Using a MOSFET in reverse

(I thought I posted this here earlier, I don't know what happened to it?)

I have an app I want to use a very high current/low rds on N-channel MOSFET. However, for isolation reasons, the intrinsic body diode is going the wrong way and will create problems.

So, recalling that MOSFETs are inherently symmetrical devices, can't I just reverse the source and drain and use it in reverse? What are the consequences of doing so? Does it change the gain or what?

Here's the background: I have a camper van with a secondary deep storage battery, an isolator (they're made of std high current diodes), and a huge 160 amp alternator. The alt's regulator references the main battery voltage. Now the problem is, the secondary battery gets deeply discharged at times, yet the float voltage used by the reg isn't enough for the high charge rate it could make use of, and it takes many hrs to charge. Since the isolator's diode drops about 0.7V (over 1V when high current is involved), I can boost the voltage by bypassing the isolator on the deep cycle's side. I'd use a MOSFET in satuation and some supplemental stuff to turn off the gate when it would drain the battery rather than charging it. However, the intrinsic diode still allows this, so I'd like to reverse it so it's just in parallel with and facing the same way as the isolator's diode.

how can it that the diode goes in the wrong way..???

if you want to shut down that diode.. wy not add a power full diode in the oposite direction..??/

maybe a shooky diode fits your needs like a U1620R..???


TKS

The whole point is to make a low voltage drop pass around the 0.7-1V drop on the isolator. The charging current that the secondary battery will load with is highly voltage dependant. The drop of a diode, even a Schottkey, at high currents is completely out of question. The whole point is to fix the drop problems of a diode which is already in the isolator.

I'm looking at using several IRF3703 in parallel. Each has an 0.0028 ohm Rdson and an ultimate current capacity (including the diode) of 210 amps. Several will keep the heat lower.

I'm still worried about where the heat would go if the MOSFET were off but the current goes through its diode instead of the isolator's diode. This might require a bigger heatsink than I would want to have. So I'm even wondering if I might use another MOSFET in series, with the drain/source in the conventional direction, which would double the Rdson but would mean there's a body diode going either way so it would ensure the diodes could never conduct when the gates are on.

Maybe I should give up and just go for one of those huge relays. I don't have good confidence in the reliability with such high currents being involved, and I might want to do PWM if certain issues come up, which the relay cannot do. I seem to recall seeing once that they took an inordinate amount of coil current, I'm trying to check up on that but among dozens of vendor pages not one has mentioned it (kinda suspicious).

If I understand what you are trying to do correctly you want to use a FET to lower the voltage drop across the diode at high currents. Some switching regulators use a fet like that. To prevent the body diode from ever turning on you can use a schottkey diode in parallel the schottkey will always turn on before the FET diode.

Make sure you don't exceed the charging current of the battery, 160 amps going into a battery is almost certain to blow it up. You might want to add a temp sensor as protection for overheating.

Brent

Keep in mind that the gate has to go more positive than the voltage on the battery you are charging. How much higher (10 volts or so?) depends on how hard you want to turn it on.
Other than the fact that P-channel MOSFETs tend to have higher ON resistance, could you use several of them in parallel? The advantage here is you can use GND as your gate drive - no boost required. You would still have to reverse drain and source, and a Schottky in parallel, as suggested by Brent, might be a good idea. However, you should look at the leakage current spec in a high current Schottky.
The series back-to-back scheme sounds like a good idea. I seem to recall seeing this scheme in power supply hot-switching circuits.

Why not just use a relay?.

here's some relays to look at https://www.aromat.com/pcsd/cb.pdf
The coil currents are also listed..The highest I saw was 1.8w the rest were in mw.. Contact rating of 40a should suffice ...

Contact rating of 40a is way too small! Duh... PAC/Stinger 200 amp relay is a minimum. It must have a rating equal to or greater than the alternator's rating.

