-
Using an N-Channel Mosfet as a reverse polarity protection diode
Posted by waymond91 on 09 May, 2024 09:08 -
Ok, I think I am getting myself turned around here.
I have a circuit that uses 8 18650 lipos in parallel.
I was previously going to use a schottky diode for reverse polarity protection, and reverse current protection on each the battery cells.
However the power losses are pretty substantial for the application.
I was then going to use a p-channel mosfet as high-side reverse polarity protection.
Then I came to believe that if I do this, if one of the cell voltages is higher than the other, the lower cell might have some current flowing into it...
After poking around and asking chat gpt, I now want to use an n-channel mosfet on the high side with the source and drain shorted. I believe this should functionally behave as diode with minimal forward voltage drop, but this seems too good to be true...
This seems very simplistic, and I cant quite pin down the behaviour of the body diode. Im under the impression the body diode goes from
SOURCE -> DRAIN and therefore will not be backflowing current into the battery cell if a cell is under charged.
I know this should be basic but Ive got my wires crossed before - just hoping a real human being could point me in the right direction on this one
-
As shown the MOSFETs would never turn on. There are missing the gate voltage higher than the source to turn on. This would need an extra more postive voltage and control circuitry.
For the positive side p-channel MOSFETs are easier, though usually higher R_on.
One could use n-channel FETs on the low side / negative side of the batteries. -
With that setup you're using just the MOSFET's body diode. It's basically the same as using any other diode. Something like an ideal diode controller could be what you're looking for.
-
You state that the power losses using Schottky diodes are substantial...I would think they would be less than using the body diode of a FET. Please provide part numbers and data.
-
What you want is something like this, the zener is optional, only needed for voltages exceeding the G-S breakdown.
-
What you want is something like this, the zener is optional, only needed for voltages exceeding the G-S breakdown.
That will work as reverse polarity protection, but isn't safe for connecting two batteries in parallel.
An ideal diode circuit is required. Here's one with a P-MOSFET. The BJTs should be matched pairs, such as the DMMT5401.Ideal diode discrete.asc
-
A PFET on the high side can be used for this but usually my company does not due to lack of low Rds parts and cost. Usually use a NFET with a ideal diode controller like a LM5050.
I used one of those on a project a few years ago that had to do about 500A in the correct direction but block reverse. There are some incredible fets out there. -
If the positive side can be common, then the circuit will work with N-MOSFETs and NPN transistors.
-
A PFET on the high side can be used for this but usually my company does not due to lack of low Rds parts and cost. Usually use a NFET with a ideal diode controller like a LM5050.
Uh. Please tell us more about your requirements, because sure in general there are more NMOS with low Rdson and lower cost for similar characteristics, but I have never has any issue finding appropriate PMOS for this for applications from a few mA to a few A, and that would still be cheaper than a NMOS + a LM5050. Of course there are always exceptions, but making this a company-wide rule sounds pretty counter-engineering. Oh well. -
If the positive side can be common, then the circuit will work with N-MOSFETs and NPN transistors.
Good example, there are a couple of points to watch
* For single cell use, you need to choose a low threshold fet (sub 2V?) - you need RDS still low at the cell discharge.
eg You do not want a 3.5V threshold fet.
* For higher voltages, watch the reverse VBE rating of the NPN pair.
ROHM and Toshiba make muting transistor pairs with VBE ratings up to 20V, tho they do not define the VBE matching.
-
Wow guys, thank you so much for the awesome feedback! I had no idea that there was such a circuit as an "ideal diode" so it was quite interesting to dive in a bit.
Because of my inexperience, anytime I see more than 2 transistors in a sub-circuit, or I get overwhelmed with bias resistors, or I get a spooky feeling my circuit needs hysterisis, I really just start looking for an IC
So what I landed on was the MAX40200: https://www.analog.com/media/en/technical-documentation/data-sheets/max40200.pdf
It seems like an appropriate choice here. My application draws about 1.5 - 2 watts (but maybe has higher transients...) so the 1.5 max amperage should have me covered.
The input voltage range looks right for a single cell lipo (2.5 - 5)
It lists the quiscient current at 15 uA max which is nice because the device will be "off" for long periods of time.
I guess one of my big concerns besides jsut generally am I doing this right:
1. The device lists a load capacitance of .3 uF -> 100 uF. Is this just referring to the output capacitor? Or do I need to be checking the input capacitance to the switching regulator this all feeds into?
