Reverse-polarity protection comparison: diode vs. PFET vs. a smart diode solution

When you fly on an airline, you’ll occasionally hear “Thank you for flying with X; we know you have many choices when you fly.” Similarly, engineers have many choices when it comes to picking reverse-polarity solutions. Some of the choices include diode, P-channel field effect transistor (PFET), and TI’s LM74610-Q1 plus N-channel field effect transistor (NFET) (called a smart diode solution). In this post, I’ll highlight some key aspects of all three solutions with regards to automotive applications.

I will pick a couple of application specific parameters for comparison purposes: thermals, dynamic reverse polarity, voltage interruption, continuous power-line disturbance and quiescent current (Iq), and leakage current.

1) Thermals

Thermals are usually the deciding factor when choosing between a diode solution and a FET solution. Figure 1 shows the board I used to compare. The smart diode is the “coolest” solution and shows the lowest temperature rise seen in Figure 2. The temperature rise of PFETs is much higher if the input voltage drops to 6V (similar to start-stop voltages), since the Rdson increases dramatically at lower input voltages. Of course, take one look at the diode temperature and you’ll say “ouch.”

 Figure 1: Board used for comparison of thermal

Figure 1: Board used for comparison of thermal

 

 Figure 2: (Schottky) Thermal measurements with 10A current and VIN = 6V

 Figure 2: (PFET) Thermal measurements with 10A current and VIN = 6V

 Figure 2: (LM74610-Q1 + NFET) Thermal measurements with 10A current and VIN = 6V

Figure 2: Thermal measurements with 10A current and VIN = 6V

2) Dynamic Reverse-Polarity

The dynamic reverse-polarity requirement comes primarily from standards testing requirements such as ISO7637 and OEM-specific requirements. You can use diodes, PFETs and NFETs but their performance is related to other component values. Figure 3 illustrates how diodes and a smart diode solution are very similar in performance, whereas with the PFET solution the voltage goes negative (depending on the output capacitor used). Due to the slow response time, you will need bigger output capacitors, as shown in Figure 4.

 

 Figure 3: (PFET) Dynamic input voltage reversal, 12V to -20V, Cout = 4.7µF, Io = 0.1A

 Figure 3: (Schottky) Dynamic input voltage reversal, 12V to -20V, Cout = 4.7µF, Io = 0.1A

 Figure 3: (LM74610-Q1 + NFET) Dynamic input voltage reversal, 12V to -20V, Cout = 4.7µF, Io = 0.1A

Figure 3: Dynamic input voltage reversal, 12V to -20V, Cout = 4.7µF, Io = 0.1A

 

 Figure 4: Dynamic input-voltage reversal, 12V to -20V, Cout = 2200µF, Io = 0.1A

Figure 4: Dynamic input-voltage reversal, 12V to -20V, Cout = 2200µF, Io = 0.1A

 

3) Voltage Interruption

The voltage interruption requirement comes specifically from OEM specifications and is meant to simulate loose contact at the battery terminal or module wiring. As shown in Figure 5, diodes and the smart diode solution are very similar in performance, whereas with the PFET solution the voltage dips very low. The name of the game is to keep the output voltage high for small voltage interruptions. In the case of interruptions, reverse current flows from the output capacitors back into the input. In the case of diodes and the smart diode solution, the reverse current is blocked. The PFETs allow reverse current back and hence deplete the output capacitor. In practice, this means less hold-up time for an interrupt event.

 Figure 5: (Schottky) Voltage interruption, 12V to 0V, T = 2ms, Cout = 100µF, Io = 0.1A

 Figure 5: (PFET) Voltage interruption, 12V to 0V, T = 2ms, Cout = 100µF, Io = 0.1A

 Figure 5: (LM74610-Q1 + NFET) Voltage interruption, 12V to 0V, T = 2ms, Cout = 100µF, Io = 0.1A

Figure 5: Voltage interruption, 12V to 0V, T = 2ms, Cout = 100µF, Io = 0.1A

 

4) Power Line Disturbance/Superimposed AC Voltage

The continuous power-line disturbance requirement comes from the fact that alternators create noise/AC ripple on DC voltages. The AC ripple at different frequencies can cause high ripple current if reverse currents are not blocked. The repercussions of high ripple current are lower reliability for the electrolytic capacitors and higher temperature rise. Figure 6 shows the difference in ripple currents: 5KHz for a diode, 7A for the smart diode solution and 25A for the PFET.

