• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 3 answers
  • Subscribers 64 subscribers
  • Views 770 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

LMG5200: Using 5V bootstrap pin as floating power supply for ADC

Kiron Sen
Prodigy 60 points
Part Number: LMG5200

I am designing an inverter using the LMG5200 and want to measure in-line phase currents using shunt resistors. To measure the shunt voltage, I'm considering using the LMG5200's bootstrap pins (HB and HS) as a floating 5V power supply to power an AMC1303 delta-sigma modulator ADC. I have some questions:

1. Would it be possible to supply a max of 10mA using this pin?

2. If so, would I need a larger bootstrap capacitor than what is recommended by the datasheet?

3. Would the noise coupled by the FET switching through the floating power rail have a negative effect on my ADC readings?

Thank you!

  • 0
    TI__Genius 16390 points

    Hi Kiron,

    That's a very interesting question. I will gather more information and circle back soon.

    Regards,

  • 0
    TI__Genius 16390 points

    Hi Kiron,

    1. It depends on the duty cycle of the low-side device. The 10 mA current could be handled, however you will need a larger min low-side on time to keep the bootstrap charged. If we say the peak current that the bootstrap  diode could handle is on the order of 0.5A during its charging time, then this means that the min low-side on time would have to increase by 10mA/0.5A = 2% to compensate for it. So for example, if you are switching at 100 kHz, the min low-side on time would increase by 200 ns. So the min low-side on time required to keep the bootstrap charged may go from say 50 ns to 250 ns.

    2. For the bootstrap capacitor size, the same formula found in section 9.2.2.2 in the datasheet still applies, except instead of using the quiescent current for the driver, it has to be the quiescent current for the driver+ADC current. 

    3. You can start by trying out the supply on the ADC to see if the noise is in an acceptable range. If it is a problem, a ferrite-bead + cap could filter the supply for the ADC, or maybe even a R-C if the voltage drop could be tolerated.

    Hope this helps!

    Regards,

  • 0 in reply to Yichi Zhang.
    Prodigy 60 points

    Hi Yichi,

    Thanks for the detailed reply!

    I was wondering where you got the 0.5A peak bootstrap diode current number from?

    Also, if I were to use a floating power supply instead, do you have a recommended TI isolated power supply with a +24V input that supports the same transient immunity as the max slew rate of the LMG5200, 50 kV/us?

    Best,

    Kiron

  • 0 in reply to Kiron Sen
    TI__Genius 16390 points

    Hello Kiron,

    The 0.5A is an approximation from Qcb/tmin where Qcb is the total gate charge plus the charge for quiescent current, and tmin is the smallest on time. An approximation of 25nC and 50ns yields the 0.5A.

    For isolated power supply, the EVM we are using for LMG3410 series is transformer based followed by voltage quadrupler and diode for voltage clamping. The transformer controller is SN6505BDBVR from TI. However, feel free to contact other TI teams for parts recommendations for isolated power supply.

    Here's a link for reference of LMG3410 EVM: https://www.ti.com/lit/ug/snou165a/snou165a.pdf?ts=1592354981565&ref_url=https%253A%252F%252Fwww.ti.com%252Ftool%252FLMG3410EVM-018

    Regards,

  • 0 in reply to Yichi Zhang.
    Prodigy 170 points

    Hello Yichi, 

    I am Minkyun Noh, working with Kiron on a project using LMG5200. I have a follow-up questions on your last answer.

    1. It seems that 0.5 A is the current 'needed' to charge Qcb = 25 nC for tmin = 50ns. Can we say this is the maximum current that the bootstrap diode can conduct?

    2. As for Qcb = 25 nC, I found from the datasheet p.6 that Total Gate Charge is Qg= 3.8 nC and Output Charge is Qoss = 21 nC. The sum of the two is Qg + Qoss = 24.8 nC, which is approximately 25 nC. Is this what you refer as Qcb?

    3. Where does the number tmin = 50 ns come from? I cannot find it from the datasheet. I am trying to connect your answers back to the datasheet information.

    4. More fundamental question we want to address is the following: Can we use the 5V bootstrap rail as a floating voltage source for an external device, such as AMC1303M2520. This is for in-line shunt measurement, so its ground will reference to the half-bridge pole voltage. If the bootstrap diode does not have a max current limit, can we place an over-sized bootstrap capacitor and let it be a stiff voltage source? Are there any potential problems you can see in this concept? 

    I will really appreciate your help!

    Best regards,

    Minkyun Noh

  • 0 in reply to Minkyun Noh
    TI__Genius 16390 points

    Hello Minkyun,

    Let me try to answer the questions:

    1. Yes, the definition is correct. However, this number was arbitrarily chosen as an example. Looking at the datasheet, the actual current that will be flow through should be much lower. Having another thought, the example of 0.5A is probably is over the abs max current allowed through the diode. From the datasheet, at 100 mA current, the bootstrap diode can see 1V drop and up to 2.8 Ohms dynamic resistance (slope of the line at that point) If we assume the dynamic resistance is constant, then at 0.5A you would have a max of (0.5-0.1)*2.8+1V = 2.12V drop, which will trigger the UVLO if the current is 0.5A. Now again from the datasheet, the gate charge and quiescent current is given, you can estimate the current with a minimum on time (Qg+(T-ton)*Iq)/ton, and the number should be much smaller.

    2. The tmin of 50ns is another arbitrarily chosen number as well. The minimum on time depends on your application.

    3. Tie back to the original question, theoretically it could work. By using big size cap, you will need to control the time for the initial charge. However, there are other factors such as the load current. If the current is flowing out of the FET, then the IL*Rdson adds to the bootstap voltage, if it is flowing into the half-bridge, that term subtracts from the bootstrap voltage. In this case, it can be very difficult to keep the bootstrap up.

    As you can see, there are lots of factors which are not always easy to figure out. In the end you have to build it and test it (at cold where the bootstrap drop is the worst) to confirm your bootstrap is staying up.

    Regards,

  • 0 in reply to Yichi Zhang.
    Prodigy 170 points

    Hello Yuchi, 

    Thank you for the answers. These are very helpful! I appreciate the point that the fully-charged bootstrap voltage can be affected by the load current. This is something I completely missed!

    Best regards,

    Minkyun