You are quite right that PMOS has a higher rdson, thus after reviewing 50+ spec sheets I found NMOS to be a better choice. I will be using a voltage booster (MAX1044) to get more positive voltage. This is part of a microcontroller driven coulometric charge controller which also takes into account temp and voltage. Coulometrics accounts for amp hrs in vs amp hrs out to create an accurate guess of how much charge is needed. Battery voltage, when evaluated based on temp and current load will also be taken into account. In case of transistor or battery overheating, I can also just disable the gate and let the existing over-the-counter diode isolator take over for awhile.

The alternator's regulator holds the main battery voltage at ~13.8V. If the main battery doesn't need much current, the isolator drops to min 0.7V so 14.5V is available on the alternator output, but the other isolator diode will drop between 0.7V-1V to the deep cycle. That's not enough voltage for a bulk charge, it's more like a trickle.

I'd like to run the accessories/fridge all day, and if I'm camping and not driving around, I want idle the car for as little time as possible to restore it every day or two. Stupid to run a 200 hp engine and a 160 amp alternator to only produce 10 amps into the battery! That huge battery should be able to take 50 amps if it's not in overcharge. Charging at 14.5V is very healthy and a far more appropriate charge voltage.

But, also remember I want to be able to run a 75 amp inverter load while the engine is running. The battery will still regulate its own load and may be pulling 30 amps, so the circuit may have more than 100 amps on it even though the battery isn't drawing that.

Now I do have some huge 240 amp Schottkeys:
https://www.electro-tech-online.com/custompdfs/2004/06/245nq015.pdf
The reverse leakage should not be a problem, as I can turn off the MOSFET under those conditions. However, getting bulk charge currents is a matter of tenths of volts, I will already be dealing with the MOSFET rdson drop and I don't think it'll perform with an added Schottkey drop. Also, the Schottkey's in a large, weird package and I don't intend to build a 25W+ heatsink for it. Not a good idea unless the MOSFET just can't be done.

Brent, do you have any links or remember if there was a specific term for this arrangement? I need to know if the technology has negative consequences associated with doing this, the spec sheets never list properties for doing this.

What size wire do you have running from your alternator to battery?
Probably isn't much larger than #6.... Can't push 200a thru a #6..

The reverse leakage should not be a problem, as I can turn off the MOSFET under those conditions. However, getting bulk charge currents is a matter of tenths of volts, I will already be dealing with the MOSFET rdson drop and I don't think it'll perform with an added Schottkey drop. Also, the Schottkey's in a large, weird package and I don't intend to build a 25W+ heatsink for it. Not a good idea unless the MOSFET just can't be done.

Click to expand...

I think the idea on the Schottky diode was to put it in parallel with the MOSFET and the existing isolator diode, to ensure that the intrinsic MOSFET diode is not damaged by overcurrent when the MOSFET is off.

The intrinsic body diode of the IRF3703 is rated at the full 210 amps. The heat may be a problem especially since paralleling MOSFETs will not guarantee the current will be shared between devices when the diode is conducting.

Paralleling the Shottkey would not work. The battery would be left in overcharge during long term driving as the supply is stuck 0.4V higher than normal. We could replace both isolator diodes with Schottkeys, but then the leakage could result in killling the main battery if the deep cycle is drained. Also they are less constant on the voltage dropout vs temp/current, and this should result in poorer charge regulation.

I believe it was 4 ga wire I put in there... maybe 2 ga. I forget. The wire from the battery to the inverter is 2/0 (or "00") ga for low voltage drops over a relatively long run as well as high current handling abilities. I didn't mess around.

Have you read note 6 on p.8 of the datasheet?

 Calculated continuous current based on maximum allowable
junction temperature. Package limitation current is 75A

Click to expand...

My interpretation of this is that 210 amps continuous is not possible for one device. Three in parallel would be marginal.

Thanks, I didn't see that note, but I knew the current was limited by thermal capabilities of the pkg.

