Diagnose and rectify high current draw and implement loud headphone amp

Discussion in 'Power Electronics' started by freeflyer, Sep 9, 2016.

  1. freeflyer

    Thread Starter New Member

    Sep 9, 2016
    12
    1
    My project uses a 3.7V 850mAh battery which is rapidly being discharged both during standby (within hours) and especially when in use (within 10 minutes) !!!

    I am at a loss as to why this is happening and looking for help to diagnose and rectify the issue.

    I designed the schematic, PCB (4 layer) and the embedded software.

    The main components on my PCB are:

    • Microchip 16-bit micro (dsPIC33EP256MU806)
    • Bluegiga bluetooth module (WT32i)
    • Texas Instruments CODEC (TLV320AIC32)
    • Texas Instruments DC-DC converter (TPS65136)
    • Texas Instruments headphone amplifier (TPA6120A2)
    • Spansion 32 Mbit external flash (S25FL032)
    • Bosch pressure sensor (BMP280)
    • Maxim battery fuel gauge (DS2745U+) - I have not implemented the I2C software for this device yet

    The project receives audio via the bluetooth module and which is then sent to the headphone amplifier.

    This is all handled by the dsPIC using the digital audio format I2S, which controls the playback of the bluetooth audio and also plays back 16-bit PCM audio stored in the external flash. The pressure sensor is used to measure altitude.

    The schematic is attached.

    The CODEC has an internal amplifier but I dont think this will be powerful enough.

    The project will be used in a noisy environment, so this is the reason why I have used the external amplifier (TPA6120A2) which uses the DC-DC converter (TPS65136) to supply a +/-5V supply.

    I also bought a pair of Tork Xpro helmet speakers...

    http://www.torkworld.com/helmet-speakers/xpro-helmet-speaker.html

    The specifications of these speakers is 500mw (it doesnt state whether this is peak or rms), 32 ohm.

    But during testing, the battery was completely discharged within 10 minutes of use.

    I don't know whether the battery capacity or voltage is suitable and the sound from the speakers is not very loud.

    So I need to find out:

    1. Why the battery is being drained so quickly
    2. Why the speakers are not very loud
    I don't know whether there is an issue with my design or with the assembly.

    I have limited understanding and experience when it comes to batteries and audio amplifiers, so was hoping someone could help.

    I put the dsPIC in sleep mode when the switch is used to turn everything off and also disable the bluetooth module and the DC-DC converter.

    The battery had been on charge overnight, so I am at a loss as to why it drained so quickly.
     
  2. wayneh

    Expert

    Sep 9, 2010
    12,126
    3,048
    Well there are a couple of approaches. One is a paper analysis of the current (and power) consumption of the various components compared to the power capacity. Your battery is rated to provide 3.7V • 0.85A = 3W for one hour. In reality it can't do that and I think a 20 hour discharge rate is commonly used to define the amp-hour capacity. So it would be more like 3.7V • 0.0425A = 0.16W for 20h. Subtract from that the losses in the converter (probably 10% or so) and the power consumption by the integrated circuits and I think you'll find that your battery is too small.

    The other approach is to get out your multimeter and measure where the battery current is going, compared to what you would expect from the schematic. You may have something wrong and are wasting a lot of juice.
     
    absf, ronv, freeflyer and 1 other person like this.
  3. freeflyer

    Thread Starter New Member

    Sep 9, 2016
    12
    1
    Thanks Wayneh

    I have taken some preliminary current measurements:
    • ~180mA when the device is initially powered up
      • dsPIC running at 60MHz
      • Bluetooth module enabled and out of reset - no bluetooth connection
      • CODEC out of reset
      • DC-DC converter enabled
    • ~190mA when the device is powered up and bluetooth connection is made
      • dsPIC running at 60MHz
      • Bluetooth module enabled and out of reset - bluetooth connection but no audio playing
      • CODEC out of reset
      • DC-DC converter enabled
    • ~220mA when the device is powered up, bluetooth connection is made and audio is played
      • dsPIC running at 60MHz
      • Bluetooth module enabled and out of reset - bluetooth connection and audio playing
      • CODEC out of reset
      • DC-DC converter enabled
    Once again, the battery lasted minutes, the voltage dropped too low and everything reset and powered off.

