In numbers, there are probably more chargers around for NiMH than for Lithium-ion or Lithium-polymer batteries. If you wanted to use the latter you’d have to integrate the charger circuit, making the ‘battery pack’ more expensive and complex. Keeping the batteries separate from the enclosure still leaves the option to use Liion or LiPo. Luckily their initial voltage (3.6–3.7 volts) is almost the same as three NiMH batteries in series. Also, by using a separate battery holder, you’re able to exchange discharged batteries with fresh ones without having to charge first or open the enclosure. That’s a big plus if your phone, tablet or e-gizmo is in serious need of charging and you’re in the middle of nowhere.
How many batteries to use?
The circuit has to produce 5 V and be able to deliver up to 1 A of output current. Four freshly charged NiMH batteries can have a voltage well above 5 V, so it seems prudent to keep the number at three. However, in USB speak ‘5 V’ is nominal, the actual range being 4.35 V to 5.40 V. Although that’s in favor of four batteries again, we still wish to produce a 5.00 V supply that’s accurate, if only because some designers use the USB voltage as a reference (keep that limited to non-critical applications). Three.
Boost converter TPS61030
The lower voltage of three batteries implies the use of a smaller battery holder but also the need for a boost converter. There’s an excellent device available from Texas Instruments, the TPS61030. It’s a synchronous boost converter with an internal 4-amp switch and an efficiency of 96 % (dependent on input voltage and output current of course). The converter also has an (optional) Low Battery Comparator to prevent deep discharging of the batteries. An extra undervoltage lockout (1.6 V) prevents the converter from malfunctioning. The internal reference voltage is 0.5 V, making it easy to calculate the voltage divider for the correct output voltage. Here 1.8 MΩ is used for R3, and 200 kΩ for R4. According to the datasheet, only if R4 is significantly lower than 200 kΩ then an extra capacitor for stability is necessary in parallel with R3. Here a 10 pF cap is used just to be sure. Resistor R2 should be low enough to eliminate the input current of the comparator (about 10 nA). A value of 500 kΩ is recommended. The comparator level is around 500 mV with a hysteresis of 10 mV. A threshold of 1.1 V was chosen to define one fully drained battery. Values of 1.8 MΩ for R1 and 330 kΩ for R2 result in a theoretical threshold of 3.23 V. If the total battery voltage drops below this threshold the output of the comparator goes Low (LBO). This output is used to disable the output circuit.
The decoupling of the input voltage by C1, C2 is in accordance with the recommendations in the datasheet. The decoupling of the output voltage depends on the maximum output ripple. A few millivolts is ideal, but the ESR of the capacitors in particular, as well as the board layout will result in a higher value in practice. Theoretically the ripple should be around 1 mV with an output buffer capacitor of 220 µF. In practice about 60 mV was measured across C5 (3.50 V in, and 1 A load). C5 has a rated ESR of 20 mΩ at 100 kHz. The switching frequency of 600 kHz is a lot higher, and the higher switching current accounts for the higher ripple voltage. To suppress switching noises a ferrite bead (L2) is placed in series with the output circuit. This way the ripple voltage is also reduced. The final output capacitor (C8) reduces the ripple voltage even further. For the calculation of the inductor, a change of 10 % of the maximum average inductor current was taken into account. At 3.20 V the average inductor current is close to 2 A. Given de formula in the datasheet (SLUS534E), this gives an inductor value of about 10 µH. The Sync pin can be used to operate the converter in different modes. We selected Power Save by connecting Sync to the ground, which improves efficiency at light loads (the device then operates discontinuously). The converter only operates when the output voltage drops below a set threshold. On the downside, the output ripple voltage increases slightly. With no load, an 80-mV sawtooth was noticed with a 150-ms period. But it got better rapidly with increased loading.
TPS2511: glue for USB:
A special IC type TPS2511 is used for controlling the output. Texas Instruments calls it a USB Dedicated Charging Port Controller and Current Limiting Power Switch but we still liked it. Here’s why. It’s often not enough to just put 5 V on a USB connector and get a device to work.
It’s great that phone & gizmo manufacturers increasingly fit their devices with USB connectors as the charge port, but chargers are unlikely to be compatible between fruit and non-fruit platforms, and different manufacturers. For example, some devices expect specific voltages on the data lines, or simply a connection (resistor) between the data lines to recognize a charger (Dedicated Charger Port or DCP). The TPS2511 supports three of the most common protocols:
• USB Battery Charging Specification, Revision 1.2 (BC1.2);
• Chinese Telecommunications Industry Standard YD/T 1591-2009;
• Divider Mode.
