ADuM7234 H-Bridge for induction heating

Hi

I'm building an H-Bridge power stage switch to drive and RLC circuit which features an induction coil and capacitor tank coupled by an 10:1 toroidal transformer to the H-bridge output, working in the frequency interval of [60,90] KHz. To maintain the RLC locked on its resonance frequency I used an PPL that reads the frequency of the coil and locks on it.

I'm trying to use the ADuM7234 to drive the MOSFETS or IGBTs at 80V maximum and maximum current that can reach 120A, limited by the power source. I don't believe that so much current will be needed because I already built an H-Bridge based on the Linear LT1162 IC that when working drained 20A at 40V.

I intend to use this induction heater to heat 25um thick, 1cm2 copper foils until 1050ºC as fast as I can inside a reactor with controlled atmosphere.

So the problems and questions:

- The output of the PLL (CD4046) and the input in the BNC plug are a 12Vpp square wave, so I put the TLP2955 and TLP2958 opto-couplers to reference the signal to 5V and invert it also. Do you have other component suggestion that make the same result but with less delay and low noise sensitivity?

- The output of the opto-couplers (Vopp) decreases when I plug them to the ViA and ViB and I believe that it is an impedance mismatch since when I remove the 1K resistance (R3 and R4) the signal gets better. So, what is the input impedance of the ADuM7234 and does it varies when on and off?

- I'm still testing the assembly in the breadboard so I understand that elevated parasitic impedances can be present but the chips overheat too fast, can this be from a problem in the circuit topology?

- I'm using a computer power source to get the 5V and 12V that have a common ground, should I use two separated power sources to have isolated grounds? Should I use separated power sources, one for each ADuM7234?

- The Disable pin needs to be grounded for the drivers to work?

- The gate resistors are 3.3Ohm and not 1K, can I put just a ferrite instead of the gate resistor to reduce the initial turn on ripple?

- The LT1162 IC used a resistor between the upper MOSFETs gate and source, but from the Analog application note ( Inside iCoupler Technology: Driving an H Bridge with ADuM3220 Isolated Gate Drivers) it's not included, is there a reason for this, or should I include one close the loop during in the charging cycle of the bootstrap capacitors?

- When I increase the power (VPP), the gate signal gets distorted, what can be the reason for this?

- Is the ADuM7234 prepare to handle big gate capacitances?

I hope it was clear in my questions

Best regards

Jorge

attachments.zip
  • Hello Jorge,

    - The output of the PLL (CD4046) and the input in the BNC plug are a 12Vpp square wave, so I put the TLP2955 and TLP2958 opto-couplers to reference the signal to 5V and invert it also. Do you have other component suggestion that make the same result but with less delay and low noise sensitivity?

    It sounds like you are using the optocoupler as a level shifting inverter. Since the ADuM7234 itself is an isolator, I don't know if isolation is needed at the location you are using the opto-couplers. Could you use a standard inverter instead, and use a resistor divider to reduce the 12 V square wave to something around 5V? Or, could you do an open drain output with an N-FET with a 5 V supply?

     

    - The output of the opto-couplers (Vopp) decreases when I plug them to the ViA and ViB and I believe that it is an impedance mismatch since when I remove the 1K resistance (R3 and R4) the signal gets better. So, what is the input impedance of the ADuM7234 and does it varies when on and off?

    The ADuM7234 has CMOS inputs, and when it is powered up, should look like a high-impedance. When the ADuM7234 is not powered up, putting voltage on the input pins may parasitically power the ADuM7234 through internal protection diodes, so it is no recommended to feed the input pins when the unit is unpowered on VDD1.

     

    - I'm still testing the assembly in the breadboard so I understand that elevated parasitic impedances can be present but the chips overheat too fast, can this be from a problem in the circuit topology?

    If the chips are overheating, there might be a problem somewhere else. I don't think the parasitic inductances should add that much power dissipation to the system.

     

    - I'm using a computer power source to get the 5V and 12V that have a common ground, should I use two separated power sources to have isolated grounds? Should I use separated power sources, one for each ADuM7234?

    Each highside output of the ADuM7234 needs it's own power source. This could be as simple as a bootstrap capacitor if the circuit meets duty cycle and frequency requirements. The circuit note on the ADuM7234 uses a special configuration because the bus voltage is low. In your case, with a bus voltage of 80 V, we need to find a way to power the high sides independently of the bus voltage. Here is an app note about that subject:

    http://www.analog.com/media/en/technical-documentation/technical-articles/Powering-the-Isolated-Side-of-Your-Half-Bridge… 

     

    - The Disable pin needs to be grounded for the drivers to work?

     

    Yes, the Disable pin needs to be grounded. If left floating, it can float high, thereby shutting the output off.

    - The gate resistors are 3.3Ohm and not 1K, can I put just a ferrite instead of the gate resistor to reduce the initial turn on ripple?