2. My application typically draws about 1.5 watts -> 2 watts. I believe with this 1.5A spec a single battery should suffice, but I am a little confused about how putting some many devices in parallel really plays out. I assume I should be able to draw more than 1.5A from the battery cells in a pinch
-
1. The device lists a load capacitance of .3 uF -> 100 uF. Is this just referring to the output capacitor? Or do I need to be checking the input capacitance to the switching regulator this all feeds into?
2. My application typically draws about 1.5 watts -> 2 watts. I believe with this 1.5A spec a single battery should suffice, but I am a little confused about how putting some many devices in parallel really plays out. I assume I should be able to draw more than 1.5A from the battery cells in a pinch
Maximum output capacitance is literally the maximum capacitance this IC can have on its output before it might become unstable and oscillate due to the internal feedback loop becoming unstable. This means, in general, that you should keep the combined output capacitance under 100uF. I'd even encourage doing a test and observing how the IC behaves in your application.
"In a pinch," for example, in a transient, you have the output capacitance to handle most of those.
From the datasheet: "These capacitors (Input & Output) also improve the surge capability of the power supply. In general, for higher Output Capacitive Loads [e.g., COUT = 10µF], then CIN should be kept to COUT/10 (µF) for optimum transient response."
-
Ok thank you for the response.
So in our case, since we have 8 of these diode controllers, putting a 10uF ouput capacitor on each one doesnt seem like a hot idea...
The output bus formed by these diode controllers feeds through a mosfet and then goes into a switching boost converter.
I am assuming the diode controller can "see" the input capacitance of the boost converter (44 uF) but cannot "see" the output capacitance of the boost converter?
Heres my switching regulator below:
If the above is true and I change the output capacitor to each diode to controller 1uF, this would bring the output capacitance to about 52 uF, so in that case I should add a 5.2uF input capacitor to each diode controller? -
Most circuits here for RPP have been over complex. The circuit in this post is simple and it works very well.
https://www.eevblog.com/forum/beginners/another-reverse-polarity-protection-question/msg5239122/#msg5239122
David -
Most circuits here for RPP have been over complex. The circuit in this post is simple and it works very well.
https://www.eevblog.com/forum/beginners/another-reverse-polarity-protection-question/msg5239122/#msg5239122
David
it won't work for ORing supplies because once the fet is on it won't act like a diode any more, it'll conduct in both directions
that's the point of the schematic Zero999 posted -
Most circuits here for RPP have been over complex. The circuit in this post is simple and it works very well.
The MOSFET only turns off when the power supply voltage drops below the threshold. When the battery is connected, it will bias the gate on, allowing current to flow through the MOSFET in both directions. Here's a simulation showing the battery voltage, when a voltage source is connected to the output and swept from 3.3V to 4.3V. The battery voltage simply follows the source voltage, minus the losses due to the MOSFET's on resistance.
https://www.eevblog.com/forum/beginners/another-reverse-polarity-protection-question/msg5239122/#msg5239122
David
Notice how the current through V1 changes direction, when the output voltage rises, indicating it's being charged. LTSpice shows negative current for discharging and positive current for charging.
-
Quote
Uh. Please tell us more about your requirements, because sure in general there are more NMOS with low Rdson and lower cost for similar characteristics, but I have never has any issue finding appropriate PMOS for this for applications from a few mA to a few A, and that would still be cheaper than a NMOS + a LM5050. Of course there are always exceptions, but making this a company-wide rule sounds pretty counter-engineering. Oh well.
Oh, not a rule but things usually cost out that way. For low currents, a regular or schottky diode. There have been some with a PFET over the years too. But, over a couple of amps things get to be a problem as most of our designs are to no less than 85C and many are 105C or 125C so there is not much room for temperature rise. No forced cooling either in these applications. Then the space constraint. For a given Rds, a PFET is more costly than a NFET too. My most recent design where I used a ideal diode controller and NFET for this application I had no room to fit a larger package PFET needed to get an acceptable low power dissipation as the Rds was too high compared to a NFET in a smaller package. Just this past week, a colleague was asking me about this as they have a D2PAK Pfet on a product that looks massive compared to the rest of the parts and they are looking to get rid of it somehow (as the equivalent controller and Nfet stands to be 1/2 of the BOM cost). They want to reclaim space for feature creep too I suspect.