• State Resolved
  • Locked Locked
  • Replies 16 replies
  • Subscribers 30 subscribers
  • Views 4976 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TPA3118 shuts down with inductive load

Ray Buck
Prodigy 210 points
Other Parts Discussed in Thread: LM1875, LM3886, TLC04, TPA3118D2

I have designed a TPA3118 board to use as a PA amplifier. The power supply is a 24 VDC, 3 amp unit. There is a transformer connected to the amplifier output that steps the voltage up to 70 volts. The transformer primary is 8 ohms and the secondary is 330 ohms. This translates to 15 watts at 70 volts.

I connect a 300 ohm load to the transformer secondary and drive the amplifier with a 1 kHz sine wave. I can drive the amplifier to full power for an extended period of time (over 15 minutes) and it continues to function properly. Testing with the signal generator, the output varies from 68.6 volts at 300 Hz to 74.4 volts at 15 kHz.

However, when I connect the voice announcement input source to the amplifier, I have problems. If the output exceeds roughly 7 or 8 watts, the amplifier goes into shut down. It will cycle back on after about 5 seconds. But continuing to make the voice announcement, the amplifier will re-trigger the shutdown cycle. Also, the source is switched into the amplifier input with a relay. If the volume level is set for more than about 8 watts output, the amplifier goes into shutdown when the source is switched on. Again, it comes out of shut down after about 5 seconds.

If I remove the transformer and connect an 8 ohm, 50 watt resistor to the amplifier, I do not experience the shut down issue. In fact, with the voice announcement source, I can drive the amplifier to 30 watt peaks with no shut down problems. Driving the amplifier with a 1 kHz sine wave to 30 watts for 15 minutes, there is no shutdown.

The transformer works fine with a class AB amplifier (LM1875 and LM3886) with the voice announcement input source. Using the LM3886, I can load the transformer secondary with 200 ohms (25 watts), drive it with a 1 kHz sine wave, and the output looks fine on my scope. So I don't think transformer core saturation is an issue.

I am using 10 uH inductors and the same resistor, capacitor values as shown in Fig 27 of the data sheet for the PBTL sub woofer circuit. But there has to be some kind of inductive kick back from the transformer that causes the TPA3118 to shut down.

Any suggestions will be appreciated.

  • 0
    TI__Expert 5000 points

    Hi, Ray

    I am on traveling this week, will look into it later.

  • 0 in reply to Peter Cao
    Prodigy 210 points
    Thank you Peter. I have ran additional tests. If I connect any source of varying material to the amplifier it goes into shutdown with the transformer connected. I drove the input with music and voice from a FM tuner. As I increase the input level, it eventually gets to a point where the amplifier starts cycling in and out of shutdown. At that point the amplifier is still well below the 15 watt output level.
  • 0
    Genius 3295 points

    Hi Ray,

    I'm pulling this out of my behind; TI has implemented in almost all of their Class D devices output protection that can seem a bit "overzealous", that is to say it works well, really well, to protect the device and speakers.

    If I were to guess, I'd say that you are seeing a net DC imbalance in the output when the transformer is connected. With a transformer as a load, I suspect the signal is not symmetrical. You could try putting a pot across the transformer with the wiper to ground and see if you can balance the output. 

    Mike Tripoli

  • 0 in reply to Mike Tripoli
    Prodigy 210 points
    Mike, thanks for the reply. Unfortunately that will not work. Each location the amplifier would be used in would have a different load. In fact, the load can change from one announcement to the next if different zones are used. Each zone would have a different number of speakers with matching 70 volt transformers. It may be that class D amplifiers are just not suitable for driving into a transformer load. I may just have to stick with class AB amps unless TI has a solution.
  • 0 in reply to Ray Buck
    Genius 3295 points
    Hi Ray,

    Gotcha'. You say you've ruled out transformer saturation; have you seen this paper?

    www.ti.com.cn/.../sloa133.pdf

    Mike Tripoli
  • 0 in reply to Mike Tripoli
    Prodigy 210 points
    Mike,

    No, I had not seen that paper. It makes for interesting reading and does indicate a class D amp is much more sensitive to transformer saturation than a class AB amp.

    I ran additional tests based on the paper. I drove the transformer to 70.7 volts out (into 330 ohm load) at 1 kHz. Then I started reducing the frequency on the signal generator. The generator is a HP 3314A and has a constant output level. When I got to 142 Hz,the amp started going into shutdown. This probably indicates core saturation. I then restarted the test at 1 kHz and went back down in frequency. The transformer was putting out 66 volts when the amp went into shutdown. The frequency was at 143 Hz at that point.

