The iRobot 02004 Rapid Charger

Last modified:


The Rapid Charger remains in use today, mid 2007, for re-charging newer, higher-energy content NiMH batteries, a task for which it was not specifically designed. Its continuing attraction is that it can do that job independently of the machine which expends the battery's charge to do work--i.e., a robot which has its own, similar charging capability. The Rapid Charger has become a valuable trouble-shooting resource for ailing battery and charging-system sections because it isolates a suspect battery from the robot while demonstrating that a suspect battery will or will not properly re-charge!
Looking back, now, at the Rapid Charger's brief advertisement, placed in the Roomba™ FloorVac's Owner's Manual, almost no technical data, or information about how to use the charger was printed. Of course the Rapid Charger was an accessory, that had to be special ordered. Unfortunately, whatever words of user-advice that shipped with a Rapid Charger has been lost; however, nothing significant can be recalled. It was therefore fortunate for the user that operations were essentially plug & play, then wait a couple hours for the green-charging light to extinguish. Yet, it would have been nice to know about its various 'features', such as: what meaning was indicated when its green pilot-LED started flashing! iRobot never said anything about that! And, naturally we, fringe techno-nerds, always want to know more about every item of equipment we own, therefore, an objective of this document is to provide as much technical detail about the Rapid Charger, as can be gleaned by inspection, measurement and analysis.


Circa, end of 2002 or beginning of 2003, iRobot released, for sale, a stand-alone battery-charging appliance which would re-charge one of their Roomba™ floor cleaner's batteries several times faster than the charging power supplies that plugged into the Original Roomba's charging port. The appliance was named "Rapid Charger". It requires the battery to be removed from the Roomba, so the battery may be plugged onto the top of the Rapid Charger. The following illustration shows an external view of a Rapid Charger.

Figure 1. 02004 Rapid Charger and Roomba™ APS-Battery

A Rapid Charger would re-charge a depleted (black battery, as issued with the Original-Roomba™ series robots) Ni-MH battery in about two and a half hours, whereas the predecessor charger took seven to 12 hours to accomplish the re-charge. How may the Rapid charger do that faster charging? According to the Rapid Charger's label, it delivers one ampere into a battery, while the Original-Roomba's power-supply output was only ½-ampere, yet it (plus minimal related electronics in the Roomba) took much more than twice the time a Rapid Charger would take to accomplish a re-charge.

The key ingredient has to be differences in 'charging-control methods'! In 2003 (and up to 2004, Q2), Roomba's charging control was very simple (its details are unknown, but that primitive control-circuit used only a small number of discrete parts), and did not consider temperature of the cells under charge. On the other hand, the Rapid Charger uses battery-temperature and voltage data to optimize, and keep things safe during the charging process. Here is an overview of the Rapid Charger's charging-control features:

Rapid Chargers are designed to use two modes of charging: a) high-rate (or "fast"), and b) trickle (or "maintenance") rate.

When a battery is connected, and the unit is AC-powered, high-rate charging mode begins after passing these gates:

Once high-rate charging has been initiated, it is continuous at a regulated, configured current level. There are five conditions that will interrupt the high-rate mode:

  1. Battery temperature exceeds the configured maximum.
  2. The configured time-out period for fast-charging ends.
  3. Charging completes, as determined via a peak-voltage detection process.
  4. The configured maximum battery voltage is exceeded.
  5. Power is interrupted; via either AC-mains-power, or battery-disconnect.

"Configuring" noted in (1), (2) and (4) is the result of choices which the Rapid Charger's designer made after selecting a commercially available integrated-circuit charging-controller package, that could do the sorts of checking implied by the above list. It is the use of tailored, and dynamic biasing for certain of the device's inputs that set the overall prescription. Details stemming from the controller choice, and then its configuring, will be revealed and expanded within, appropriately, the "Design Details" section, below.

Immediately following high-rate termination (other than by (5), above), the Rapid Charger's second mode of operation begins: viz., trickle-charging. Trickle charging is accomplished as the Rapid Charger's controller periodically applies a very narrow voltage-pulse to the battery. Roughly speaking, the time of voltage application is just a little less than the time taken to flash two motion-picture frames on a theater's screen. That pulse is then repeated at intervals slightly in excess of one second. Trickle-charging mode continues until power is stopped (as in (5), above).

