Circuit protection modules using mosfets

[symmetal]

I've made several modules to help protect circuits from power supply problems, to see how well they perform. They all use mosfets with a very low 'on' resistance, to minimize their impact on being used.

1: Over voltage, under voltage and reverse polarity protection

This uses the LTC4365 IC which is only available in a small surface mount package. I first used surface mount resistors and mosfets which made a small module but found the mosfets got a bit warm over 5A as they tend to need thermal vias through the board for power dissipation which is tricky for home made boards. As size isn't a big consideration I used standard components where possible. I used 2 resistors in series for each voltage threshold to enable a more accurate voltage to be set. With the values shown if the power supply drops below 10.5V or over 15.0V the Mosfets will turn off isolating the power supply. The circuit will withstand a power supply going from -40V to +60V and only pass the voltage if it's between the threshold values. There's around 1V hysteresis after shutting down so an overvoltage trip at 15V won't reset until the input volts drops to around 14V. This avoids it rapidly switching on and off, if the volts hovers around 15V.

If polarity protection isn't required then only mosfet Q1 is needed, and Q2 can be replaced with a shorting wire link.

My test load is a 1.5R 100W resistor fixed to a thick aluminium plate which is why 8.8A is the test current (from a 13.8V supply). I soldered wires on the current carrying tracks to reduce their resistance but they aren't really necessary below 8A, just a thick layer of solder would do. The mosfets just act as a low value resistor so the voltage drop depends on the current. Without the wires the total resistance was 0.015R compared to 0.010 with them. The heat sinks on the mosfets aren't necessary up to 15A or so but added them as I had them. As it is at 8.8A everything stays cold as the whole module only dissipates 0.8 watts. The tiny LTC4365 is on the back of the board and I thought I'd have trouble hand soldering it but I've made around six modules using ICs this size and they've all worked. 😀

🙂

I'll add the other modules I've made at a later time. 😀

Alan

[rl]

This is a very good chip that I have used in other applications. Analog Devices do a demonstration board which saves you some of the soldering problems...part number DC1555C available from Mouser and Digi-Key for about 25 quid. It's pricey (especially compared to a simple diode), but not as pricey as trashing your mount, and it copes with a lot more failure modes. . Don't bother omitting the extra FETs....they're so cheap it makes no sense. 

https://www.analog.com/media/en/technical-documentation/eval-board-schematic/dc1555c-2-sch.pdf

Excellent suggestion from symmetal...I've been working with these for months and the astro application never occurred to me!

[symmetal]

I chose the mosfet from looking through the Farnell online catalogue for one with a low on resistance in a TO-220 package which are easy to work with and went for the PSMN2R5-60PL as it has only 0.002 ohm on resistance and 60V breakdown voltage. Its transfer characteristic shows it's fully turned on at 3.3V gate-source. I'm amazed by the current ratings of these mosfets, max drain current of 150A with a momentary peak of 1002A. 😬 The drain current flows through the source pin and I wouldn't like to put 150A through the small pins. Soldering 150A wires to them would be a bit of a challange. 😁

I used the metal tab as the drain connection and used a stainless steel M3 nut and bolt for it. On checking it said that stainless steel needs a strong acid flux to solder but I used standard tin/lead solder with rosin flux and the nut soldered fine.

Here's the Kicad board layout

Alan

Edited August 18, 2020 by symmetal

[rl]

They are certainly very impressive, driven largely by the automotive market. But the specs need to be treated with a bit of a pinch of salt.....150 amps at 0.002 ohms is still 45 watts of heat which needs decent cooling. The specs often apply with the case artificially force-cooled to 25C. The on resistance usually doubles as the die heats up towards its maximum temperature which makes the heat sink even more important. And clean switching on the gate is important, or the instantaneous power will exceed the pulse rating. But for more civilised currents they are indeed fantastic components. 

[symmetal]

2: Reverse polarity protection using P-Channel Mosfet

This is the simplest method which avoids the the voltage drop, and associated heat dissipation requirement of using a diode particularly at higher currents. P-Channel mosfets are generally not as efficient as N-Channel mosfets and the range of P-Channel devices are more limited. Recent developments for the automotive industry have created low on-resistance P-Channel ones so here's an example. The heat sink isn't really needed at currents below 15A or so but I have a box full of them to use up. 😁

The circuit's very simple.  IPP120P04P4L-03 data sheet. P-channel devices seem to be less tolerant of higher gate-source voltages than N-channel devices so the zener diode protects the mosfet, especially when the supply is reversed. Maximum drain-source voltage 40V, max gate-source voltage limits +5 and -16V, on resistance 0.0031 ohms, max current 120A (with suitable cooling 😉). The resistance I measured on my module includes the track and connector resistances. As before, the mosfet when turned on acts as a resistor so the voltage drop depends on the current.

 

Alan

[symmetal]

3: 'Perfect' Diode for polarity protection and/or connecting supplies in parallel

This uses the LM74610-Q1 IC to make a N-Channel mosfet behave as a diode without the forward voltage drop. The IC has no Gnd reference so the module appears as a two terminal device labelled as anode and cathode. This also enables it to be used to OR power supply outputs together such as a back-up power supply.

It's a small surface mount IC only and only needs the IC, a N-channel mosfet and a capacitor to function. I used the same mosfet as before and the heat sink isn't required at currents less than around 15A. I used 2 pole connectors as they are mechanically stronger than a single pole if just soldered in.

I put the 'perfect' in commas as it's perfect for 98% of the time and is a standard diode for the other 2%. The capacitor determines the duration of this cycle. Using 2.2uF it's perfect for about 2 secs then a standard diode for the next 40mS. It then repeats this cycle. This means the output voltage has 40ms 0.6V or so dips, around every 2 secs. For most uses this is not a problem. It's the momentary 0.6V standard dip across the IC which powers it for the next 2 seconds. It generates around 6V to the gate to turn the mosfet on. When this voltage drops to around 5V it turns the mosfet off and repeats the cycle. Mosfets have a body diode in their design shown as the diode in the symbol and this body diode is the diode which supplies the load during the 40mS periods. As it's used for such short periods in total this body diode can cope with high currents without overheating.

Here's an application using 2 of these modules to have a back up 12V battery alongside a 13.8V supply. With a charge resistor across the battery 'diode' module, the 13.8V PSU will act as a float charger and charge the battery up to 13.8V and no more so the battery can't be overcharged. If schottky diodes are used instead the battery can only charge up to around 13.4V so can't be fully charged this way. Using a 20 ohm resistor a discharged battery at 11V will start charging at only 140mA and reduce to zero once it reaches 13.8V. A 20 ohm resistor only dissipates a maximum of 0.4W so a small 2W resistor will be fine, and it can be connected across the spare anode/cathode terminals of the battery 'diode'. Don't use a totally dead lead acid battery as the resistor will dissipate 9.5W. 🙂

Here's a dual module I used using the above circuit but 2 single modules are just as good.

Alan

Edited August 19, 2020 by symmetal