H-bridge with two logic-level GPIOs drawing high current

Question

I am working on controlling a bi-stable latching solenoid with an H-bridge.

The microcontroller that I am using for my project doesn't have enough GPIOs for me to use four pins for the H-bridge, so I designed a circuit with PNP and NPN transistors:

I tested the circuit in simulation before making my board and it appeared to work. Setting H1B to high set the direction of the voltage correctly and vice versa.

I built my PCB and discovered that when 24 V is applied to the power rail, my power supply goes into constant current mode (even when H1D and H1B are both at 0 V). I added some control to the H1D and H1B lines but can't get either of them to close Q13 or Q10.

What could be causing this to short with H1D and H1B both low? Do you see any issues with this circuit?

Latest Simulation

I rebuilt the model in LTspice to get more accurate component models. I am seeing some strange behavior. The solenoid+ is seeing 12 V before any FETs are opened. How would that happen? I also moved the PNP resistors to below the NPN so that they don't pull any current and activate PNP. Changing the diodes to VS-E5TH1506 fast-recovery diodes helped bring the end voltage back to 12 V.

Answer 1

With only two more transistors (eight vs six in your circuit), only one GPIO pin is needed to drive the solenoid. The GPIO pin has three states: INPUT, OUTPUT HIGH, OUTPUT LOW.

• INPUT: solenoid off, 150μW power consumption.
• OUTPUT HIGH: solenoid +
• OUTPUT LOW: solenoid -

The firmware has to be written such that direct transitions between the OUTPUT states are not possible, but rather the INPUT state is always transitioned.

^{simulate this circuit – Schematic created using CircuitLab}

Q1 and Q2 are current switches that steer GPIO pin current through either R6 (OUTPUT LOW) or R7 (OUTPUT HIGH). Q3 and Q6 get turned on in the respective states. R8 and R11 drive the bridge switches Q4 and Q5 during the OUTPUT LOW state. R12 and R15 drive the bridge switches Q7 and Q8 during OUTPUT HIGH.

The flyback diodes can be back-to-back Zeners as shown, or switching diodes to rails as in your original design.

The solenoid can be driven with about 20-30mA of current. As shown, the driver thus supports 0.5W coil power rating. For higher coil power ratings, the resistors R8..R15 need to be adjusted.

Answer 2

I tested the circuit in simulation before making my board and it appeared to work. Setting H1B to high set the direction of the voltage correctly and vice versa.

No, your circuit in the simulator does not work correctly. It appeared to work because you used the default diodes with a lot of leakage current. I suggest replacing them with 1N4004 diodes. Another bigger problem in your simulation is the use of a voltmeter in place of the solenoid. This makes the Solenoid+ and Solenoid- terminals open. To remedy that, put an inductor to represent the solenoid (see the image below).

What could be causing this to short with H1D and H1B both low? Do you see any issues with this circuit?

You missed the base resistors on both PNP transistors. Without them, H1A and H1C are shorted to GND when both bottom-most NPN transistors are turned ON, as you have observed.

Here is the amended circuit.

Edit In your actual circuit, please avoid using 1N4004 diodes due to their slow reverse recovery time. Instead, consider using diodes like the ES1A. I suggested 1N4004 above because the Falstad simulator does not have many other options.

Adding a pull-up resistor at the base of each PNP transistor, as suggested by Vir (and highlighted by RussellH), can slightly improve turn-off time. This is due to faster removal of excess charge carriers in the base region through the resistor when the PNP transistor is turned OFF. However, it’s acceptable not to add the resistor if you’re using NPN like the 2N3904 and 2N2222 for those bottom most transistors (left and right). These NPN transistors have very low collector leakage current (5nA to 50nA). The resistor is needed if the collector leakage current is significant; otherwise, the PNP transistor might partially turn ON even when the NPN transistor is OFF.

Regarding the solenoid, I didn’t include coil resistance in my previous suggestion because my intention was to identify the issues and provide minimal adjustments to get the simulation running correctly.