The Rapid Charger remains to be a useful charging system, one that is totally independent of the robot and its internal charging-controller, thus retaining the Rapid charger's secondary utility as a valuable trouble-shooting resource to help sort out ailing battery vs. ailing charging systems.

This concludes the general review of the Rapid Charger. If more technical information is desired, continue into the Design Details section.


Construction Overview

From a user's viewpoint, the Rapid Charger appears as a black-blob, somewhat pear-shaped if viewed directly from above, and which has an upper-side concavity that is shaped to receive one end of a Roomba battery and a portion of the battery-case. At the upper-housing's higher end, the same configuration of electrical contacts (same as those seen in the Roomba) are visible, and it is of course there that a battery to be charged makes electrical connection to the Rapid Charger. It is also there, that the battery's spring-latch will grab hold of the Rapid Charger's upper-housing, once the battery has been fitted in place. Instructions which pertain to battery insertions and removal are printed nearby those features. There is a green (LED type) indicator on the highest portion of the upper-housing, next to the "Ro•mba" label. The final set of upper-housing features to note are the approximate three-dozen vertical grooves which are distributed around its periphery; many of which penetrate the housing's wall to provide ventilation for the electronics.

The lower portion, or housing-base, is a dish-like casting that not much taller than half an inch. Once we get inside the R-C we will see that the two major PWB's (printed wiring board assemblies) are secured to the base. Hence, it is to the base that the 120VAC line-cord is attached. The housing-base also has ventilation slots around its perimeter, and there are four, rubber foot pads.

Getting Inside the Rapid-Charger: Removal of six base-screws will release the R-C's upper-housing from its base-casting; however, two short, interconnecting cables prevent full separation until their connectors have been de-mated. In addition to those restraints, a bit of mechanical interference may be experienced as an odd patch of fishpaper (attached to the top of a heat-sink screwed to the base) catches on hardware screwed to the interior of the upper-housing. Inspection, will guide the release of that snagging. BTW, re: 'accessing base-screws', four of their six pockets are covered with the R-C's rubber foot-pads. The pads are not at all difficult to remove, since each is a simple cylindrical disc which has an adhesive coating on one side... very much like a spot of adhesive tape. The purpose in mentioning this 'adhesive' feature is to emphasize its non-drying nature, and suggest it will be prudent to protect each sticky side of those removed pads, so the sticky faces don't pick up a lot of debris while dismounted, and then refuse to adhere to the base-casting, when later replaced.

PWB-Modules / Functions

Once the upper-housing clears the housing-base, the two main PWBs will be in view. The PWB connected to the 120VAC pigtail is the R-C's dc-power supply--a switch-mode power supply (SMPS) extraordinaire!; and the smaller PWB (about 2/3rds the board-area of the SMPS) contains the charging-controller circuit.

Upon overturning the upper-housing to view its innards, two tiny PWBs will be seen fastened to its interior, and a plastic retainer-'plate' will be seen holding the battery-contacts against the housing. The larger of those tiny PCBs mounts a trio of interlock switches, while the smallest PCB simply functions as a 'terminal-strip' and mounting-bracket for the green-LED, poking its head through the housing wall. All of these PWBs are interconnected via short, de-mateable cables.

Other than identifying these assemblies, the purpose of this section is to provide details needed to gain an understanding of how the Rapid Charger goes about sampling battery-status, then re-charging it.

  1. Switch-Mode Power Supply, (SMPS) Assembly...

    The Rapid Charger's SMPS will not be detailed at this time, except to relate its construction to Power Integrations' Application Notes, and to accept it as a 24.0Vdc power source, which is capable of delivering at least one ampere of current. Note: Power Integrations is the mfr of the SMPS' integrated-circuit, power controller, and a designer of SMPS systems. It is currently unknown whether current limiting has been provided in this SMPS. This SMPS-PWB is the R-C's largest sub-assembly, and, while attached via four screws, there are seven which must be removed to release the SMPS-PWB AND attached RFI-control items from the Rapid Charger's base casting.

