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.

BQ25756: Higher temperatures than expected

Part Number: BQ25756
Other Parts Discussed in Thread: UCC27714

Tool/software:

Hello,

I would like to develop a charger for LiIon battery packs with 12S, i.e. 50.4V max. I want to charge them with up to 20A and and 48V input voltage.

I designed a custom PCB for this purpose using the BQ25756. Testing the design showed, that the inductor (Würth Elektronik 74435580680) and the Mosfets (Vishay SIR880BDP-T1-RE3) get rather hot. Charging with only 3A gets the Mosfets to about 55-60°C. Charging with 6A lead to extremely high temps in the inductor of about 170°C measured directly on the windings, and rising. Also high temps on the mosfets of about 110°C. These tests even included some small heatsinks and a fan for cooling the components. As the Inductor is supposed to have a DCR of 2.1 mOhm, I was not expecting thermals like this.

Attached you can find a picture of my oscilloscope. X axis spacing is 1us/div. It shows the voltage across the inductor at 3A charging current. Input was 48V, V_BATT was 46.5V.

Can somebody tell me if this behaviour is expected? If yes, what could be done, to drastically improve the performance, so I will be able to charge with 20A?

Thanks,

Max

  • Hello Maximillian,

    Getting up to 170°C with only 6A of charging current is not expected behavior. I've got a few questions and test suggestions to help us debug this:

    • If possible, can you send me your schematic/layout?
    • Can you take an oscilloscope capture of LODRV/HIDRV and SW?
    • Do you see a change in inductor performance when you use an inductor with a higher DCR?
    • Can you measure the REGN and DRV_SUP voltage?

    Best Regards,
    Ethan Galloway

  • Thanks for the answer.

    Archiv.zip Those are the schematic and layout files.

    The following images are taken at 3A charging current and were measured on the gate pins of the mosfets.

    LODRV1This is LODRV1.

    HIDRV1This is HIDRV1.

    LODRV2This is LODRV2.

    I lost the image of HIDRV2. I only remember it was constantly "high".

    I was not able to test with a different inductor. Do you have resommendations what to buy and test?

    Unfortunately I am on holiday this week, so I can only deliver the REGN and DRV_SUP data next week.

    I hope this information already helps.

    Max

  • Hello Max,

    Thanks for the new information. I'll take a look at this later this week and get back to you.

    Best Regards,
    Ethan Galloway

  • Hello Max,

    Here are my thoughts for the schematic and a few test suggestions:

    • Does the circuit behavior improve if you lower the switching frequency?
    • Does the BTST voltage look good?
    • Does the circuit behavior change if you replace the IC?

    The schematic looks good. By the way, the circuit has 1320µF of capacitance on the input and output of the charger? This will work, but you might be able to save on the BOM cost by reducing this capacitance.

    The layout looks good. By the way, I wouldn't route HIDRV and LODRV so close to each other. HIDRV's reference is SW and LODRV is referenced to GND. I don't think this will majorly impact the circuit performance though.

    Best Regards,
    Ethan Galloway

  • Hi Ethan,

    - I think the circuit gets a little better with 100k instead of 57k6 on Pin 36. When charging with 7A, the inductor temperature rose slower than before. But it still gets to 100°C within ~1.5 minutes and continues to increase.

    - I attached some images below sjowing the BTST pins during charging.

    - I tried with two different PCBs of the same design. Both behave identically, so I think it is not a bad component.

    I wasn't really sure how much capacitance I need and thought for the first try it couldn't hurt to have more than enough capacitance.

    "Bad" routing of the DRV lines is also something I noticed after ordering the PCBs. What is the recommended thickness/width and distance between the traces?

    BTST1 2A 450kHzBTST1 2A 450kHz

    BTST2 2A 450kHzBTST2 2A 450kHz

    BTST1 7A 300kHzBTST1 7A 300kHz

    BTST2 7A 300kHzBTST2 7A 300kHz

    Kind Regards,

    Max

  • Hello Max,

    Thanks for the new information. I'll look over your results tomorrow.

    By the way, the BTST voltage looks normal. The last image you have captures the clock jitter on BTST2 and this is expected behavior. The clock jitter isn't a problem for the battery charger system.

    Also, are you using an external gate drive for DRV_SUP?

    Best Regards,
    Ethan Galloway

  • Hi Ethan,

    yes, I am using the 5V output from a LMR38015SQDRRRQ1 buck circuit on DRV_SUP. Should I try to disconnect that? The 5V output might be not the smoothest. It was no problem for the ESP32 so far, as its 3V3 input voltage gets regulated by an additional LDO. The DRV_SUP is directly connected to the 5V buck output.

    Kind Regards,

    Max

  • Hello Max,

    Thanks for the new information.

    I have a few suggestions:

    • Can you measure the signals on LODRV again to make sure those signals are good? - The previous signals of LODRV don't look very good. I think a bad voltage on DRV_SUP might be to blame.

