adum4135 DESAT setup and calculate issue

 dear ADI FAE team 

could you please help these question as below, thanks. 

#1.      If around 40A current flows through Q2 Drain-Source, it looks that it detects DESAT.

          Is it correct?

          If so, the limitation of detecting voltage seems to be low.

          Is my understanding correct?

          If not, how much current does the DESAT turns on (works) for APTMC120A20CT1AG or how much voltage at MP point on the circuit?


#2      Would you please explain relation between DESAT detecting time and components like C39, D11, D7 and D15?

           How does DESAT detecting time by C39 C39, D11, D7 and D15?


#3.     Vss2 is different from HV_RTN. Would you please let me know why do you separate these voltages?


#4.     Do you have any simulation result of voltage at input voltage of DESAT ?    

           Data of current condition 20A and 10A at Q2 short will be helpful.


  • the schematic from AN-1536 and co-work with APTMC120AM20CT1AG (Rds(on)=17mohm).

    1, about the DESAT, how much is its short circuit protect circuit?   

        how to calculate?

    2. the R47, C39 it is connect to "GND-ISO-L", that is right?   

        I had survey other adum4135 design from Microchip & wolfspeed gate drive EVB.

        and Adum4135 datasheet suggest circuit, it should be connect to normal GND (as SOURCE-L).

    3. the tblank time is deping on R35, C39, R47 right?

        but the R35 / R47 will have a partial voltage about 8V, right?    maybe that will effect DESAT?

        I refer other adum4135 design, that only have R35+C39 or R47+C39, that don't have see R35+R47 design.

       it is a mistake for design?


  • 0
    •  Analog Employees 
    on Nov 4, 2020 8:57 PM 2 months ago

    Hello Seokor,

    1) Desat detection is not a direct measurement of the drain current of the power device. The Desat detection method is able to detect when the drain to source voltage combined with the voltage drops across the high voltage blocking diodes and desat resistor are above the threshold. In the schematic provided, there is also a Zener diode whose drop should also be added. The BZX84C6V2LT1G effectively lowers the threshold voltage by around 6.2 V. This means that when the voltage drop across R46, D7, D15, and the Drain to source of Q2 is around 2.8 V, the desat will trigger. I wasn't able to find the exact switch datasheet online, but I found the APTMC120TAM17CTPAG, which is probably similar. In this part, a drain current of around 40 A will result in a drain to source voltage of around 1.5 V, which is probably enough to trigger Desat in this particular circuit If you want to raise the current that the desat will trigger, you can change or remove the D11 Zener diode.

    2) C39 is the blanking capacitor. It is one of the components that can change the detection time, sometimes called the blanking time. A value of 220 pF is a good value for many IGBTs. For SiC, the detection time to comply with the desired short circuit withstand times usually results in a blanking capacitor in the 50-100 pF range. A larger C39 results in more noise immunity, but slower reaction times.

    D11, D7, and D15 create a voltage drop during the on times of the power device. D11, D7, and D15 effectively lower the perceived threshold voltage. D7 and D15 are required to block the high voltages from the drain of the device. D11 is optional, and it's Zener voltage can be changed to tune the detection trip point for the power device. Two high voltage diodes (D7 and D15) are often used to reduce the parasitic capacitance, as well as require lower voltage devices to save cost.

    3) VSS2 allows the ADuM4135 to deliver a negative voltage to the source of the power device. All that is required is to connect the GND2 pin to the source of the power device. As long as the voltage supplies feeding VDD2 and VSS2 are floating, the ADuM4135 will be offset by the voltage seen on VSS2 to GND2, and a negative gate voltage will be able to be delivered when the off state is requested from the ADuM4135. A negative gate voltage is sometimes desired to reduce the effects of the Miller injection. If the Miller injection is not too large, a unipolar setup could be used, and the application might not require a negative voltage. In that case, VSS2 can be connected directly to GND2.

    4) I don't have any simulations of the ADuM4135 alongside the APTMC120TAM17CTPAG. The ADuM4135 does have an LTSPICE model that does a nice job of simulating the part. You could create a simple simulation with the free LTSPICE circuit simulator.


  • 0
    •  Analog Employees 
    on Nov 4, 2020 9:11 PM 2 months ago in reply to seokor

    Hello Seokor,

    1) The calculation of the desat blanking time will be different if the option pullup resistor (R35) is used or not. There is an internal current source within the ADuM4135 that will feed the blanking capacitor with around 0.5 mA of current. The blanking capacitor will fill up linearly with I=C* dv/dt.

    For a 220 pF capacitor, if we don't use the optional external pullup resistor, the voltage on the Desat pin (and blanking capacitor) will reach 9 V in approximately:


    dt = C * dv / I = 220pF * 9V / 0.5 mA = 3.96 ms

    If we add the external pullup resistor, we can estimate the effect as being around V/R = 20/4.7kΩ = 4.2 mA. Adding this to the earlier current source:


    dt = C * dv / I = 220pF * 9V / (0.5 mA + 4.2 mA) = 421 ns

    The ADuM4135 has some comparator delay on the Desat pin, so we can guess that detection will occur in within around 150 ns of the above numbers. Again, this is using no zener diode in the above analysis. The Zener diode will reduce these times even more, assuming a fault under load case (FUL). For a high side fault (HSF) case, the timings will need the inital masking time of 300 ns be added to the above numbers.

    2) R47 and C39 are both referenced to GND2. This is also the connection that goes to the source of the power device being driven.

    3) Tblank depends heavily on C39 and R45. R47 is a bleeder resistor that doesn't really come into play. It guarantees the capacitor will be discharged when the part is unpowered, but for the most part can be ignored. R45 is option as well, as this adds more pull-up current for the desat circuit, but the internal current source within the ADuM4135 sufficient for many designs.


  • dear RSchnell 

    nice meet you, thank for your feedback. some question that is clear now.

    1. yes, about DESAT we use it only for short circuit protect, not for Id current detect.

        about 40A short circuit protect threshold setting, because we found the AN-1536 circuit that maybe have some mistake of design..

    APTMC120TAM17CTPAG Rds(on) is 17mohm, and that show the continuous current capability is 108A.

    I see the D11 is BZX84C6V2LT1G it is 6.2V, and D7, D15 Vf about 1.17V.

    So protect current calculate = (9V-6.2-1.17-1.17) / 17mohm (Rds(on) of APTMC120A20CT1AG)

    the current about 27A, so the short circuit protect threshold setting is too low?   we can not to work to Id=108A, right?

    that maybe to reduce the D11 clamp voltage to increase the threshold voltage setting.

    2. about D13, I see it be connected to "GND-ISO-L" (VSS2), same with C39, that should be connected to GND2, not VSS2.

         as my know, that is right?

  • 0
    •  Analog Employees 
    on Nov 16, 2020 10:37 PM 2 months ago in reply to seokor


    1. Yes, if you want to operate to a higher current, you can change the Zener diode to a lower voltage.

    2. D13 and C39 should be connected the GND2 pin on the ADuM4135. The Desat pin is internally referenced to the GND2 pin, and should not be pulled lower than GND2.