    At that high-voltage end of the PWB, an AC-power cord terminates; and is seen to penetrate the housing's wall via a strain-relief device, to then become available for connection to AC-mains-power. At the opposite end of the SMPS PWB, the low-voltage section, the SMPS's 24Vdc output voltage connects to two-pins of a four-pin connector. A very short, two-wire interconnect-cable is then used to interface the SMPS's output to the Charging-Controller-PWB.

    Until such time that more detail becomes available for this SMPS, readers may gain a great deal of insight into its circuit details and physical construction requirements, by reading:

    1. about the TOP224Y at TOPSwitch® Product Family page,
    2. then, in the lower portion of that page, select & down-load the "TOP221-227" data-sheet (472kbytes), to see in that document its Figure 8, "Schematic Diagram of a 20 W Universal Input TOPSwitch-II Power Supply..."; which provides an approximation to the Rapid-Charger's SMPS circuit.
    3. App-Note: Application Note 21 provides application guidelines for the TOPSwitch-II family of devices.
    4. App-Note: Application Note 15 provides reasons for various odd looking parts and construction tactics seen in SMPS units.

    Considering the example SMPS-design, 20-watts is four short of what the R-C requires, and that supply's voltage is half of the R-C's, however, those differences can be overcome by transformer changes and use of a different TOPSwitch package. While I do see more components on the low-voltage end of the SMPS-board than I see in the sample schematic, new discoveries await us regarding the R-C's SMPS' actual circuit. In the meantime, however, that example will give insight into the R-C's SMPS. Keep in mind, for the initial intent of this document all we need to know is the SMPS' output voltage, which we do, and that output current is at least one ampere--which it must be to properly support the regulated operation of the following assembly.

  2. Charging-Controller PWB Assembly

    Second in size to the SMPS, this Controller-PWB also mounts (via four screws) onto the base casting. Three individual cables connect it to the other three PWBs within the Rapid Charger (to the SMPS, to the triple safety-switch module, i.e., "Battery-Isolation Switches Assembly", and to the "Green-LED Indicator PWB"). It is the Charging-Controller that determines whether a battery should be charged, and if so, then charging power is controlled in a particular manner until the circuit determines that fast-charging should be cut back to a trickle rate. The elements of this circuit require more thorough explanation than any other Rapid Charger circuit; so, to satisfy that need, a special section for the Controller begins after completing overviews of the two smallest PWBs found in the R-C.

  3. Battery-Isolation Switches Assembly

    The Battery-Isolation Switches Assembly is composed of a small PCB (approx. one-inch square) with three each snap-action (MICRO Switch™ type-) switches soldered to it, and two each three-wire pigtail cables. One of the pigtails has a three-contact connector at its far end that mates to a connector on the Controller PWB, and the other cable terminates at the two blade-contacts, (+) & (-) battery connections, and to the temperature-sensor's side-contact button. Those three contacts (which are replicas of those seen in the Roomba robot) mount to the housing at a high up position, while the Switches-PWB fastens to the housing at a slightly lower location. A loose plastic-pad is retained in a rectangular via directly above the three switch levers, so, as a battery is fitted onto the R-C, the battery-case pushes the pad down to depress all three levers together and close the switches. This switch-bank of three single-pole / single-throw safety switches was most likely put into the R-C's design to satisfy Underwriter's Laboratory requirements--i.e., ensure no exposed electrical-contacts have voltage on them when the battery is not in place.

  4. Green-LED Indicator PWB

    Of course this is the green indicator which will be seen lit when high-rate battery charging is underway. The LED has a T-1 (3mm) envelope, and is the sole device mounted on a ½ x 1-inch PCB which mounts (two screws) along side the battery-connector. A two-wire cable with connector-interface to the Controller PWB is pigtailed off the board.

Charging-Controller PWB

To help get acquainted with the Charging-Controller PWB, I will segregate it into functional blocks, followed by discussing the role each block plays in first authorizing battery-charging, then accomplishing two sequential rates of battery-charging. An attempt will be made to discuss those block-elements in the order they come into play.

