ChargingCircuits

The info and circuits below are intended for and have only been tested with NiMh batteries of 700-750mAh capacity (C).

0. FACTS AND DESIGN CONSIDERATIONS

Battery data:

  - Top charging voltage of a battery pair:       TCV < 3.1V
  - Recommended "fast-charge" current:            FCC = C/4
  - Recommended trickle-charge current            TCC = C/20  

AgVr3 data:

  - Resistance of the "serial cable" power lines: CR ~= 2 Ohms
  - Maximum safe voltage (without batteries):     MSV < 3.6V
  - Approximate current drain:
    * suspend:                              <1mA
    * suspend, serial cable connected:       5mA
    * on, idle, serial cable connected:     50mA
    * normal work, serial cable connected: 100mA

The MSV is given by the absolute maximum ratings of the AgVr3 [MAX1676] Step-Up DC-DC Converter, which specify it as VOUT + 0.3V, where VOUT is set at 3.3V. This suggest setting MSV not higher than 3.57V.

The value of TCV depends on many things, but a main factor is the charging current. The value of 3.1V corresponds to FCC = C/4. For a pair of NiMh batteries with C = 750mAh, FCC = 188mA and TCC = 38mA. 188mA entails a voltage drop of 0.38V over the 2 Ohm power lines. For a total voltage limited to 3.57V, this leaves almost no room for regulation (3.57-(3.1+0.38)V; i.e., approximately 0.1V if you really want to reach full TCV). This strongly suggests lowering TCC to 150mA, which entails a 0.3V drop and leaves a margin of 0.2V for both voltage limiting and current regulation. This current is also enough for normal Agenda use while (slow) charging. The 38mA TCC, on the other hand, seems quite adequate for keeping the batteries fully charged, assuming the AgVr3 is suspended and connected to the serial cable (which will draw ~5mA).

It should be noted that the 0.2V margin is quite tight; mainly if the charging circuit needs to be a real (high impedance) constant-current source, even when the charging process approaches TCV. Such a top-charge high-impedance behaviour is specially important if you plan to use the ChargerController to control the charger by the flbatt program running on the AgVr3 itself. The 0.2V margin can be considerably loosened by installing the charger into the cradle. This way, the cable resistance and the corresponding 0.3V drop disappear, leading to a much wider margin of 0.5V.

To summarise, the following nominal values will be assumed in the design of the charging circuits described below:

   * Maximum voltage:       3.57V
   * Fast-charge current:   150mA (not critical)
   * Trickle-charge current: 35mA (not critical)

1. SIMPLEST CIRCUIT

This circuit, shown as Circuit 1 above, relies on using a 5V stabilised adaptor to set the reference voltages needed both for a ~160mA constant current source and for a ~3.6V voltage limiter, implemented by the PNP and NPN transistors, respectively. Unfortunately, the exact values of these voltages depend on the base-emitter diode offset voltage of the transistors which, in turn, depend on many factors, including the operating temperature. Therefore, to be on the safe side, the voltage-limiter reference voltage should be adjusted so as to get a rather low output value such as 3.4V or 3.5V. As previously discussed, this will prevent real constant-current source operation during the topping phase of the charge process. Another circuit which also had this drawback was previously posted as SimpleChargingCircuit.

The components for this circuit are not critical, but the values of R2 (or R1) and R3 (or R4) may need to be adjusted to meet the required limiting voltage and charging current, respectively. The two transistors must support at least 200mA, 1Watt and R3 should be able to dissipate at least 1/4 Watt.

2. ACCURATE VOLTAGE LIMIT

A simple manner to obtain a quite accurate voltage limit is to use the popular, low-cost LM317 IC, which is available from several manufacturers ([NS], [Motorola], [Fairchild], etc.), wired as a constant voltage source. In Circuits 2-5 it is adjusted to 3.57V. The output voltage is computed as:

     Vout = Vref·(1+R2/R1) + Iadj·R2

where Vref is an accurate internal reference voltage factory set at 1.25V and Iadj (<0.1mA) is the current needed for the feedback based regulation. Iadj and, mainly, Vref are very stable; so you can really trust on the stability of the regulated Vout against environmental variability, including a wide range of input voltages and operating temperatures. However, under our operating conditions Vin-Vout must be larger than ~2V, which imposes a minimum input voltage of ~6V. Also, to ensure regulation, the output load current should be larger than ~1mA, which imposes R1 + R2 < 3.6K. The value of R2 = 1.72K shown in the circuit can be obtained, for instance, by connecting in parallel two resistors of 1.8k and 39k (1%). In any case, it is advisable to check the resulting voltage carefully with a (trusted!) tester and modify the values of R1 and/or R2 if necessary.

A safe and rather adequate charger can be built using a stabilised wall adaptor, the LM317 wired as discussed above and a suitable series resistor, as shown in Circuit 2. With a stabilised input voltage of 6V and a (1/4Watt) 6.8 resistor you get a charging current ranging from ~160mA (with flat batts) to ~100mA (with fully charged batts), and a voltage which is strictly limited to 3.57V.