    Does the current consumption sound about right for this circuit and is the 3.7V 850mAh battery capacity suitable (it seems not but I dont understand why).

    This is what confuses me, all these ratings and specifications are not really valid because in reality you never get anywhere close to them.

    So an 850mAh battery won't actually give you 850mA for an hour !? In this case, the battery could only supply ~220mA for minutes before loosing its charge !!!

    The same goes for amplifiers and speakers. e.g. an amplifier rated at 100mW rms wont actually give you 100mW rms and a speaker rated at 500mW rms wont actually give you 500mW rms !? And there are so many other variables that affect these ratings.
     
  4. markdem

    Member

    Jul 31, 2013
    75
    38
    Just throwing it out there. Did you get this battery from eBay?
    Before doing anything else, connect a load (high power led works well) and monitor the current and voltage. I would not be surprised if the battery is not what you think it is, or is not been charged right.

    Have fun.
     
    absf likes this.
  5. freeflyer

    Thread Starter New Member

    Sep 9, 2016
    12
    1
    No, the battery came from a development kit (DKWT32i) made by Bluegiga who manufacture the bluetooth module...

    http://www.silabs.com/products/wire...es/wt32i-bluetooth-audio-development-kit.aspx

    I bought the development kit from Digi Key.

    The part number on the battery is LP603048, I have attached the only datasheet I could find.

    The first time when I realised the battery was discharged I measured 0V across the terminals, but I had never charged the battery since I bought the kit in April this year as I had been using USB to supply the power.

    The bluetooth module (WT32i) has a built in charger, so I left the battery on charge over night using my board, not the development kit.

    I am a bit concerned that the WT32i datasheet states "Do not short circuit the battery or discharge below 1.5 V", does this mean the battery could now be damaged ? It still doesnt explain why initially after only minutes of use it completely discharged the battery ??

    I am now using the DKWT32i development kit to charge the battery, in case there is an issue with my board design/assembly.
     
  6. markdem

    Member

    Jul 31, 2013
    75
    38
    It might. I don't know too much about that type of battery but it could be damaged or faulty from day one.

    Good idea using the dev board to charge it, it will isolate your charging circuit.
    You need to work out if the issue is with the battery or you device so I would still do a load test on it so you can see how long the battery will last with a simple load.
     
  7. markdem

    Member

    Jul 31, 2013
    75
    38
    I just looked at the datasheet of the battery. Looks like it has protection against over and under voltage. It will not allow it to go under about 3 volts, so I don't think it has been damaged.

    My point still stands, the battery might be faulty and needs to be tested.
     
    wayneh and freeflyer like this.
  8. freeflyer

    Thread Starter New Member

    Sep 9, 2016
    12
    1
    I charged the battery using the development kit (DKWT32i) and then plugged the battery into my PCB.

    With audio playing at full volume, the battery lasted just under 15 min and the average current (measured using a DC multimeter) was ~220mA.

    A scope trace of the battery voltage is shown below. It only shows the last few minutes before the voltage dropped to 0V.

    The voltage can be seen to decay over just these few minutes. I assume I can rule out the charger having tried charging the battery using the development kit ? So does this mean there is an issue with the battery or my PCB design/assembly ?

    The battery shouldnt discharge this quick with an average current of 220mA should it ?

    What current load should I put on the battery to test it ? 10mA? 100mA ? 200mA ? 500mA ?

    [​IMG]
     
  9. wayneh

    Expert

    Sep 9, 2010
    12,126
    3,048
    That's the self-protection circuit kicking in. It's protecting the battery from excessive discharge.

    It may be a little overly protective, but anyway I think this is proving that you need more battery capacity.