Looking at a schematic diagram of this PWB, I envision the overall circuit may be segmented into eight functional blocks. Here they are, simply listed:

  1. Low-power, +5VREG-Block
  2. Temperature-Signal Creation Block
  3. Four-Node, State-Summing Block
  4. Battery-Voltage Sensing Block
  5. Charging-Controller-IC Block
  6. Charging Control-Switch Driver Block
  7. Charging-Current-Regulator and Power-Elements Block
  8. Charging-Status Indicator Block

And here they are, overlayed on a schematic diagram of the charging-controller's circuit:

Figure 2. Charging-Controller Schematic, with Blocks Outlined

In the following sections, having like titles, the known details of each block's function / operation within the R-C will be presented. Although the 'Charging-Controller-IC' has its own section, I believe it will be beneficial to identify its inputs and outputs prior to engaging the four sections which precede it, because those sections must refer to I/O pins of that integrated-circuit. For brevity, the controller-IC will be referred to as "U2", its PCB identity.

U2 is an 8-pin, DIP, SMD-device. Three of its pins are devoted to fixed-duty (supply-voltage of "Vcc" = +5Vdc on pin-6, power-return, "Vss" = 0Vdc, on pin-4, and its Mode-Select pin "TM" ("Timer mode"), at pin-1. That pin is biased (to Vcc / 2 = +2.5Vdc, NOM.) to select U2's mid-mode. "Mode" is important, more information will be given about that later. The remaining pins are: three INPUT pins, and two OUTPUT pins. The latter five serve these purposes:

NOTE: All voltages are with reference to U2-4, Vss. The odd levels indicated above, are due to invoking data-sheet tolerances on stated nominal levels. Such calculations frequently must be based on the actual value of '+5VREG'; however, U3's data-sheet tolerance on its five-volt output (which is expected to be not worse than ±1%) was not convolved in the above estimates.

CAUTION: The above I/O introductions may not list all aspects / characteristics of U2's I/O pins. At this point, only enough information is provided to support reference to those functions in the immediately following sections. When a more complete function-description is required, additional data will be given.

Now, on to the treatment of each of the eight blocks.

Low-power, +5VREG-Block

In this block a regulated +5.0Vdc power-form is developed from the 24.0Vdc SMPS input power-form via U3, a 78L05 (TO-92 case), and with a SMD capacitor, C2 filtering the +5Vdc output.

Temperature-Signal Creation Block

The Temperature-Signal Creation Block performs two duties. As its name suggests it does develop a signal voltage which is proportional to battery temperature, but it initially functions as a 'watchman' that ensures that a reasonable representation of a Roomba-battery's temperature sensor has been connected to the battery-connector on top of the R-C's housing. If a reasonable resistance value is not detected, this block has the power to inhibit onset of charging.

I'll cover that latter topic first. +5VREG is used here to bias two similar voltage dividers whose tap-points are differentially monitored by a voltage comparator. The two dividers form a classic 'resistance-bridge', except the comparator's output is bi-level; either zero or near +24Vdc. Both upper arms of the bridge are 3.3k resistors, and one of the lower arms is fixed at 27k, while the other lower arm is a variable resistance, ranging from (usually) about 12k down to between 3k to 4k ohms. This latter resistance is, of course the battery's thermistor, temperature sensor. Using some part-names here may help keep the players straight! In the static, reference side of the bridge, R16 is the 3.3k upper resistor, R19 is the lower, 27k resistor, and their junction connects to the inverting input of the comparator. On the opposite side of the bridge, the other 3.3k ohm resistor is R8, and it connects to the BUC's "R_T_batt" (a made up name for use here, but not used on the schematic diagrams). The junction of those two connect to the comparator's non-inverting input.

The comparator is ¼th of an LM324, quad op-amp IC, and its A1 section is used in this 'watchman' circuit. The LM324, U4, has positive feedback via R5(1M) to its non-inverting input, and is powered with the 24Vdc from the SMPS, so, when U4-A1(+in) is more positive than U4-A1(-in), the output pin, U4-A1(out) goes up to about 23Vdc; and that is what happens when, e.g., the R-C is powered and no battery is connected to it; or when a battery is connected, but its thermistor resistance is greater than 27k ohms. That corresponds, roughly, to a thermistor temperature of 5°C! Not a real thermistor, or ambient temperature in which a Roomba-battery might be found! The 'watchman' will inhibit the start of charging such a 'battery'.

