• State TI Thinks Resolved
  • Replies 3 replies
  • Answers 2 answers
  • Subscribers 30 subscribers
  • Views 93 views
  • Users 0 members are here
Support feedback
Options
Options
Related

TPA3220: Overcurrent triggers early

Dale Ulan
Prodigy 10 points
Part Number: TPA3220
Other Parts Discussed in Thread: TPA3221, TPA3221EVM

Tool/software:

I have a design using the TPA3220 in PBTL mode, parallel connection is made before the filter. It is being used for a musical instrument application, not for mixed audio. Originally I had two TPA3220s on a 4-channel amplifier  PCB and it worked fine until testing with a real speaker load. My specification was to drive a 4 ohm load and it did that, but in my testing with the 'real load' showed an impedance that dropped as low as 2 or 2.5 ohms due to mechanical resonances in one particular speaker design. So I redesigned the PCB to use the same IC, just four of them in PBTL mode thinking that should do the job - but it didn't. One thing that I needed to do was run the modulation mode in the lower-efficiency one that idles the output at 50% duty cycle; the nature of the audio signal from this instrument would create distortion in the high efficiency modulation mode. The four amplifiers are set for master/slave operation.

I was expecting the amplifier to trigger overcurrent at around 18-20 amps (2 ohm impedance at 26V PVDD) but it is tripping at around 10 amps or 3.5 ohm impedance using a dummy load resistor. The amplifier is being driving to hard clipping (musicians seem to do this often) through the ill-behaved speaker as I described earlier. I was concerned about my PCB layout - although it worked fine in normal BTL I was a bit nervous about the issues that show up at higher current levels when I redid my layout in PBTL so I purchased the TPA3220 micro EVM and strapped the outputs to the same configuration (PBTL tied before filter) and I see the same thing happening.

The overcurrent protection appears to happen at a current level earlier than I would expect, but also this is happening during hard clipping - perhaps 30% of the waveform is clipped - so I don't know if that contributes to the shutdown. I can send images later on today or tomorrow if that is useful. It takes around 3 seconds to shut down although this varies by the impedance of the load. If the load impedance is set to 'just trips' then it takes about 25-30 seconds to shut down. If I lower the impedance, than with the same signal it takes shorter and shorter time, and at a 2 ohm setting it takes about 2-3 seconds to trip. My test signal is a triangle wave, but about 30% of the tops and bottoms are set to be hard-clipped by the amplifier.

Also, I notice that if the amplifier is unloaded it tends to trip as well.

Thank you.

  • 0
    TI__Genius 17560 points

    Hello,

    This device was designed to drive signals within the PVDD to ground rails. With that, the device has both overload and DC speaker protection features. I believe the main issue is that you are driving a very clipped signal which is causing the amplifier to fault. I believe increasing PVDD to a level such that the signal is within PVDD and ground (no clipping) will resolve the issue. Can you test this? You could try with max PVDD. 

    Also for future use you can use the below table to understand which specific faults are triggering. 

    Regards,
    Sydney Northcutt

  • 0 in reply to Sydney Northcutt
    Prodigy 10 points

    I can test that. I am just a bit puzzled as to why I don't really get to much lower of a load impedance capability in PBTL mode compared to BTL stereo mode or why the clipping should cause the overcurrent protection to trip. I suppose that amount of clipping is a bit unusual, or if device temperature was used internally to adjust the overcurrent threshold, these observations would make a lot of sense. Most the drive-by-wire H-bridge ICs I use (I design automotive ECUs most of the time) have that temperature-dependent graceful degradation function built in. I did check and I am getting OLP status above (/FAULT is low, /OTW_CLIP is high - though it pulses low during clipping), probably not UVP since I've probed for voltage drops on PVDD already.

    Also I see that there is a pin-compatible TPA3221 and the specs look pretty much the same except the minimum load impedance, maximum power ratings, and the location of the heatsink pad. Would it be worth it to use the TPA3221 instead of TPA3220? Or is it the same device but just with a heatsink on top instead of on the bottom? Thanks again!

  • 0 in reply to Dale Ulan
    TI__Genius 17560 points

    Hello,

    The clipping does not necessarily lead to over current, but it can lead to issues causing the amp to turn off the output. When clipping you are essentially pushing the amp to its max capabilities which causes it to possibly hit OT and or OC. The amp is essentially not designed for this kind of output. The OTW_CLIP pin signals clipping so it can be handled before the device reaches an OT or OC issue. 

    Yes PBTL allows for higher current capabilities which allows you to use smaller load impedances. The device still has it's same max thresholds in terms of temperature, but because the load is smaller you are able to push more power. Because you see the same result in PBTL and BTL, it may be the case that temperature is the issue here. 

    These two devices have the same OC threshold so this portion will not change. If it is an OC condition here, switching devices will have no change. Overall they are the same except for the thermal pad. TPA3221 is a pad up device so you can have better thermals with a heat sink on top which allows you to push more output power. It OT is the issue here I would recommend swapping devices. Maybe the best way to test this is with a TPA3221EVM? 

    Regards,
    Sydney Northcutt