    • Can you increase the external gate drive supply to 7V or 10V? This might help with the conduction losses.
    • Also, can you measure the voltage on DRV_SUP with an oscilloscope to make sure the voltage is good?
    • Can you lower the switching frequency to 200kHz? This might improve the system thermals further.
    What is the recommended thickness/width and distance between the traces?

    The HIDRV, LODRV, and SW traces need to be at least 20mil. SW is the return path for HIDRV. So, HIDRV and SW need to be routed close to each other. GND is the return path for LODRV. So, LODRV and GND need to be routed close to each other.

    Best Regards,
    Ethan Galloway

  • Hi Ethan,

    in the pictures below I made some more measurements. I noticed, that between 5A and 7A charging, there is a huge difference in LODRV1 and also the spikes in the DRV_SUP got alot bigger. I don't know if that could be the reason/problem. DRV_SUP is drawing 2mA at idle and around 25mA during charging according to the PSU.

    I am not able to easily change from 300kHz to 200kHz, as I don't have the needed 200kOhm resistor at the moment. I could try to hack something with two 100kOhm resistors, but I didn't think that is necessary, as the difference between 450kHz and 300kHz didn't not show a massive improvement. If you think that could still make the needed difference, let me know and I will of course try it or order 200kOhm resistors.

    I hope the pictures are helpful. Let me know if you need anything else.

    Kind Regards,

    Max

    LODRV1 5A LODRV1 5A 300kHz with internal 5V at DRV_SUP

    LODRV2 5A 300kHzLODRV2 5A 300kHz with internal 5V at DRV_SUP

    5V DRV_SUPinternal 5V DRV_SUP  at 5A

    LODRV1 5A 300kHz with 7V at DRV_SUPLODRV1 5A 300kHz with external 7V at DRV_SUP

    external 7V at 5Aexternal 7V DRV_SUP  at 5A (KORAD KWR103 PSU)

    LODRV1 7A 300kHz with external 7V at DRV_SUPLODRV1 7A 300kHz with external 7V at DRV_SUP

    external 7V DRV_SUP  at 7A (KORAD KWR103 PSU)external 7V DRV_SUP  at 7A (KORAD KWR103 PSU)

  • Hello Max,

    Thanks for the new information and thanks for doing all these tests.

    DRV_SUP does have a lot of noise on it. I think maybe this capacitor is bad. Can you send me the part number for the DRV_SUP capacitor? Do you see a difference in the signals with a higher capacitance of about 10µF on DRV_SUP?

    I recommend using the external gate drive voltage of 7V or 10V. The charger will need this to reduce the thermal losses at high currents.

    Do you have a current probe to measure IBAT by the way? It might be useful to see SW, LODRV, IBAT, and VBAT on the same plot.

    Best Regards,
    Ethan Galloway

  • Hi Ethan,

    I have some good news! I spoke with an engineer at Würth Elektronik, and he was also very helpful. He immediately pointed out that the originally chosen inductor was a poor fit for my application.

    I didn’t fully understand all the technical details, but essentially, everything about this inductor contributed to excessive heating. The flat windings resulted in high capacitance, the large air gap caused eddy currents (if I remember correctly), and the core material led to inductance fluctuations with temperature. This meant that as the inductor heated up, the problem only got worse.

    In the end, he recommended using two 74439370047 inductors in series. I soldered them in today, and that actually solved the issue! While charging at 15A, the inductor temperature only reached 40°C, which is a huge improvement.

    The only remaining issue is that the MOSFETs are still running a bit too hot at 15A and become far too hot at 20A. So, I’m considering upgrading them to reduce cooling requirements. I was thinking about trying the NTMFS6H800NT1G as a replacement for the currently installed SIR880BDP-T1-RE3.

    Do you see any potential issues with running these MOSFETs? They have a slightly slower fall time and a slightly higher total gate charge (Qg), but apart from that, they seem like a good fit. If you don’t see any major concerns, I’d go ahead and order them for testing.

    Thanks,
    Max

  • Hello Maximilian,

    That's great to hear! Thanks for letting me know about the inductors.

    Do you see any potential issues with running these MOSFETs? They have a slightly slower fall time and a slightly higher total gate charge (Qg), but apart from that, they seem like a good fit.

    The higher gate charge and slower fall time will increase the switching losses of the NTMFS6H800NT1G. The lower RDS(ON) will reduce the conduction losses. I think these FETs will work to reduce the heat generated when the switching frequency is low.

    Also, you'll need to use an external gate drive with the NTMFS6H800NT1G because the threshold voltage is higher than the SIR880BDP-T1-RE3. The internal LDO will have trouble driving a FET with a threshold voltage between 2V to 4V.

    Best Regards,
    Ethan Galloway

  • Hi Ethan,

    Do you have any other recommendations for MOSFETs? Since the chip is rated for 20A, I assume you have some test setups with MOSFETs capable of delivering 20A. The SIR880BDP-T1-RE3 gets too hot even at 200kHz, but all other MOSFETs I found perform worse, according to the Excel calculator document (https://www.ti.com/tool/download/BQ25756-DESIGN-CALC).