On the other hand, if the thermistor's resistance is less than R19's 27k-ohms, the comparator's non-inverting input will be less positive than its inverting input, and U4's output will swing down close to zero volts, a state which is passed on to U2's INHibit input as a NOT_INHibit signal.

Getting back to *battery temperature*, this Temperature-Signal Creation Block also passes the signal-voltage from the junction of R8 and R_T_batt to the TS-input of the U2 controller IC. As stated earlier, it is when this signal falls through half of U2's Vcc level (half of Vcc is nominally 2.5Vdc), that triggers over-temperature, and U2 cuts off fast-charging. That happens when the thermistor's resistance becomes equal to or less than the 3.3k-ohm value of R8.

What is that battery-temperature which halts fast-charging? It is not possible to answer that at this time, and for a couple reasons:

All of the variables are at hand to calculate a fairly accurate temperature-value and its associated error, but since we do not know the exact thermistor which is used in any Roomba battery, an accurate value will remain unknown until someone puts a bunch of actual Roomba-battery thermistors into a temperature-chamber and measures their resistances as a function of a calibrated temperature. For those reasons of uncertainty, you will have to be willing to accept an interim estimate for the actual, design-value, high-temperature cut-off point.

We may only use the thermistor functions of likely-candidate thermistors, e.g., those found in thermistor-manufacturer's data sheets. One such device, the BetaTHERM "10K3A1B" thermistor which has a resistance of 10k-ohms when at 25ºC, seems to be a fair candidate. BetaTHERM provides an on-line R-T table from which we can estimate that it is 52ºC which correlates to the Rapid Charger's nominal temperature cut-off point.

Four-Node, State-Summing Block

This so called "summing block" is simply a clever voltage-divider that accepts two inputs and provides two outputs; and, depending on what those states are, the gate to fast-charging will be closed or open, or if fast-charging has already been started, a divider-node will be driven to set the green pilot-LED ON. In the schematic, you will see four nodes of the five-element divider are lettered A, B, C, & D. Those letters are there to support this discussion, they are not marked on the PCB. Here are details about how it works with ancillary circuits and the U2 charging controller:

Battery-Voltage Sensing Block

Earlier, in the overview of the U2-controller's "BAT" input (pin-3, which 'sees' 1/11th of the battery-voltage), it was stated that this input is designed to sense terminal-voltage of the battery under charge (BUC). That is true, but U2 also evaluates voltage at the BAT-pin prior to start of charge, as well. If no battery is connected to a powered Rapid Charger the +24Vdc fed down through D1 and R13, into the R10, R11 voltage divider, will raise the voltage on the U2-3 BAT-pin to a level greater than 2.0Vdc ± 5%, resulting in U2 holding off starting fast-charging. Yet, with an acceptable battery connected to the R-C, there is a mid-level voltage zone through which V_BAT may range from 1.0 to 2.0Vdc, and which U2 interprets as acceptable, hence it authorizes fast-charging. U2 is designed to also accept its PVD calculation results when V_BAT is in that range.

Of course U2 keeps tabs on battery voltage all the while fast-charging is in process. If at any time the voltage on the BAT pin, U2-3, exceeds the maximum cell voltage, "VMCV", which is nominally two-volts at pin-3, but 22V at Roomba-battery terminals, high-rate charging is terminated.

More likely, high-rate charging is terminated when the U2-controller decides the battery has reached full charge. If battery-voltage (divided down and presented to the "BAT" pin, U2-3) droops below previously logged voltages by 2.5 ± 2.5mV, high-rate charging is terminated. This is the PVD (peak-voltage detection) function.

U2's data-sheet suggests that, when a battery of cells are to be charge-controlled, V_BAT_nom be set to V_BAT_spec = V_batt / (N-1), which would be, nominally, 14.4 / 11 = 1.3Vdc, for a Roomba battery (wherein N =12).

And one final action, which relates to testing the level of V_BAT prior to authorizing charging: When V_BAT is less than 0.88Vdc ± 20%, charging will be inhibited.