 

 Figure 6: (Schottky) Superimposed AC voltage, 2V peak to peak at 5KHz and Io = 3A

 Figure 6: (PFET) Superimposed AC voltage, 2V peak to peak at 5KHz and Io = 3A

 Figure 6: (LM74610-Q1 + NFET) Superimposed AC voltage, 2V peak to peak at 5KHz and Io = 3A

Figure 6: Superimposed AC voltage, 2V peak to peak at 5KHz and Io = 3A

 

5) Quiescent Current

Quiescent current is a big, big deal for automotive applications. A complete module is allowed a maximum 100µA for standby currents (Iq). The smart diode solution mimics a diode with zero Iq. With a PFET solution, the Iq can be higher than 100µA unless you use additional circuitry to turn off the gate-drive bias.

6) Leakage Current

Leakage currents are often overlooked in most single input applications. How it plays a role in real-world scenarios is sometimes confusing. Let’s take an ORing application with two diodes. If one of the inputs turns off, you would expect the voltage at this point to be zero. Due to some leakage current, however, the voltage from the output leaks back into the input and you see some voltage there. If you draw more current from the input than the leakage spec, the voltage will drop to zero. Schottky diodes have low leakage currents at 25°C but go much higher at high temperatures. For example, a MBR1545 diode has 1.5µA of leakage at 25°C and 1mA at 125°C. PFET solutions have infinite leakage, since they allow current to flow back into the input if the FET is not turned off. In comparison, the smart diode solution has leakage of 60µA at 25°C and 110µA at 125°C.

After making all of these comparisons, it can be seen that the smart diode solution (a LM74610-Q1 plus NFET) is superior to both diodes and PFET-based schemes used in reverse-polarity designs for automotive applications. As I had mentioned in the beginning, when you are ready to dive into your next design of a reverse polarity solution, I hope you will do a test drive of the smart diode solution considering the various advantages it brings to the table.

  • Hi, I am confused about the leakage currents of smart diode. According to LM74610-Q1 datasheet, it is a "zero Iq" smart diode controller. But here, you say the leakage current is 60uA at 25°C. Can you explain more about this?

  • Hi Robert,

    Zero iq refers to current from Vin to GND. Reverse leakage is the current that flows from the output back into input. Consider the case you apply voltage to the output of smart diode and there is no voltage at the input (typical of ORing scenario) then you will have 60uA of current flowing from output to input(not to ground). For comparison purposes a schottky diode can have the reverse leakage current of 1mA at high temperatures.

    Mathew

  • Hi,

    The main, and possibly only, advantage of this over a P-FET (with a 10k resistor & 18V zener on the gate) is that there is no reverse current.  This can be fixed for the P-FET circuit by adding a '3904 and a 2N7002 FET.   The entire circuit using a 100A, 6 mOhm 5x6mm SMT P-FET cost less than $0.50.  (without the anti-reverse current parts < $0.45). This circuit has been tested at 3 different labs using eMtest & Tesseq transient test equipment with a source impedance of 0.5 Ohms, to ISO and numerous vehicle manufacturers' standards.  The equipment using this draws up to 30A in normal operation.

    How much does the proposed solution described here cost at 10k quantities?

  • Imre,

    One of the other key advantage is Iq,  a PFET solution has a higher Iq compared to the 0 Iq with the smart diode. 30A and a PFET is the ideal candidate for replacing with a lower cost NFET. If Iq is not a concern and you are able to build a robust solution with 2 transistors for reverse current detection (keep in mind its not reverse voltage but reverse current and the speed of detection and turn OFF that the smart diode offers) then it may not make sense to replace your solution at these volume levels. If your system has high input voltages >8V then its okay but if the voltage can drop low then the PFet Rdson can be quite high. Please feel free to contact me for further details at mathew.jacob@ti.com.

  • I may be wrong, but isn't the "Voltage Interruption" test incorret? From Figure 5, the test performed seems to be simulating an input short circuit condition, not a voltage interruption in the sense of a loose connection to the battery. In a voltage interruption condition, the impedance in parallel with the input is very high, and it is unable to deplete the output capacitor so fast.

  • Regarding voltage interruption test. The test case shown here is when there is a generator connected which has a low impedance. However you are right in the classic case of voltage interruption when there is a loose connection the impedance is high and there will be no reverse current. This would be a useful test to add also. Thanks for pointing this out.

    Mathew