| Electronics > Projects, Designs, and Technical Stuff |
| Sanity check on Li-Ion charger power path. |
| (1/2) > >> |
| photomankc:
I'm working on a custom design for my mobile robot for power distribution and battery charging. Previous prototypes didn't include this because of the difficulty of charging Li-Ion and I did not trust my original designs with charging a fairly hefty 3 cell pack. I'm now working on the second incarnation of the robot and want to avoid needing to remove the pack for charging all the time. The new design plans on the use of the MPS MP26123 charging chip. The support components seemed reasonable and it allows me to charge a 3 cell pack. They have a reference circuit for powering the device from the AC adapter while charging the batteries and in studying that I have a question. The circuit on pg12 shows two MOSFETS controlling the power path to Vsys. The problem for me is that they show two P-Channel depletion mode MOSFETS in the digram which seem to be something like unicorns. Mouser drops to three results with "P-Channel" and "Channel Mode: Depletion". The text of the datasheet really doesn't clarify much. They are using the body diode of "M1" to block VIN from feeding back to VCC when /ACON is high as when the charger supply is not present. In this condition the battery can definitely supply "Vsys" through the body diode of "M2" and I'm suspecting the intention is that this depletion mode MOSFET is "on" because a positive voltage is not present providing a better path than the diode when no charger voltage is present. If /ACON is pulled low for a valid charger being present then MOSFET M1 is 'on' and conducts providing a better path than the body diode. It appears even if /ACON is not pulled low but a charger is connected the body diode will pass the current at whatever penalty in voltage drop you take for that. So, my question is could I substitute P-Channel enhancement mode MOSFETS instead. For M1 (Q1) in my schematic I think it's already fine. When /ACOK is low the MOSFET is 'on' because that line is pulled low. When /ACOK floats it is pulled high by R6 and that will have the MOSFET "off". For the battery MOSFET M2 (Q2) in my schematic I believe a change is needed. The MOSFET will no longer default to "on" in the absence of a gate signal so a pull-down is needed to switch it "on" when there is no charger voltage provided this is R11 in my schematic. If a charger is plugged in then Q2's gate is made high and the MOSFET is now switched 'off' and the body diode blocks that path to the battery allowing the IC to perform its voodoo. Does that seem correct? Sometimes I get wrapped around the axle on MOSFETs, especially P-Channels so I'm hoping someone can validate me on that. Attached are the datasheet for the charger IC and a screenshot of my charger block. The battery in mine is attached via a board connector that is beyond the fuse F2 on the left side of the schematic. Sorry, I know that is not the easiest to follow now. |
| Peabody:
You might want to look at this thread: https://www.eevblog.com/forum/projects/my-new-2-3s-lithium-battery-charger-burned-out!!!!!!!!!/ |
| Siwastaja:
These chips don't do desaturation or SOA protection for the MOSFETs they drive, when they really should; so it's completely up to you to do the SOA analysis. Application notes do not even hint about this, so you easily fail this part. This protection is fairly easy to integrate on the driver IC, but as they just don't do it, it's harder to add externally. There is no easy way out, if you have any significant bulk capacitance like you probably have in a robot. Here's one case I encountered (by my own stupidity, following appnotes without giving it enough thought): https://www.eevblog.com/forum/projects/bq24610-power-management-blown-fet-issue/msg1461488/ |
| photomankc:
--- Quote from: Peabody on December 08, 2018, 04:59:23 am ---You might want to look at this thread: https://www.eevblog.com/forum/projects/my-new-2-3s-lithium-battery-charger-burned-out!!!!!!!!!/ --- End quote --- Wonderful. :-\ |
| photomankc:
--- Quote from: Siwastaja on December 08, 2018, 10:03:36 am ---These chips don't do desaturation or SOA protection for the MOSFETs they drive, when they really should; so it's completely up to you to do the SOA analysis. Application notes do not even hint about this, so you easily fail this part. This protection is fairly easy to integrate on the driver IC, but as they just don't do it, it's harder to add externally. There is no easy way out, if you have any significant bulk capacitance like you probably have in a robot. Here's one case I encountered (by my own stupidity, following appnotes without giving it enough thought): https://www.eevblog.com/forum/projects/bq24610-power-management-blown-fet-issue/msg1461488/ --- End quote --- Thanks for sharing your experience there. In this case a separate system handles switching on the main load and the additional capacitance on the reset of the board would not be present on the initial insertion of the AC charger. The mystery burn-out in the other thread while running on battery does concern me. |
| Navigation |
| Message Index |
| Next page |