Charging-Controller-IC Block

During the course of explaining how the Rapid charger pre-tests a battery, and then sets out to charge it, practically all aspects of this Charging Controller integrated circuit have been revealed, and yet, the device has not been properly introduced! It, "U2", is the UNITRODE / Texas Instruments, part-number bq2002G, NiCd/NiMH Fast-Charge Management IC. Basically, the contents of this IC perform all the heavy thinking required to make decisions pertaining to authorizing fast-charging, terminating fast-charging, and then defining timing of trickle-charging pulses. Of course it must be fed data in prescribed formats for that "heavy thinking" to have meaning. The bq2002E/G Data Sheet provides all of those formatting requirements; so, if the reader is interested in fully understanding how this device may function, it will be necessary to Google that part-number to find a no-cost Data-Sheet down-load site.

Once again, fixed-levels on U2's pins will be addressed, followed by the probable sequence of HI or LO biasing which charge-enabling inputs undergo, and the later analog-threshold crossings which other inputs receive as charging progresses, will dictate the order of presentation. Pin names from the bq2002G data sheet will be used here.

Once started, fast-charging continues until any one of these events occur: time-out, voltage signal on the TS terminal falls below its nominal 2.5Vdc threshold, U2 invokes cut-off via PVD, or charging-power is interrupted. When fast-charging termination is due to U2's action, trickle-charging ensues. There is no time limit on the trickle-charge phase.

But What About Those Faults That Prevent Start of Fast-Charging? This subject matter, alluded to in the Introduction, is related to the panel-LED, on one hand (because it displays a visual warning), and to the controller-IC, U2, on the other, since it senses the fault(s) and responds by pulsing two of its pins. This appears to be the appropriate place to identify the faults that cause LED flashing.

After connecting a battery to the R-C, and AC-powering it, unbeknownst to its user, the R-C's controller briefly goes about its business of checking battery temperature and voltage, but holds off on starting fast-charging while doing the assessment. If the assessment yields certain out-of-range conditions for either parameter, a visible warning is switched ON to indicate something is wrong.

Those 'certain out-of-range conditions' were ultimately found in the referent bq2002E/G data-sheet, but they are obscure. While the unilateral out-of-bounds expressions are clearly stated in at least two places (DS-pp.5 & 7), the notions that U2's 'NOT_LED' and 'CC' pins will be pulsed at the trickle-charge rate, simply do NOT stick out in your face! Here are expansions regarding those findings:

  1. Battery-Temperature: Examination of the schematic clearly shows its bq2002G controller-IC, U2, will be inhibited if there is no temperature sensor connected. However, since the IC is indeed responding to a fault and causing a flashing-LED warning, then U2 is not inhibited from going to work, therefore, a temperature-sensor must be present. Hence...
    1. If that is so, then the only temperature related fault that is able to inhibit charging and flash the LED, is that of a too-high battery-temperature. Tests on a real R-C have verified its refusal to start fast-charging a hot battery.
    2. Upon power-cycle (includes battery R&R), U2 follows this high-temperature rule:
      If temperature-signal V_TS =< 0.6 * Vcc, (± 5%), the battery is too hot to start fast-charging it.
      In the R-C, Vcc = 5.0Vdc, so V_TS = 3.0Vdc, or less, on that U2-5 input. Hmmm, that level is just half a volt above the high-temperature threshold which terminates normal fast-charging; yet as expressed as a 'voltage' it has little intuitive meaning. What might the equivalent temperature be? Two steps are needed to get to that temperature-value: First, the thermistor-resistance which causes the TS-pin to receive a three-volt signal will be calculated, then the less-certain process of transforming that resistance to a temperature value will be used.
    3. The current, i_R8, though bias-resistor R8 may be expressed as: i_R8 = (5.0 - 3.0) / 3300 = 6.06E-04A; and that same current flows through the unknown R_T resistance, which has developed three-volts across it; therefore, upon substituting for i_R8 : R_T = 3.0 / i_R8 = 3.0 * 3300 / 2.0 = 4950, or about 5000 ohms.
    4. As was done earlier, I assume I know the characteristics of the nominal 10k25C thermistor in the Roomba battery, so I can look up the temperature which causes that thermistor's resistance to become 4950 ohms. Based on BetaTHERM's R-T table, the flashing warning should show when the battery is hotter than 42ºC.
    5. If the user knew about that high-temperature constraint--and what the flashing LED might mean, s/he could pay attention to the battery's warmth, while loading it onto the R-C, to determine whether it seems hot. Actually, that 'hot' condition is only 4-degrees above body temperature! Hence, the word "warm" might be more applicable! If so, the hot battery could be put in a cool place for a couple hours, to cool. If the battery does not feel hot, there is no "thermal-problem", but there may still be a "temperature-sensor-problem".
  2. Battery-Voltage: In addition to temperature, the bq2002G-IC looks at battery-voltage to decide if it is within an acceptable 'window'. Two voltage-fault types exist. Both are handled by U2, but only one flashes the LED. If the V_batt, terminal voltage, is too high, i.e., in excess of 22Vdc, i.e., 11 * 2Vdc, U2 simply inhibits itself from proceeding. But, when V_batt is too-low, U2 will follow its own voltage-rule and provide the visual warning. Testing has shown the LED to flash when voltage is too-low; whereas when battery-voltage is too high, which is an unreal situation for a battery, the LED will remain dark. The low-voltage rule follows:
    1. Upon power-cycle (includes battery R&R), U2 follows this low-voltage rule:
      If signal V_BAT =< 0.175 * Vcc, U2 will divert to its fault-mode.
      Thus, when voltage V_BAT is less than (0.175 * 5.0V = 0.88Vdc, (NOM)), the fault-mode begins. V_BAT is the signal into pin: U2-3. To reflect that signal back out to the battery-terminals, i.e., where "V_batt" may be measured by the user, we multiply by the R10:R11 division ratio of "11". Hence, V_batt_too_low = 11 * 0.88Vdc = 9.7Vdc, (NOM).
  3. Accompanying Trickle-Charge: No matter whether the LED flashes due to excessive temperature or low-battery voltage, each LED-pulse is also a trickle-charging pulse.

