LT3741-1 Bottom Gate Never Turns On

I have fabricated and assembled a PCB that contains eight LT3741-1 regulators in parallel in order to do high current testing on PCB traces. I am at an early stage where I just have one of the regulators populated with components. When enabling the regulator only top gate fires off. I have tried various resistive loads and the activity remains the same. Absolutely no activity on bottom gate is detected.

Power integrity appears fine on VCC_INT. Vref comes up just fine. VC goes to 3.0V and sees sawtooth shaped drops. Pulse width on top gate is around 100ns and the frequency is oddly around 85kHz.

LTSpice shows a nice sloping rise on Bottom Gate upon asserting Enable. This is not observed in hardware. Any help is much appreciated.

Target max output current will be up to 50A per regulator but for now I am just trying to get a single regulator to run at any current.

Top Replies

Parents
  • +1
    •  Analog Employees 
    on May 6, 2020 4:52 PM

    I looked at your schematic.  My guess is that you have no bottom gate drive because your RSENSE (R4) is so small - 0.4mOhms.  I did this post for the similar LT3763 for someone else who was not seeing BG pulses.  The post is titled "LT3763:  gate driver doesn't pulse".

    The 50A output current of your circuit is relatively high and I assume the output voltage of your pcb trace load is relatively low so it might be a good idea to get a reference design running first, and then adapt the reference design to your requirements.  The image below shows the closest reference design that I could find:  the LT3743 driving an LED at 40A.  You can make a version of the circuit without the LT3743 features for your purposes using the LT3741-1 instead of the LT3743.  It may be a good idea to use LT3741 demonstration circuit DC1602A to make sure the pcb layout is correct.  Then, I'm sure it will work to put your pcb trace load in series with the LED in the reference design.  Then the challenge will be to get the circuit to run smoothly without the LED and just your pcb trace as a load.  It is a challenge because a 12VIN buck may have a problem with a very low output voltage due to minimum on time limitations.  The high 50A output current can make stability a challenge also and that is why I recommend starting with a reference design.  

    LT3743 DS example app. 40A_5.png

Reply
  • +1
    •  Analog Employees 
    on May 6, 2020 4:52 PM

    I looked at your schematic.  My guess is that you have no bottom gate drive because your RSENSE (R4) is so small - 0.4mOhms.  I did this post for the similar LT3763 for someone else who was not seeing BG pulses.  The post is titled "LT3763:  gate driver doesn't pulse".

    The 50A output current of your circuit is relatively high and I assume the output voltage of your pcb trace load is relatively low so it might be a good idea to get a reference design running first, and then adapt the reference design to your requirements.  The image below shows the closest reference design that I could find:  the LT3743 driving an LED at 40A.  You can make a version of the circuit without the LT3743 features for your purposes using the LT3741-1 instead of the LT3743.  It may be a good idea to use LT3741 demonstration circuit DC1602A to make sure the pcb layout is correct.  Then, I'm sure it will work to put your pcb trace load in series with the LED in the reference design.  Then the challenge will be to get the circuit to run smoothly without the LED and just your pcb trace as a load.  It is a challenge because a 12VIN buck may have a problem with a very low output voltage due to minimum on time limitations.  The high 50A output current can make stability a challenge also and that is why I recommend starting with a reference design.  

    LT3743 DS example app. 40A_5.png

Children
  • ARad,

    Thank you for this very helpful answer. I have explored higher valued current sense resistors but still was unable to get the circuit up and running. In the interest of time I chose to abandon the project and produce high currents using alternative methods. Should I choose to come back to this I will very likely take the design route you suggested.

    Thanks again!