     

    The external series gate resistor not only serves to dampen the turn on and turn off edges, but also serves to help spread the power dissipation of the gate drive outside of the gate driver. 3.3 Ω is a suitable external series gate resistor size. A ferrite bead can sometimes help, but it is still recommended to have an external series resistor in place.

     

    - The LT1162 IC used a resistor between the upper MOSFETs gate and source, but from the Analog application note ( Inside iCoupler Technology: Driving an H Bridge with ADuM3220 Isolated Gate Drivers) it's not included, is there a reason for this, or should I include one close the loop during in the charging cycle of the bootstrap capacitors?

     

    The resistors from gate to source sometimes found in gate driver schematics serves to help the system be at a known state when the gate driver is unpowered, or damaged. They are not always needed, but are a good idea most of the time. The values usually are in the 1kΩ-100kΩ range, where they won't affect the quiescent current drastically, but should something go wrong, or the gate driver become unpowered, the MOSFET gate will be discharged to the source voltage through this resistor.

     

    - When I increase the power (VPP), the gate signal gets distorted, what can be the reason for this?

     

    Do you have a waveform capture of this?

     

    - Is the ADuM7234 prepare to handle big gate capacitances?

    The power dissipated within the gate driver is directly proportional to the gate charge of the device being driven, as well as the switching frequency the system is running at. The ADuM7234 should definitely be able to run the power device you mention, with a 120 A target current. This all depends on the max ambient temperature the system will be running at as well.

    RSchnell

  • One more note: Since propagation delay is important to your setup, look at the ADuM3223. It's pinout and function are similar to the ADuM7234, but the propagation delay is much lower.

    RSchnell

  • Hi RSchnell

    Thank you for you answers.

    I made some changes in the circuit:

    - Took the opto-couplers off and substitute them with a cd4041.

    - Introduced two new power sources, so now I have:

          - 1, 5V power source, that powers the CD4041 and the signal side of both ADuM7234

          - 2, 12V power sources, one for each ADuM7234

          - 1, 80V power source, for the power MOSFETS

    - Connected the Disable pins to 5VGND

    - Connected both 12VGND with the 80VGND

    I believe I burned one of the ADuM7234 because when I turn the board on (as you said, first powering the ICs and then injecting the signal) it starts to heat and there is a DC component on the signal, as you can see on the pictures, that also increases with time. To be sure it wasn't an assembly driven mistake, I interchanged both ADuM7234 and the same IC overheated.

    Fig1: Signal after the CD4041 inverting/levelling stage disconnected from the ADuM7234

    Fig2: Signal after the CD4041 inverting/levelling stage connected from the ADuM7234 (CH1 on yellow connects to the ADuM7234 that overheats)

     

    I'll receive the new components and a prototype socket for the ADuM7234 in the end of the week to carry on with the tests and check if I still have the wave distortion on the MOSFET gates with increasing Vdd.

    Questions:

    - What do you believe on the changes that introduced in the circuit, do you believe they are relevant?

    - The CD4041 introduces a slope in the square wave signal: will the ADuM7234 follow this signal, meaning that the MOSFETs will be driven slowly during the turn off and on time, smoothing this periods or will the ADuM7234 generate a square wave when the signal passes a certain threshold, resulting in a square wave with a reduced duty cycle?

    Is this signal characteristic good or it will affect the MOSFETs lifetime, stressing them?

    - In the breadboard I'm constantly seeing noise in the 15.4MHz frequency, I tried to attenuate it with a ferrite which helped, and with a 10MHz low pass filter that introduces too much distortion in the square wave, I believe because it filters vibration frequencies needed to create the square wave itself. Are there other techniques?

    - Following to your answer to my second question (

    The ADuM7234 has CMOS inputs, and when it is powered up, should look like a high-impedance. When the ADuM7234 is not powered up, putting voltage on the input pins may parasitically power the ADuM7234 through internal protection diodes, so it is no recommended to feed the input pins when the unit is unpowered on VDD1.

    ). Can you pass me some article to better understand the process you described?

    My best regards

    Jorge

  • Hello RSchnell

    So I mange to connect all the circuit with independent 12V power sources for each ADuM7234.

    Results:

    One of the ICs seems to be working accordingly but the other has a lot of noise and just the low gate output is on as you can see in the pictures:

    fig1: High side gate (yello), inverse signal in (blue) - h-bridge power off

    fig2: Low side gate (yello), inverse signal in (blue) - h-bridge power off

    fig3: Low side gate (yello), inverse signal in (blue) - h-bridge power on

    Do you have an idea about what is happening? The chip is new and it's assembled is the exact same way as the other.

    Best regards

  • Hi,

    I don't know what is causing that. Is it possible to get a photo of the test setup with the probes you are using?

    RSchnell