LTC4125 strange behaviour

I'm using the LTC4125 power transmitter for a wireless charger in conjuction with LTC4120 device.
In attachment you can see the schematic of TX unit.
I notice a strange behaviour.
With no RX device coupled, When I apply the power source (5V), the LTC4125 absorbs the maximum current determined by R1,R2 and R6 (1.8A).
I have set R6 to 470K to reduce current to 380mA for further testings.
In picture 1 and 2 you can see voltage on both sides of inductor L1.
Frequency is about 106 KHz.
Every 4 seconds voltage changes as in picture 3 and 4, for a very short time and current drops to a few mAs.
I think the behavior should be exactly the opposite: normally the LTC4125 should absorbs a few mAs. Current should raise when he has been coupled with the RX inductor.
L1 is a 8uH inductor (760308102207 from Wurth).
I've tried different types of inductor for L1 (760308100110, 760308101103, 760308101105) with no success.
Some one can help me debugging this circuit?

attachments.zip
  • 0
    •  Analog Employees 
    on Jun 6, 2018 12:44 AM over 2 years ago

    I am working on a number of customer issues right now.  I will respond to this question as soon as I am able.

  • 0
    •  Analog Employees 
    on Jun 19, 2018 10:12 PM over 2 years ago

    I've set up the DC2330A in the lab, and I recorded the waveforms I observe during power search.  You can see that the higher-input-current portions of the waveform will persist for about 15ms before the search exits.  Between searches, the current consumption of the circuit is low.  I cannot say for sure why you seem to see a long period of high current followed by a short period of low current.  Be sure the connections and component values are correct.  Also, during your testing, ensure no conductive materials are in the vicinity of the coil.  I see no problems with the schematic, so if you wish, I can review your layout.  For the DC2330A, for example, we had to remove etch in the vicinity of the coil-wire attachment region to prevent eddy-current losses in that part of the board.  Once the power-search waveforms look similar to the ones posted here, you should be able to begin 'tuning' the detection circuits for proper power-search exit criteria.  (More follows.)

    DC2330A Operation over Showing Two Transmit Power Searches with No Receiver Present (Blue trace is the voltage at the node connecting the resonant coil to the resonant capacitor.)

     

    Zoomed-in View of the Highest-Current Coil Drive during Power Search

    It is apparent from your component selection that you used the DC2330A design as a starting point which is good.  The main difference is the characteristic impedance of the tank circuit which is less in your circuit by a factor 3.  As a result, the tank-circuit voltages will be less by a factor of ~3, and the tank-circuit currents will be greater by a factor of ~3.  Since the sensing on the DFH and FB pins is voltage-based, a good place to start in adjusting the circuit would be to increase the resistor-divider ratio of R7, R8 by a factor of 3.

  • Dear WATaylor,

    many thaks for you answer.
    I'm really confused about this circuit.

    1) I've checked all the connections and components value. It looks like ok.
    2) How you can see, my PCB layout (PCB picture) has copper bands along the inductor in the inner layers VCC and GND (green). I don't think this is a problem, anyway I've cutted out the PCB around inductor eliminating NTC sensor (see Top, Bottom and Cut pictures).
    3) I've verified the Frequency Threshold function (FTH pin). Approaching a metal object near the inductor, the frequency raises and when it's near 119KHz, current drops down and STAT pin goes high. This agrees with R4 and R5 value as stated in datasheet pag.19. So I think there is no problem with foreign metal objects.
    4) In your first picture I can see the waveform is composed by: a small signal for about 3.7secs, an 8 steps ramp for about 150ms and again a small signal. In my circuit waveform (see Wave1 and Wave2 pictures) is composed by: a small signal for about 140ms, a 6 step ramp for about 100ms, and a continuos signal with the same amplitude of sixth step. I think this means that optimum power search stops for a valid exit condition, although it should not. In effect STAT pin goes low. Probably the exit condition is meet because VImon exceeds 0.8V. I've tried different values for R2 and R6 to increase current limit (now about 500 mA) but the behaviour remains the same.
    5) I've tried to increase the resistor-divider ratio of R7, R8 by a factor of 3 (R8=1K8) but nothing changes. Inductor signal amplitude remains the same.
    6) A very strange thing. If I touch with a finger C6 pole, connected to CTS pin, with +IN signal (IN pin), waveform changes (see Wave3 and Wave4 pictures): zero signal for about 4 secs, a six steps ramp for about 100ms and again zero signal. I've tried changing values of C6 and C5 varying T1, T2 e T3 (see datasheet pag.20) but the behaviour remains the same.
    7) I've tried to change L1 value to 24uH, reducing C4+C11+C12 to 100nF (resonant frequency 103KHz). L1 signal amplitude is much higher but the behaviour remains the same.

    I can't understand how to calibrate this circuit to works properly.

    PCB Layout

    Top

    Bottom

    Cut

    Wave1

    Wave2

    Wave3

    Wave4

  • 0
    •  Analog Employees 
    on Aug 2, 2018 4:05 PM over 2 years ago
    This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.

    Thank you,
    EZ Admin
  • 0
    •  Analog Employees 
    on Apr 12, 2019 1:59 PM over 1 year ago

    Hi caryone, I can see that your R7 anf R8 sets the voltage limit of the resonant tank to be 100V. If that 100V limit is not triggered but the voltage on IMON reaches to 0.8V, you will see this situation of LTC4125 drawing high current for seconds. 

    LTC4125 has the search algorithm that will step up the duty cycle to look for a valid load. During this 'search' period', duty cycle will increase step by step and each step last for a period of time set by CTS capacitor. Once a exit condition is met (usually means a valid receiver is found) , this search will stop and duty cycle will stay at that certain level for a period of time set by CTD capacitor. If a fault condition is reached during the 'search period', duty cycle will be pulled down to almost zero, and LTC4125 should draw low current from the input. 

    One of the exit condition criteria is when IMON pin reaches to 0.8V. One of the fault condition criteria  is when IMON pin reaches 1.2V. During 'search period', IMON pin voltage (in proportion to input current) gradually goes up as search step increases. When IMON pin reaches 0.8V, exit condition is met. The duty cycle keeps at the level at the moment when exit condition is met for seconds. As duty cycle cannot go up anymore, IMON pin can never go up to 1.2V to trigger the fault condition. That is why you see high input current for long period of time and I believe the input current level translates to 0.8V-1.2V on IMON pin. 

    To prevent LTC4125 falsely enters this exit condition, we should properly set another fault condition indicator. Another of the fault condition criteria is when FB pin reaches VIN, which is 5V in most cases. We have to guarantee that FB reaches 5V before IMON pin reaches 0.8V when receiver is removed. In this way, before IMON pin reaches 0.8V to trigger the exit condition, FB has already triggered the fault condition to pull the duty cycle low. 

    In your case, you have high capacitance at your resonant tank. From your waveform I suggest you increase R6 to at least 50kΩ as I can see your resonant tank voltage is at around 15V when the IMON exit condition is met.

    Hope this will solve your issue!