• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 30 subscribers
  • Views 741 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.

TPA3121D2 class D amp occasional No sound on power up for several seconds problem

Vic Elias
Prodigy 110 points
Other Parts Discussed in Thread: TPA3121D2

We are having trouble with some of our radios which use a TPA3121D2.

On some of our units, there is a delay before we hear sound out the output on power up.  This delay can range from 4 seconds to 40 seconds.  We see audio going into the chip and don't hear anything out during this time.  It also seems to be somewhat dependent on the VCC voltage.  On the problem units, the delay usually doesn't occur at 11 volts Vcc, but almost always happens at 14 volts Vcc.  This delay usually does not occur after the product has been powered cycled.  I have to wait 5 minutes with the power off to get it to happen again. 

We are controlling the -SD shutdown input via a microcontroller at power on.  I'm not sure if there is some minimum length or any restrictions on that input at powerup (I know there is a 'application report' on the TPS3123D2 regarding this, but it says the problem was fixed in the 3121D2.)  The mute signal is also microcontroller controlled, but is always held low.

I read that the BYPASS capacitor affects the startup time of the chip, but there is no specifics about this in the datasheet.  We are using the standard 1 uf cap for that.  We have pretty much copied the standard application circuit for this chip.

We are having our customer send us some of the problem radio units with this chip in it that they have discovered for further testing.

Does anyone have any ideas or suggestions?

Thanks.

  • 0
    TI__Guru**** 189401 points

    Hi, Vic,

    Sorry to hear of your troubles. 

    I would look in two areas:

    1. Output filter - is it drawing too much current at start-up?

    2. Charge on the BYPASS cap. I believe there is a comparator that won't let the part start until the BYPASS cap is discharged. So, if you cycle the power rapidly, I could see what you're describing happening.

    -d2

  • 0 in reply to Don Dapkus
    Prodigy 110 points

    Here is the schematic we are using (except the -SD pin now goes to a microcontroller output).

    We set the -SD pin high about 600 mS after power up.  I can watch the charging on the BYPASS cap go to about 1.75 v which makes sense (VCC/8), our VCC is about 14 volts.  It charges in about 324 mS.  One odd thing I noticed on all our boards is that the output is always muted when I have a x1 scope probe on the bypass cap until I take it off, a  x10 probe works ok.  I can't see any difference in the waveforms between boards that are ok and boards that have the problem.  I was thinking that maybe the bypass voltage just barely reaches the limit to turn the chip on, on the problem boards, its hard to measure much difference.

    I thought maybe the output is shut down because of the output filter or something is shorted but I think that is ok.

    I do notice there always seems to be a small voltage 140 mv or so on the bypass pin even when there is no power on the board.  The strange part of this problem is that once a problem board is repowered, it works fine until it is disconnected for 5 or 10 minutes.

    I think it might be a good idea that we get a EVM board for this chip, to make comparisons with, they should have gotten one in the first place. 

    Anyway are there any suggestions, or something that doesn't look right about our circuit?

  • 0 in reply to Vic Elias
    Prodigy 110 points

    I am thinking our problems with our units are related to the 1uF Bypass capacitor.  I think this must be a critical component.  We are using a 1 uf ceramic chip capacitor X5R dielectric +- 10 percent (Kemet C0603C105K4PACTU)

    I was reading in posts about other chips in this family, suggestions that there is flux or bad solder connections that can screw things up and to clean the board etc.

    One of the malfunctioning units we had had the 10-40 second power up problem, I soldered on a wire to the cap to make measurements and the unit seems to work fine now.  Another problem unit I noticed there would be no sound on the output whenever I put a DMM on that capacitor to measure where its leveling off.  When I took the DMM probe off that capacitor the sound came back.  I measured it charging to 1.61 volts and slowly going to 1.68 volts.  VCC was 14.08 volts.  So 14.08 / 8 = 1.76 volts.  It seems it never quite makes it to the 1.76 volts.  The meter seems to affect the charging so it never gets there, but on working units the measuring it works fine and doesn't mute the output.

    I was going to replace that capacitor on that problem board but the 3121 blew up, smoke etc when I was measuring VCC, don't know why.

    So my question is how sensitive are things to this bypass output?  Could it be that our capacitors are crappy or on the edge?  Its hard to isolate since its only 1 in 5 boards or so. 

    We will try replacing that capacitor in problem units and see what happens.

  • 0 in reply to Vic Elias
    TI__Guru**** 189401 points

    Hi, Vic,

    Great!!! I had actually forgotten about this issue. It's a good thing you used e2e to find it!

    We had a big customer who was using this device and having similar problems. They had used an 0402-size bypass cap (1uF) and the leakage/IR of the cap was too great. We had them to change to an 0603 or 0805 cap which had less leakage (IR) and probably was also less sensitive to flux, etc, causing problems.

    I assume that you are using a no-clean soldering system. First thing I'd do is scrub the area around the cap to remove junk and see if it fixes slow boards.

    -d2

  • 0 in reply to Don Dapkus
    Prodigy 110 points

    Thanks for the info.

    I will figure out if that's the problem or not, we are now having trouble finding a board that is having the problem which I guess is a good thing.  One of our problem boards blew up and the other one is the one that mysteriously started working again.  We are going through our supply of 100 or so boards to find one with the problem and our customer is supposed to be sending back a couple that have it.  The capacitor is a 0603 but its still pretty tiny and difficult to work with.  They normally use larger packages for other products here.  I don't know much about the clean or no clean process they use its all subcontracted out, but I can find out. 

    I think the plan will be if we get another bad board to clean it, see if it fixes it, reflow the solder with soldering iron, see if that fixes it, then replace it and see if that fixes it.

    I will give an update in a week or two to see if that's it or not.

     

     

  • 0 in reply to Vic Elias
    Prodigy 110 points

     

    After getting a couple of boards back that had this problem from the customer, we cleaned them by hand and problem went away.  Sounds like it is just sensitivity to contamination on that power-up bypass capacitor.  I think the board layout that was used on our unit probably wasn't the best either (there is a ground plane fill under the cap and all around the traces to that cap, maybe this made it more sensitive to contamination?  Also its connection is thru vias to the other side of the board, not right next to the pin like the EVM).  We are using 0603 size, but probably should have used 08 size like our other products use, though things were tight with space so using a physically larger cap might not have fit.

    Anyway, looks like the problem is solved !