    The transformer is designed for 15 watts at 300-3000 Hz. By saturating at 142 Hz, that agrees with what the TI paper says about transformer design. I think that by feeding audio in from an FM radio source or microphone, frequencies lower than 142 Hz are being introduced into the amplifier. It also means that a full range class D PA amplifier (50 Hz-15 kHz) would require a much larger output transformer than a class AB one. The class AB amp just generates distortion on transformer saturation peaks, whereas the class D one shuts down.

    I have been testing with just a microphone and mixer console. I have circuit boards on order for the actual preamp that will be used with the TDA3118 board. The preamp has a high pass and low pass filter that is supposed to limit the frequency range to 300 Hz-8 kHz. When the boards come in and one is assembled I will run additional tests. If I still have shutdown problems I will try changing the high pass filter components to 400 Hz.

    Also, maybe Peter will have additional input about the problem.
  • 0 in reply to Ray Buck
    Genius 3295 points
    Hi Ray,

    I think you're onto a solution. It warms my heart to see someone actually testing their design. I've done a great deal of design with "sound chips" (these are "spec'd at the "telephone bandwidth" you mention); rarely, if ever, do people need the bandwidth they "think" they need. This is obviously a "PA" type of system; no one is going to notice the bandwidth limiting you propose.

    If I can help with anything, drop me a note.

    Mike Tripoli
  • 0 in reply to Ray Buck
    Prodigy 210 points
    Peter, I have bread boarded and tested a high pass filter as a means to prevent the TPA3118 from going into shutdown mode. I have found that I need to have a fourth order filter in order to prevent going into shutdown. A second order filter does work but I have to set the cutoff frequency to 500 Hz to get the necessary stopband performance at 100 Hz. This does affect the sound at the low end more than I would like. The fourth order filter has a sharp enough drop that the 100 Hz performance is acceptable and low end response is not adversely affected. Unfortunately the PCBs that are on order only have a second order high pass filter section implemented.

    Do you think I would be successful in stopping shutdown if I used the PLIMIT pin to reduce the output voltage? Currently PLIMIT is tied to GVDD. But it would be fairly easy to cut that trace and install a resistor. If I understand the data sheet correctly, limiting output voltage swing does not introduce clipping or cause the chip to enter shutdown. Is this in fact correct?
  • 0 in reply to Ray Buck
    Genius 3295 points
    Hi Ray,

    If your budget can afford it for the design, have you considered a switched cap filter such as the TLC04?

    www.ti.com/.../slas021a.pdf

    Mike Tripoli
  • 0 in reply to Ray Buck
    TI__Expert 5000 points
    Hi, Ray
    I am not sure what is your transformer's spice model. TPA3118D2 has DC protection/Over temperature protection/Over current protection can cause it shutdown. You can use a oscilloscope to check if there is a large output DC offset when the fault pin of TPA3118D2 change to low.
  • 0 in reply to Mike Tripoli
    Prodigy 210 points
    Mike,

    Thanks for the suggestion. But the fourth order high pass filter will do the job at a lower cost.

    Ray
  • 0 in reply to Peter Cao
    Prodigy 210 points
    Peter,

    I haven't tried to measure the DC offset but I assume that is the problem. The high pass filter solves the problem. My only other question is whether it could be solved by limiting the output voltage. If so, that would be a much simpler solution. I only need 11.7 volts out to drive the transformer to the required 15 watt output. If output voltage limiting doesn't introduce distortion that would work.

    Ray
  • 0 in reply to Ray Buck
    Prodigy 210 points
    Peter,

    I ran an additional test using the PLIMIT pin to limit output voltage. I used 43K and 53.6K ohm resistors to limit output. I measure 3.2 volts on the PLIMIT pin. It does in fact limit the output of the TPA3118. However, the amplifier still goes into shutdown, just at a lower output level. So I believe the fourth order high pass filter is the only solution that will work. If you have any other ideas, please let me know. Otherwise, I will create a new PCB design using the fourth order filter.

    Ray
  • 0 in reply to Ray Buck
    TI__Expert 5000 points
    Hi, Ray
    TPA3118D2 can config to 1SPW mode, in this mode, DC detection is disabled.
    If you can make sure the current DC output does not destroy your speaker, you can have a try.
  • 0 in reply to Peter Cao
    TI__Expert 5000 points
    In order to enter into the 1SPW mode, you can only change PIN1 (MODESEL)'s status.
  • 0 in reply to Peter Cao
    Prodigy 210 points
    Peter,

    I will test the 1SPW mode. The transformer primary (8 ohms) is wound with #20 gauge wire so there is no danger of it being destroyed. I will determine whether to use 1SPW mode or to use BD mode with the fourth order filter.

    You can consider this issue closed as I will use one of the two as my solution.

    Ray