    That also matches my observations when testing the NTMFS6H800NT1G. The calculator indicates that it has about twice the MOSFET losses, and I would estimate that it heats up roughly twice as fast as the SIR880BDP-T1-RE3.

    Kind regards,
    Max

  • Hello Maximilian,

    Do you have any other recommendations for MOSFETs? Since the chip is rated for 20A, I assume you have some test setups with MOSFETs capable of delivering 20A.

    We've found that we have to active or passive cooling to keep the FETs cool at high voltages and high currents. Is it possible to attach a heat sink to the FETs on your board?

    At lower voltages, we've used the AONS66408 for >10A charging, but these won't work for your application's voltages.

    Best Regards,
    Ethan Galloway

  • I have already tried adding small to medium-sized passive heatsinks to improve thermal performance on top of the MOSFETs and bottom of PCB, but unfortunately, the results have not been sufficient. My goal is to keep the temperature rise below 50 K.

    As I posted before the BSC033N08NS5SC is only expected to perform 50% worse then the SiR880BDP, but would allow for direct metal contact to a heatsink. I am not sure if that is enough to offset the theoretical 50% higher power loss, whats your take on this?

    I am also considering using an external MOSFET driver controlled by the internal driver. This would allow me to switch MOSFETs with larger footprints and lower R_DS(on) (both to reduce W/mm^2), even if they have a higher gate capacitance. One option I am considering is the UCC27714. However, I am concerned that the HIDRV voltage might be too high for the UCC27714's inputs. Would it be feasible to use a simple voltage divider to bring the voltage within an acceptable range for the external driver?

    Additionally, I am open to recommendations for other external MOSFET drivers that could fulfill this role. If there are alternative charger ICs that might be better suited to my requirements, I would also appreciate your suggestions.

    Looking forward to your insights!

    Best regards,
    Max

  • Hello Max,

    Thanks for the new information. I have a few more suggestions and questions:

    • Have you tried the SIR680LDP in your circuit? It's got a lower RDS(ON) than the SiR880BDP.
    • Is it possible to use two BQ25756's in parallel for your application?
    • A temperature rise of under 50K might be hard to achieve. What temperatures are you currently seeing on the FETs?
    As I posted before the BSC033N08NS5SC is only expected to perform 50% worse then the SiR880BDP, but would allow for direct metal contact to a heatsink. I am not sure if that is enough to offset the theoretical 50% higher power loss, whats your take on this?

    I'm not sure if the direct metal contact with the heatsink would be worth it. The 50% worse sounds pretty extreme.

    Would it be feasible to use a simple voltage divider to bring the voltage within an acceptable range for the external driver?

    I wouldn't recommend using a external gate driver. The BQ25756's gate drivers can output a 100% duty cycle. Most external gate drivers can't support a 100% duty cycle and the converter is not going to work correctly here.

    Best Regards,
    Ethan Galloway

  • Hi Ethan,

    • Have you tried the SIR680LDP in your circuit? It's got a lower RDS(ON) than the SiR880BDP.
    • Is it possible to use two BQ25756's in parallel for your application?
    • A temperature rise of under 50K might be hard to achieve. What temperatures are you currently seeing on the FETs?

    - I did not try them, but according to the excel calculator they also perform worse. (screenshots attached)

    - In theory that is possible, but I was hoping to use them their full 20A. I would synchronize them just by sending the same I2C commands at the same time? Also I would need two seperate I2C lines as the chips address is fixed, right? Are there any other potential safety issues with using two seperate chargers at the same time?

    - I am not totally sure about that, as I am testing many different setups right now. But at 15A they quickly get to ~150°C when I shut them off. I think if I leave them running they will just burn up. I dont know the numbers for 10A, but its definitely also above 100°C.

    I'm not sure if the direct metal contact with the heatsink would be worth it. The 50% worse sounds pretty extreme.

    Ok, good to know.

    I wouldn't recommend using a external gate driver. The BQ25756's gate drivers can output a 100% duty cycle. Most external gate drivers can't support a 100% duty cycle and the converter is not going to work correctly here.

    Got it, I think two chargers would be the better option then.

    Best Regards,

    Max

    SiR880BDP

    SiR680LDP

  • Hello Max,

    - I did not try them, but according to the excel calculator they also perform worse. (screenshots attached)

    Thanks for letting me know about this. I'll look into this. I think the SiR680LDP would be better for your system though. Most of your losses seem to be conduction losses and the SiR680LDP has better conduction characteristics compared to the SiR880BDP.

    - In theory that is possible, but I was hoping to use them their full 20A. I would synchronize them just by sending the same I2C commands at the same time? Also I would need two seperate I2C lines as the chips address is fixed, right? Are there any other potential safety issues with using two seperate chargers at the same time?

    You'll need to use an I2C mux to talk to both chargers at the same time. You can synchronize the switching by using an external clock signal. The chargers will work in constant current mode just fine. You'll need to make sure one the chargers has a lower termination voltage though. This will make sure the chargers don't oscillate ON and OFF as both chargers reach the termination current at the same time.

    Best Regards,
    Ethan Galloway