For an exhaustive description of all possibilities of the TPS2511, please refer to the Texan datasheet (SLUSB18). The TPS61030 can deliver 2 A at a battery voltage of 3.3 V, and the TPS2511 can handle this current also. But at 1 A output current and 3.33 V input voltage the converter already draws 1.7 A from the batteries. At 2 amps output current this will be more than doubled, because of the higher losses. Also, the battery capacity drops at higher output currents. That’s why the TPS2511 is connected to work as a 5-watt charger. Its DP pin is connected to the D– line, and the DM pin to the D+ line of the USB connector. The current limit is set marginally higher than needed (R6 = 47 kΩ), preventing the TPS2511 from premature output voltage limiting. The Current Sensing Report pin is not used in the expected way. Instead of compensating voltage loss by changing the feedback of the converter (not really necessary at 1 A maximum output current) the pin is used to drive an LED (D1). When D1 lights up you know that more than half of the maximum output current is being drawn. The LED current is a little over 1 mA. As already mentioned, the Low Battery Comparator output drives the EN (Enable) pin of the TPS2511. This way the output voltage is cut off in case the batteries are flat. R5 is needed because the comparator output is in high impedance state when not active.
Polarity and overvoltage guard circuits:
The battery pack connection to the PCB is by way of a screw header (0.15’’ lead spacing). So in practice it’s possible for the batteries to be connected the wrong way around. To prevent damage to the circuit and still have virtually no losses when properly connected, a small n-channel power MOSFET (T1) is used, purposely the wrong way around. When connecting the batteries with the proper polarity, the body diode is in the conducting direction, and the MOSFET is fully turned on, its gate positive with respect to the source through R12. There’s no problem with the current flowing from source to drain. In case the batteries are connected the wrong way around the gate is negative and the MOSFET is turned off and the body diode effectively blocks the battery voltage. The maximum permissible gate voltage of the MOSFET used is 12 V, which also constitutes the maximum voltage the circuit will survive. At 1.7 A input current the MOSFET(s) drop a minuscule 23 mV (measured on the prototype). To avoid having to use an expensive heavy duty on-off switch, the overvoltage protection is combined with a smaller—hence much cheaper— switch. The overvoltage protection is kept simple. When the supplied voltage is too high a zener diode (D2) is used to switch on an n-p-n transistor (T2) which in turn cuts off the gate voltage of MOSFET T3, which is connected as you would expect. The 5.1-V zener diode already conducts below the specified zener voltage. At a battery voltage of 3.60 V the current through D2 is about 12 µA. At 4.25 V, it’s over 30 µA. This can easily be measured across R9, which prevents the current through the zener diode from snowballing when the input voltage exceeds 5.70 V or so. In case the overvoltage protection acts too early (due to possible tolerance of the zener diode), feel free to adapt R10, remembering that a lower value gives a higher threshold. The overvoltage protection is needed in case an AC power adaptor—hopefully set to less than 12 V out—or a 9-V battery is connected. The TPS61030 can withstand 7.00 V (absolute maximum, 5.50 V recommended). The problem with the boost converter is that the output voltage rises when the input voltage exceeds the regulated output voltage (here, 5.00 V nominal).
The PCB is specifically designed for a Hammond Manufacturing enclosure (see parts list). It’s cheap and easy to adapt to our application. The PCB is fixed with four self-tapping screws, and the top and bottom halves with two longer ones. The front and back are separate panels. In one panel only, three holes have to be drilled. The holes for the USB connector and switch should be aligned with the parts on the PCB. The same is true for two holes for the LEDs in the top cover. The exact placement of the hole for the two wires to the external battery holder is not that critical—there’s a large margin to play with. It can be located anywhere else for that matter. A power jack may also be used—there’s more than enough room in the other panel. Avoid any extra contact resistance where possible, as it will reduce the efficiency of the device as a whole. The holes for fixing the PCB are also used to connect the board’s top and bottom power planes. Be aware that the hole next to screw header K2 is not connected to ground but the net between T1 and T3. Connecting this net to ground will not cause any damage but simply turn on the circuit. The other three holes are connected to ground, however, the hole next to IC2 is specifically output ground. It’s assumed the PCB is placed in the above mentioned hard plastic (ABS) enclosure. Finally, do not touch junction R3/C3/R4 with the circuit in operation. This is a high impedance point and any hum introduced here may destroy IC1.