Load sharing mod for the 18650 battery shield v3

I submitted a new video to my local OSH group about modifying the 18650 battery shield v3 to add a load sharing circuit:

The shield has a circuit design flaw that makes it unsafe to use with projects drawing more than 60mA, and the mod fixes that. The video also presents alternatives for operating safely within the limits of the original circuit without doing the modification.

The basic problem is that when the USB charging brick is plugged in, current to charge the battery and power the load all flows through the TP4056 charger IC, which then fails to terminate charging as it should, which could be a safety issue. The load sharing mod has the brick directly powering the load so the charger can behave as it should.

I’ve nicknamed this module as the Shield of Turin because the second picture below reminds me of another Turin-related object that’s famous for how it looks in the negative.

I don’t have access to answer comments on Youtube, but can answer questions here.

jremington:
Thanks, looks useful. However, I'm confused by the schematic in the video, showing the orientation of the DMP1045U P-MOSFET. It appears that as opposed to a high side switch, the source and drain are swapped, so that the body diode will conduct current from the battery to the load, regardless of the gate potential. Is that intended to protect from a reversed battery?

Yes, the orientation is the opposite of what you normally see for a switch. But current will not flow through the diode because the diode is reverse-biased. With the power brick connected, you will have about 4.6V at the cathode of the SB220, which is also the source of the mosfet. But the maximum voltage at the battery is 4.2V. For any current to flow, you would need one diode drop lower at the source, or something like 3.5V. But it's actually 4.6V. So it's reverse biased.

But you do need the diode to block any potential flow the other way, and that's the reason for its orientation. If it's oriented normally, and the battery is discharged to below 3.9V or so, then excessive charging current could flow from the brick back into the battery through what would be a forward biased body diode.

So while the mosfet will conduct equally well in either direction if it's turned on, it's the body diode that determines how it has to be oriented. I linked to a Microchip app note below the video, and you'll see that they orient the mosfet the same way. I know it looks odd, but it does work.

And of course when the brick is not connected, the body diode is bypassed because the mosfet is on.

Thanks, and sorry for deleting the post -- I figured it out before you responded.

It is odd the the original designer did not include something like it, as that circuit is commonly used for reverse battery polarity protection as well.

Yes, but for polarity protection you don't even need the resistor. Well, as long as the battery voltage doesn't exceed the maximum GS voltage.

On the video in general, what kinda bothers me about the whole issue of "failure to terminate charging" is that it seems to be pretty much ignored even though from what I read it's a real problem. Andreas Spiess' review of this device seems to ignore the issue, as do designs shown in various Great Scott videos. It just seems commonplace to connect the output of the TP4056 to the battery and the boost converter and be done with it, and that may be a real safety problem if the load current is too high - because charging never terminates.