3. GOOD CONSTANT-CURRENT SOURCE

The R3=6.8 Ohm resistor in the above circuit provided for a very rough current-source operation for output voltages below 3.57V. Of course, a much better method is the one used in Circuit 1. Moreover, you can get rid of the requirement of a stabilised power adaptor by replacing the resistor bridge of the PNP transistor base with a fairly constant bias provided by two silicon diodes, as shown in Circuit 3. All the LM317-based limiting voltage issues discussed in the previous section also apply here. The charging current, on the other hand, can be adjusted by varying the value of R3 (or R4, though it is less recommended). As in Circuit 1, the PNP transistor must support at least 200mA, 1Watt and R3 should be able to dissipate 1/4Watt.

4. VERY ACCURATE CONSTANT-CURRENT SOURCE

The LM317 IC can be also easily wired to operate as a highly accurate constant current source, as shown in Circuit 4. This is perhaps the most accurate charging circuit you can build using low-cost, widely available components. The limiting voltage issues here are identical to those of the previous two sections. Also, the current can be adjusted by varying R3, which should be able to dissipate 1/4Watt.

5. A DIFFERENT PRINCIPLE OF OPERATION

All the previous circuits were more or less directly based on a "serial" combination of a constant current source and a constant voltage source. Therefore, both parts have to support all the charging current, with the corresponding power waste. Another principle is to implement only one of the two sources and limit the other by means of negative feedback.

This was the idea followed in the SimpleChargingCircuit, where the voltage developed by an approximate current source was approximately limited by a two-transistors negative feedback. Given the need of a very accurate limit voltage, a better idea is to do it the other way around: implement a precise constant-voltage source based on the LM317 IC and (approximately) limit the current by feedback, as shown in Circuit 5. This circuit is similar to the Conrad's 191418-77 charger kit discussed in BatteryChargingHowTo.

All the charging current, I, goes through the R3 resistor, which develops a voltage V = I·R3. If the circuit is connected to a low consumption device (such as a suspended Agenda without batteries), V is lower than the NPN transistor base-emitter offset (~0.6V) and the circuit provides a safe constant voltage of 3.57V. If batteries (lower that 3.57V) are connected, a high current would pass through R3 and V would grew (much) higher than 0.6V. This would switch the transistor ON, thereby switching the LM317 voltage source to (near) 0V. Before reaching this point, at I·R3 ~= 0.6V, the transistor starts working as a more or less linear negative feedback for the LM317. As a result, the charging current becomes (approximately) limited to I < 0.6V / R3.

As in the previous circuits, the current can be adjusted by varying R3, which should be able to dissipate 1/4Watt. In addition, the current can be lowered by injecting some current to the base of the (low power) NPN transistor. All the constant-voltage issues, on the other hand, are identical to those of the previous three sections.

6. SIGNALLING CHARGER STATUS

You can connect the Charging monitor LED circuit shown above to the point "L" of any of the above Circuits 1-4 to get some visual feedback of the status of the charging process. When the charger is powered, but it is not connected to the (Agenda) batts, the LED glows bright. If you connect low or flat batts, the LED dim. Then, as charging goes on and batt voltage grows, LED brightness grows correspondingly. This method does not work for Circuit 5, which needs an additional transistor for good signalling, as shown in Circuit 5a, below.

7. SWITCHING TO TRICKLE CHARGE

It may be useful to be able to switch the charger from normal ("fast") charging operation to trickle charge. You can do it by replacing R4 in Circuits 1,3 with any of the ''Trickle charge switches'' shown in the figure below. The value of R5 should be 180 Ohm for Circuit 1 and 12k for Circuit 3. The switch on the left is for manual switching, while the right one is intended for automatic operation under control of the ChargerController driven by the flbatt program running on the AgVr3 itself. The manual switch should be normally closed; i.e., open for trickle charge. The electronic switch needs input HIGH or LOW voltages for normal or trickle charge, respectively.

Circuit 4 also allows simple (though not very accurate) manual or automatic trickle switching, as shown in Circuit 4a, below. Circuit 5 needs yet a different method, shown in Circuit 5a. Both of these circuits need input LOW (0V) or HIGH (+6V) voltages for normal or trickle charge, respectively.

8. PROS AND CONS OF THE DIFFERENT CIRCUITS

Circuit 1 is the simplest and also the smallest charging circuit. An additional distinctive advantage is that, thanks to the base-emitter diode of the NPN transistor, it will draw no current at all if connected to the (Agenda) batts without being powered (all the other circuits would draw 1mA or more). Another advantage of this circuit is its very low power waste, which makes heat sinks on the transistors unnecessary. If you are not paranoid about the (remote, but high-cost) over-voltage risk, this is the circuit I would definitely recommend. However, due to its lack of high impedance constant-current source behaviour when approaching the voltage limit, this circuit could be problematic for its use with the ChargerController driven by flbatt.

Circuits 2-5 are essentially identical as to their highly accurate limiting voltage, but they are of increasing size and operating temperature. The difference between the voltage provided by the wall adaptor and the batts must be absorbed by the LM317 IC(s) and other components such as the R3 and/or the PNP transistor. If powered by their minimum working voltage, Circuits 2-3 and Circuit 5 do not need heat-sinks (The LM317 of Circuit 3 becomes fairly hot, though). Circuit 4 definitely needs heat sinks for both its LM317 IC's.

Consequently, while Circuit 4 is the preferred one for its high overall accuracy, it may not be the best choice if size and operating temperature are taken into account. If small size and low temperature are important, my definite recommendation is Circuit 5, powered by a wall adapter providing 6V (stabilised or minimum).