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.

UCC27524: UCC27524 range extender solution for TMS3705

Part Number: UCC27524
Other Parts Discussed in Thread: TMS3705

Hi,

I have a question in regarding to how to calculate the thermal dissipation of the range extender solution described in scba031.

It would aid me in deciding the heat sink solution for the UCC27524.

Specifically if Vcc=15V, with the default 3.3nF / 440uH coil (say coil has 10 Ohm ESR at 134.2 kHz) how would one determine:

1. The power dissipated in UCC27524?

I have read the power dissipation part in the datasheet of UCC27524, however it is for the application of driving a MOSFET where the majority of energy used for charging the gate capacitance is lost as heat. In the range extender solution there are also large amount of power become reactive and stored in the coil during each cycle.

Thanks,

Nick

  • Hi Nick,

    Today is a bank holiday here in the US, so my apps engineer will respond to your question tomorrow when our office is open.
  • Hi Nick,

    Thanks for asking about UCC27524. Im an apps eng that will help you out. Let me reach out to the designers that made this app note to hear what they have to say about this concern, let me update you asap.

    The antenna is a resistive, capacitive and inductive complex load. The datasheet specifies power consumption for a purely capacitive linear load. To find power dissipation of a complex nonlinear load we need to separate these components for superposition so we can plug them into linear equations.

    Power dissipation seen by the gate driver, driver losses, can be considered as a ratio of the internal output stage impedance with the power consumed by impedance in the charge path. With a complex load, a complex impedance model of the antenna would need to be obtained after which complex voltages and currents can be realized with simulation. After we have complex values for how the driver will behave when the antenna is switching from 15V at 134.2khz we can use linear equations.

    The max power dissipation of a capacitive load can be found from P=Q*V*Fsw (Q*V is the energy) where Q=CV for capacitor, the max power dissipation occurs when the peak current occurs during each switching cycle because the cap acts as a short circuit for high frequency and for the high current gate driver pulse the FET load Cgs looks like a short that the driver sees. Q=LI^2 for inductor which makes the power dissipation equation P=L*I^2*Fsw. Solving for this equation does not give a real answer because of the non linearity of the load, it seems that the peak current does not reach the 5A peak but a lower value determined by the complex impedance of the antenna.

    L=uH is larger than typical nF cap value -> this means for energy to charge the inductor and more energy that needs to be dissipated I = 5A peak which is more than the 1A peak from the antenna driver -> this will make the power dissipation 25times more however since theres an inductive load the switching voltage/current may overlap during turn off and slowly ramp during turn on which will complicate things further for switching losses.

    The antenna also acts as a small heatsink gate resistor due to the 10ohm ESR at frequency. This ESR will act like a resistor divider or heat divider and ratio the power dissipation or total energy to the other resistances in the path like the 4R7.

    Thanks,
  • Hi Jeffrey:

    Thanks for your answer.

    I am sorry but the answer was still a bit vague to me. 

    The complex model of antenna is just a coil with 440uH inductance and 10ohm resistance (ignore skin/proximity effect etc). So the impedance of the coil would be 10+i*2*pi*134.2k*440uH, which is paired with a resonant cap. Or basically a series LCR network.

    I understand it can be tricky given the various working mode of TMS3705 and the driving voltage is effectively square wave with higher freq components. To make life simple lets only consider the case of continuous charging.

    I would like to know in ball park how much power is being dissipated by the driver and thus PCB area for heatsink of a space confined application.  

    I do not mind complex algebra or calculus, if you cannot write the full answer here I do not mind reading tech document either.

    However please let me know the assumptions you have made in obtaining the final answer.

    Much appreciated,

    Nick

  • Hi Nick,

    I am still waiting for a reply from the designer and will update you asap.

    To update you on my response, From figure 13, Since the drive scheme puts the inputs alternating and not consistently 134.2khz I think we can consider this to be one driver in terms of power dissipation. We just need to consider how much current the driver is actually outputting.

    The inductance limits the current to ramp at a rate of dV/L, the current should reach only as high as the ramp times the duty cycle.The time constant L*R of the antenna is much larger than the period and even the R*C constant so the antenna is probably not fully discharged. Therefore the delta V the driver experiences is not large enough peak current to create excessive losses in the driver.

    A ballpark the power dissipation for a cap can be estimated by C * V^2 * Fsw assuming CV^2 = LI^2 we get L * I^2 * Fsw with L and Fsw fixed at 440uH and 134khz the only variable is how much current the driver will output. We can say the peak current is the difference in gate voltage divided by loop resistance. I_peak = dVdriver/Rdriver. If the inductor is charged then dVdriver is 5 at most so I_peak is about 300mA if R_driver is 15 ohms. This makes about 5W of power dissipation at 134.2khz (mostly dissipated by the antenna 10 ohms).

