instead of using a bipolar power supply and grounding the other side of the speaker, you would use a single power supply and bias the other end of the speaker to PVDD/2 using the voltage divider. This removes the risk of accidentally causing a short, would make using the digital signals easier, and wouldn't cause ground referenced equipment to have issues. This will also help reduce pop noise that would occur if the other terminal of the speaker was connected to ground with a single rail power supply.

    Regards,

    Ramsey

  • 0 in reply to Ramsey Foote
    Prodigy 50 points

    Hi Ramsey

    Unfortunately, in our application that wouldn't be possible. The load is already grounded and also the power supply that I use to drive the amplifier is referenced to the same ground. Though I can add an isolated power supply or bipolar power supply in between. What would you recommend in such case, assuming that having a serial DC block at the output is not a good option for other reasons?

    Thanks

  • 0 in reply to Zeng.
    TI__Expert 3785 points

    Hi Zeng,

    I would recommend that you change application. A bipolar supply might work for the device itself, but the additional complexity (low level communication like the reset pin, possible shorting of supply, risk to ground referenced measuring equipment) would almost force a re-design where adding a DC block or a voltage divider would almost certainly be easier. If you need to use a bipolar supply you can, but this has not been validated by us and you would need to be incredibly careful with your design.

    Regards,

    Ramsey

  • 0 in reply to Ramsey Foote
    Prodigy 50 points

    Hi Ramsey

    Thanks for the prompt response. Maybe other TI's class D ICs can support a bipolar supply more easily?

  • 0 in reply to Zeng.
    TI__Expert 3785 points

    Hi Zeng,

    Not by much. All of the difficulties I have mentioned could be designed around and with care can be done. To take the low level signals as an example (like the reset pin), you would need to drive the signals to -PVDD/2 + 5V since the signals are in reference to the "ground" pin which is -PVDD/2 for a bipolar supply. This is possible with level shifters, but makes the design more complex. You would also need a second power supply for AVDD and GVDD which would be -PVDD/2 +5V. Again this is possible, but would make the design more complex. 

    If you want to use a bipolar power supply, you would either need a much more complex circuit or an amp that has no communication nor interface pins which we do not have.

    The TPA3126D2 could be a bit easier as it only needs 1 power supply, but you will still need to work about the low level communication (like the shutdown/mute controls). For all of these higher power amps, a bipolar supply has not been validated so you would need to do your own testing to make sure it is to your specs.

    Regards,

    Ramsey