When a Rapid Charger flashes its LED, it is now clear that trouble-shooting the battery is what must be done. While that task is not within the scope of this document, there are obvious clues given above to guide voltage measurements and gross temperature assessment. Most R-C owners will likely be willing to take the Rapid Charger's 'word' about their duff battery, and simply go buy a replacement! Of course there is no way to determine whether flashing is caused by high-temperature, or low-voltage, without doing some instrumented tests.

Charging Control-Switch Driver-Block

All of the tests and decision making just described take place in milliseconds, or faster, to culminate in U2's CC-output shifting from a hard, LO-state to the HI-Z state. BJT transistor Q4 no longer has a place to sink its emitter current, so its collector current decays to a useless minimum. Then, R12 has nothing dragging its lower end down near ground, i.e., no appreciable current coursing through it, so the voltage (24Vdc) at its upper end becomes available at its lower end to bias Q1's gate to a high enough voltage that Q1 switches into its high-conductivity state and battery-charging begins--for real!

Q1 is an International Rectifier IRL2703, N-channel, power MOSFET which is packaged in a TO-220 case. The metal drain lug of that case is not thermally connected to a heat-sink, however, the device is mounted tombstone-style, with its case above the PCB, and is well separated from adjacent parts. Q1's case temperature was measured with a contact-thermocouple during a charging run with the R-C's upper-housing removed. At Q1's drain-tab, temperature was 55ºC.

Charging-Current-Regulator and Power-Elements Block

In between the 24Vdc SMPS charging power source, and the Q1-switch there is a stage of current-regulation. "U1" is an LM317T voltage-regulator IC that is connected as a one-ampere current regulator by dropping 1.25Vdc (NOM) across a series resistance (which is composed of R1 // R2), then feeding back the 1.25V to its ADJust-pin. The paralleled 2.4-ohm resistors net 1.2-ohms, and, therefore an I_reg = 1.25V / 1.2 ohms. While this quotient is not exactly one-ampere, its close enough for 'battery charging'!