    Or you need to reduce current consumption. 220mA seems like a lot for a headphone amp, but you've got a lot going on. I'm too lazy to analyze each IC for you. What do the data sheets say for those parts?
     
    freeflyer likes this.
  10. ci139

    Member

    Jul 11, 2016
    341
    38
    i guess these things use a "lot of power" compared to the rest of your cfg.
    ---------------------
    or perhaps also this one depending on it's output power
    ICC Supply current (each channel) mA
    VCC =±5V 13mA
    VCC=±15V 15mA
    IO Output current (per channel)
    VCC= ±5V to ±15V 700 mA
    total 1430mA ^-1 x (850mAh·50%) = 0.297h = 17 minutes 50 seconds . . . ► System Down by the power draw by TPA6120A2 only
    . . .
    you want to limit your headphone amplifier not to max.gain(<> max.undistorted.volume) - but to max amplification for rms. of std. source - it is not expressed using the std. terminology
     
    Last edited: Sep 10, 2016
    freeflyer likes this.
  11. freeflyer

    Thread Starter New Member

    Sep 9, 2016
    12
    1
    I have extracted the specified supply currents from the datasheets for the ICs in the following spreadsheet. There are separate sheets for each IC and a summary page....

    https://dl.dropboxusercontent.com/u/38813887/PowerConsumption.xlsx

    I didnt get very far summarising the current draw as I dont know how to !

    The current draw for the dsPIC will depend on the modules enabled.... the current draw for the regulators will depend on the output current.... the current draw for the headphone amp will depend on the load and the volume..... so it seems difficult to even put a rough figure for current draw on these devices.

    I would prefer not to have to use a larger battery, the one I am using only just fits into my enclosure and ideally I would like to make this device smaller and more discrete !

    Im not sure how I would reduce power consumption, the headphone amp and helmet speakers aren't quite loud enough as it is.
     
  12. freeflyer

    Thread Starter New Member

    Sep 9, 2016
    12
    1
    Thanks for this, Im trying to understand how the output current per channel can be 700mA if the supply current per channel is 13mA (the DC-DC converter is +/-5V) ? Is it like a transformer, where the secondary winding can have a higher current/lower voltage than the primary winding which has a lower current/higher voltage and vice versa ?

    So if the supply current per channel is 13mA +/-5V and the output current per channel is 700mA, then the output voltage per channel will be much lower than +/-5V ?

    You came up with the figure 1430mA (15mA supply per channel + 700mA output per channel), but I am only seeing an average DC current of ~220mA on the multimeter, not 1430mA.

    Sorry but I have no idea what you mean when you say "you want to limit your headphone amplifier not to max.gain(<> max.undistorted.volume) - but to max amplification for rms. of std. source - it is not expressed using the std. terminology"
     
  13. markdem

    Member

    Jul 31, 2013
    75
    38
    So, did you test the battery?

    Say you are only getting 10 min of driving in you car after filling the tank. After 10 min, the tank is empty. Do you start by pulling the engine apart, or do you check the fuel tank for leaks? Especially if you are only seen 220mL\h...

    It does not matter how much load you put on the battery for testing. 200mA sounds good as in theory it would take about 4 hours to discharge.
     
    Last edited: Sep 10, 2016
    freeflyer likes this.
  14. freeflyer

    Thread Starter New Member

    Sep 9, 2016
    12
    1
    I have tested the battery today with an 8R2, 7W resistor. The battery had been on charge for half a day. I monitored the current with a multimeter and it started at ~0.4A and went down to about ~0.37A. After ~ 7 minutes the battery protection turned the battery off. So it sounds like my battery is faulty then ? I will order a similar battery (RS/Farnell dont sell the same battery) and try again.

    It would still be nice if I can reduce the current consumption from 220mA if possible.

    I measured the current when the device is in standby....