Now if the MOSFET is used forwards, this could be up to 100 amps continuous (and diodes cannot be paralleled effectively, so another doesn't help at all). If it's used in reverse, it's only going to forward bias when the starting battery voltage is less than 1.4V of the deep cycle voltage. Unfortunately, that could happen on starting the engine, which could suck hundreds of amps through the MOSFET diode. This sucks... I could still put a forward and reverse MOSFET in series, and switch them on/off together so they'll never conduct together, but then the rdson is 2x 0.0028 ohms. At 100 amps that's 0.56V, as well as 28W per pkg. I was planning on knocking that down by using 3 or 4 MOSFETs in parallel, but now I'd need twice as many. And that 3703 isn't cheap!

Grrrr.... the huge ass relay is looking better and better.

Now that I think about it, I could set it up so the reversed MOSFET turns on when in danger of forward biasing the diode (starting the engine), to keep the heat within normal bounds. Realistically, I don't think I can trust the microcontroller to do this job though, sounds dangerous since this is about the time brownouts or power spikes could be resetting it.

Oznog said:

Grrrr.... the huge ass relay is looking better and better.

Click to expand...

Remember K.I.S.S. design thinking 8)

Oznog said:

Thanks, I didn't see that note, but I knew the current was limited by thermal capabilities of the pkg.

Now if the MOSFET is used forwards, this could be up to 100 amps continuous (and diodes cannot be paralleled effectively, so another doesn't help at all). If it's used in reverse, it's only going to forward bias when the starting battery voltage is less than 1.4V of the deep cycle voltage. Unfortunately, that could happen on starting the engine, which could suck hundreds of amps through the MOSFET diode. This sucks... I could still put a forward and reverse MOSFET in series, and switch them on/off together so they'll never conduct together, but then the rdson is 2x 0.0028 ohms. At 100 amps that's 0.56V, as well as 28W per pkg. I was planning on knocking that down by using 3 or 4 MOSFETs in parallel, but now I'd need twice as many. And that 3703 isn't cheap!

Grrrr.... the huge ass relay is looking better and better.

Now that I think about it, I could set it up so the reversed MOSFET turns on when in danger of forward biasing the diode (starting the engine), to keep the heat within normal bounds. Realistically, I don't think I can trust the microcontroller to do this job though, sounds dangerous since this is about the time brownouts or power spikes could be resetting it.

Click to expand...

Note 6 (76 amp limit) applies to forward current as well as diode current. I suspect the limitation is in the slze of the leads or the internal bonding mechanism (lead frame to die). That relay is looking pretty good!

Nitpick

This is a little bit of information which may just serve to confuse people, but I think proper terminology is important.
I'm offering this because Oznog (nice name, Gonzo) and others refer to the saturation region of a MOSFET erroneously. It is just the opposite of a bipolar transistor. See **broken link removed**, and the graph below. The region to the left of the dotted line is called the linear region.
I've been designing DRAM chips for 7 years, and, because of my long previous experience with bipolars, I still have to think twice about it.

Ah... I see now. I was kind of wondering how you could put 210amps through those TO-220 leads. Why did they even bother to state it had a thermal capability that would harm the connections? Isn't the pkg limit an instantaneous limit to prevent gradual electromigration damage to the connections?

Normally you see pulse limits and long term thermal limits... here they seem to have long term thermal limits which you could never use without exceeding the impulse limits. I don't get it.

Oznog said:

Normally you see pulse limits and long term thermal limits... here they seem to have long term thermal limits which you could never use without exceeding the impulse limits. I don't get it.

Click to expand...

As a famous American President once said:

"There are lies, damn lies, statistics, and specification sheets."

Or at least he would have, if he was into Electronics 8)

Nigel Goodwin said:

Oznog said:

Normally you see pulse limits and long term thermal limits... here they seem to have long term thermal limits which you could never use without exceeding the impulse limits. I don't get it.

Click to expand...

As a famous American President once said:

"There are lies, damn lies, statistics, and specification sheets."

Or at least he would have, if he was into Electronics 8)

Click to expand...

LOL!