U1 takes a bite out of the 24Vdc supply voltage, followed by another delta_V (1.25V) across the current-shunt resistance, followed by two more smaller voltage drops across Q1 and F1. Empirical data from one fast-charging run began with a 16.3Vdc level applied to the BUC, and ended with 18.1Vdc, just prior to shifting into trickle-charging. The large voltage drops across U1 and the R1//R2 pair cause them to dissipate a fair amount of power. During the already mentioned thermocouple measurements, U1's heat-sink (a robust 'Wakefield' sort of extrusion) was at 75ºC; and R1 (or R2) measured 82ºC at one of the resistor's lead-to-body junctions. Fortunately, both resistors are mounted with their bodies far above the PCB. I wonder what their temperatures might be when the upper-housing covers the base, and with the previously indicated piece of fishpaper secured to the top surface of U1's heat-sink (thus further spoiling convection)!

Some Trickle-Charging information: Trickle charging has been mentioned numerous times, but not discussed to the extent possible. This spot looks like a good place for that discussion. {Added 070901}

Trickle-charging begins as soon as U2 halts fast-charging. During trickle-charging, power application is no longer direct-current, it is a continuum of uni-polar pulses, of timed (±12%) pulse-width and repetition rate. Every 1.17 seconds, a voltage-pulse of 0.073s width, pumps current into the battery. For some (unknown) period of time after the hand-off from fast-charging, the peak amplitude of current will be the same one-ampere level as in the fast-charging period, however, when the BUC's voltage rises within a few volts of the R-C's voltage-output limit, the R-C's current regulation will starve, and thereafter a wave-train of slowly reducing voltage pulses will ensue. That train of lower-level, un-regulated peak-currents should further diminish, until trickle-charging energy equalizes with the battery's own self-discharge rate. Now, how do those variables compare with other Roomba-battery chargers? Without extensive measurements to quantify such cross-over reduction and ultimate end-point(s), I must restrict any attempt to convert this pulse train into an equivalent continuous current, to that of the high-side, i.e., the initial, one-amp peak-current regime. Therefore, equating area under the 73ms pulse to area under the 1.17s repetition-period: (1.0A * 0.073s) = (I_ave * 1.17s); and solving for I_ave, (A) = 1.0A * 0.073s/1.17s = <0.06 A>. Thus, 'I_ave' is the maximum mean-trickle-charge-level that a battery must endure while connected to a Rapid charger. Sixty milliamperes is essentially equivalent to that which the Discovery-series of Roombas apply to their batteries. {Added 070901}

Charging-Status Indicator Block

The green-LED charging-status indicator is biased with ten-volts to its anode (notice the ten-volt Zener diode, CR2). Via U4-A2 action, CR2's cathode is connected to GND to light the LED, or raised to a high positive voltage to back-bias it OFF. The inputs to U4-A2 which then determine whether the LED is lit or OFF have already been discussed.

Schematic Diagram, sans Block Graphics

For reviewers who are beyond needing the 'block'-crutches used in Figure 2, here is a copy of the charging-controller's schematic without the block-graphics:

Figure 3. Charging-Controller Schematic

I think all of the Rapid Charger's features, and what makes them work, have all been covered. Let's see if I can summarize those discussions.


I will attempt to format this summary using list-element 'bullets' to represent actions and responses which an ordinary user may take, and then experience when using the Rapid Charger to re-charge a Roomba-type battery. Three cases will be now reviewed, in which two will be 'success oriented', i.e., valid batteries and R-C unit; and, a third case will cover situations in which the battery is rejected by the Rapid Charger. Subsidiary list-elements will provide any known technical explanations related to the preceding bullet-marked action. In general, most of these technical details have never been revealed to Rapid Charger owners.

I think those three scenarios represent the only options that are open to an ordinary Roomba-battery owner. Although the Rapid Charger was designed to charge batteries having lower amp-hour ratings than ones currently in use, it is able to charge those high-capacity batteries when used properly. The Rapid Charger remains one of a Roomba-owner's best trouble-shooting tools when battery and charging system faults plague their Roomba FloorVac system(s).

"Roomba™ is a registered trademark of the iRobot Corporation.

"TOPSwitch" is a registered trademark of the Power Integrations Corporation.

Thanks go to RoombaReview-member DougInAZ for roughing out the charging-controller schematic--Spring 2007.

If you wish to return to the index, click Home

Otherwise you may reverse travel the link that brought you here by pressing the 'Backspace' key.

This page is currently maintained by G. Plews