    ~13mA to ~15mA when the device is in standby:
    • dsPIC in sleep mode
    • Bluetooth module disabled and in reset
    • CODEC in reset
    • DC-DC converter disabled
    This seems very high, so there is obviously an issue with my design or assembly. The trouble is, how do I identify where the issue is ? The only way I can think of is to cut tracks on the board to isolate power from each device, one by one.
     
  15. wayneh

    Expert

    Sep 9, 2010
    12,126
    3,048
    That may be higher than you want, but it's not the problem. ;)

    Your voltage regulator is probably burning a little current. They typically have some minimum current they need to maintain regulation. Also, are any of the LEDs lit when you measure the 15mA? The LEDs alone could use that much.
     
    freeflyer likes this.
  16. freeflyer

    Thread Starter New Member

    Sep 9, 2016
    12
    1
    Nope, all the LEDs are off. The only active devices should be the main 3.3V regulator, the dsPIC (which is in sleep mode) and the pressure sensor (in future I will put this into sleep mode too). So I expected to see uA of current when in standby, not the 13-15mA I am seeing.

    In fact the LEDs are never on permanently, I only flash the power LED intermittently when the device is on to preserve the battery. The charger LED only flashes intermittently when charging (this is controlled by the blue tooth module). The other 5 LEDs (used for the battery fuel gauge) will only light temporarily, again to preserve the battery.

    Im surprised the battery is damaged/faulty, it came with the development kit so it should be ok ?
     
  17. dannyf

    Well-Known Member

    Sep 13, 2015
    1,811
    362
    try to cut it into major functional blocks and measure current consumption for each of them.

    I would start with the power supply and the analog amplifier. 6120 is an extremely fast (and capable) amp and it wouldn't surprise me if it is oscillating into the Mhz range.
     
    freeflyer likes this.
  18. freeflyer

    Thread Starter New Member

    Sep 9, 2016
    12
    1
    This is what I was afraid off, it means cutting tracks on the PCB in order to isolate each functional block :-/

    When you say the 'power supply', do you mean the DC-DC converter ?

    Also, how do I tell if the amp is oscillating into the MHz range and if it is, what can I do to rectify it ?
     
  19. freeflyer

    Thread Starter New Member

    Sep 9, 2016
    12
    1
    Another strange thing is that the net "+3V3_WT32i" is at 2.5V when the TPS79933 is not enabled.... it should be 0V and I have confirmed this by testing the development kit (DKW32i). I checked that the 'ENABLE' pin of the TPS79933 behaves correctly, so I dont know why the 'OUT' pin is at 2.5V when the regulator is disabled.


    [​IMG]

    I also measured the current consumption of the the development kit (DKW32i) when it is powered up.

    The current ranged from 6mA to 9mA. The development kit uses the same CODEC and regulators etc, the only devices it doesn't have are the DC-DC converter, dsPIC, pressure sensor, battery fuel gauge and headphone amplifier.

    I measured the current consumption of my PCB with the DC-DC converter disabled (for a fairer comparison with the development kit) and the current was around 90mA !

    This suggests that the dsPIC/pressure sensor/battery fuel gauge (none of which are on the development kit) is consuming about 84mA.

    I had a nightmare trying to assemble this PCB because of the tiny leadless QFN packages used for the DC-DC converter and CODEC. As well as the leadless bluetooth module. I bought a stainless steel solder paste stencil and solder paste then used a laboratory hot plate (temperature controlled) to reflow these devices onto the board, following the solder reflow profile as best I could. I even used a thermocouple with my multimeter to confirm the temperature.

    I copied the schematic for the development kit (see attached file WT32i_Evalboard_schematic.pdf) but I dont know whether the high current consumption is down to my schematic/PCB design or the assembly itself.
     
  20. dannyf

    Well-Known Member

    Sep 13, 2015
    1,811
    362
    1. a scope;
    2. a rf probe. you can make one with a 1n4148 + capacitor + resistor;
    3. a zobel network on the output.
    ...
     
    freeflyer likes this.
Loading...