    You can check out an great app note on the Fundamentals of Gate Drivers section 2.7 on power losses (). Does this clear up some confusion?

     

  • Hi Jeffrey,

    Thanks for your response.

    1. "If the inductor is charged then dVdriver is 5 at most"

    what is this dV driver? how do you know it is 5V? Is this the VGS? 

    The VDS is 15V in this case.

    2. What is R_driver, how do you know it is 15Ohm? (which page in the data sheet of UCC27524?)

    3. How do you get a power dissipation of 5W? According to you (mostly dissipated by the antenna 10 ohms), so what is the power dissipated in the UCC27524 ?

    Thanks,

    Nick

  • Hi Nick,

    I was thinking VCC=15v was the driver voltage applied to the gate, (Vdriver). If the inductor part of the antenna doesnt fully discharge when the FET is off then the antenna will switch from 10v to 15v so there will be only 5v swing on antenna every cycle, this will limit the power dissipation.

    What is the driver supply voltage?

    At room temp the max power dissipation of UCC27524 is about 0.88W.

    If we consider only the resistive portion of the antenna spec 10 ohms

    V^2/R average power says that the antenna dissipates about 23W (and if the antenna voltage goes back to 0V then the transducer dissipates this power every cycle)

    According to the ratio of resistances on the gate path 66% is from the 10ohms 33% is from the 4.7ohms will dissipate in the path respectively.

    This leaves only about 1% of total resistance in the gate path coming from the internal output structure of the driver (0.5 ROL).

    1% of 23W is 0.23W which is less than the max power dissipation

    Can you sim or bench test this circuit to see the gate current waveform?

    this will allow us to know how much current the driver is sinking and sourcing and therefore also dissipating.

    We can take this current then multiply by Roh/Rol then we can know the real power dissipated internally every cycle.

    Thanks,
  • Hi Jeff:

    From my understanding:
    If we look at the figure 2 in scba031, the power supply to the UCC27524 is 12V, or 15V in this case.
    The gate voltage is 5Vpp square wave (0V or 5V biased at 2.5V) which is outputted by the TMS3705.

    The two gate switches on alternatively at 134.2kHz.
    When channel A is on or shorted to Vcc=15V, the UCC27524 charges L1. channel B is shorted to ground.
    Therefore the loop has L, C, two 4R7 resistors and ROL + ROH.
    Channel A source current and channel B sink current.


    When channel B is on or shorted to Vcc=15V, the UCC27524 charges Cres. Channel A is shorted to ground.
    Therefore the loop has L, C, two 4R7 resistors and ROL+ROH.
    Channel B source current and Channel A sink current.

    Assuming the LC is at full resonance and fully charged (which it will after few ms), the impedance of L and C cancel each other out.
    For each half cycle, the voltage across is 15V, the max resistance on the path is 10+2*4R7+1+7.5, the current is 0.54A.
    The max power dissipated in the driver is 0.54^2 * (1+7.5) =2.48W.

    Some key assumption were made such as the fact the driver output a square wave of 134.2kHz , which has higher frequency components that were not accounted for in the calculation, into a series LC network with resonant frequency of 134.2k.

    Another assumption made was the fact ROL and ROH were the only resistance or ohmic loss considered in the UCC27524. Switching loss/gate drive loss were not considered. Do you think this is OK for my application?
  • Hi Nick,

    Thanks for clearing up my understanding on how this topology works.
    The resonant or switching frequency is 134.2k? I was thinking the Fsw.
    This average current value seems ok however the only inconsistency I see is that Roh does not equal 7.5 ohms. From page 14 of the UCC27524 datasheet says Roh is dynamically equal to 1.5*ROL=1.5 ohms. This makes the current 680mA and the power dissipated in the driver about 1.2W which is about 300mW too much. I recommend increasing the 4R7 gate resistors. Increasing to 6R and beyond will decrease the drive current to .375 enough to keep the junction temp below the rec op max.

    Let me know if this does not help your issue.
    Thanks,
  • Hi Jeff:

    How do you determine the maximum power can be dissipated in the driver?

    Is that written somewhere in the data sheet?

    Thanks,

    Nick

  • Hi Nick,

    thanks for the update,

    the datasheet has this info. However its only for capacitive load.
    Check out section 9.2.2.5 and 11.3

    Also check out the first section of the IC Package Thermal Metrics (SPRA953). It explains how to find the Rja of your own system so you can calculate the max allowed power dissipation of the driver on the pcb.

    Thanks,