BATTERY-CHARGING POWER SUPPLY

Last modified:

INTRODUCTION

Scooba's battery-charging power-supply unit, PSU, is packaged in a manner similar to Roomba-Discovery's Fast-Charger unit, and supports battery charging with the battery either in, or out of the robot. Of course it uses the same, fundamental AC to DC voltage conversion technology as used for the Discovery's PSU, however, lessons learned from weaknesses in the Discovery's Fast Charger design appear to have been applied to Scooba's PSU. Those improvement should give us a more robust charging unit. Such improvements are one topic of discussion in this section. Also, a rough description of how 120VAC gets converted to the 22Vdc charging voltage is provided for those readers who are mystified by these PSUs having greater power handling capability than old-style 'wall-wart' PSUs, yet weigh so very much less. PSU construction is reviewed, and consideration is given to converting the PSU for use in countries with higher-voltage Mains power.

General

Since this PSU looks very similar to the Discovery's Fast Charger. i.e., a small box with pig-tail cables emanating from opposites ends, an overall view is not provided; however, the important 'data label' found on the base of Scooba's PSU is shown in Figure 1. From the data displayed, we see Scooba is commanding more power than Roomba -- charging current is greater by half an ampere! Input (mains) voltage is designed for the USA 120VAC power.

Figure 1. PSU's Bottom-Face Label

Notice at the top of the label, the words "Scooba Power Supply". Let us contract that phrase to "SPS" and then suffix a "U", from 'Unit', to make an abbreviation "SPSU" for use in, at least, this section -- just to save a little typing.

Going by the 'power'-data on the label, we may as well compute some unstated characteristics, such as efficiency and power being dissipated to the SPSU's enclosure (at least one or two owners might be interested!):

  1. If SPSU's power-factor, the cosine of the phase-angle between input current and voltage, happened to be 'unity', of course it is not, then input_power = 120VAC x 0.75 amp x cos(0)= 90 watts.
  2. However, the label says input_power = 45 watts; hence, we can find the power_factor by: 45W = 120 x 0.75 x cos(x); or cos(x) = 45 / (120 x 0.75) = 0.5 = power_factor.
  3. The label says output_voltage = 22Vdc, at 1.75A output_current. The (peak) output power is then P_out = 22 x 1.75 = 38.5W. This is power being delivered to the cells at full charging current.
  4. Since 45W is being supplied to the SPSU from the mains, but only 38.5W is leaving the box -- to be dissipated in the cells, (45 - 38.5)W = 6.5W remain in the box, to be conducted / radiated to its enclosure and dumped into the room. That is quite nice, the box will probably not be as hot as a Fast-Charger's housing.
  5. Efficiency may be stated as eff = 100 x 38.5 / 45 = 86%.

Just How IS that Power-Conversion Accomplished?

The SPSU uses a switch-mode type of DC-DC power-conversion technology.A brief description of a "switcher's" elements follows. Think of the description as a verbal block-diagram. There are two major 'blocks', a high-voltage section and a low-voltage section. Given the standard practice of beginning review of a circuit at its input end, we start at with the high-voltage block, then progress to the low-voltage output:

  1. HV-Block, Mains Input & Conditioning: A pig-tail, line-cord brings 120VAC mains power onto the SPSU's PWB. One side of the the line passes through a fuse, then both sides are shunted together by two excess-voltage-control devices. The first is a metal-oxide varistor, and the second, a capacitor. The AC-power then passes through a bifilar-wound inductor (at times called a Balun) to suppress high-voltage line-noise.
  2. HV-Block, Rectification and Filtering: Exiting from the Balun, the 120VAC enters a full-wave, diode-bridge rectifier, by which it is converted to high-voltage, (170V-peak) unidirectional current pulses. That train of voltage pulses is then filtered / ripple-smoothed via an electrolytic capacitor to then 'store' a slightly lower, average DC-voltage.
  3. HV-Block, Switching DC-power Through Primary Winding: The smoothed, high DC-voltage is then presented to a switching device whose job is to repetitively switch ON then OFF, to alternately pass, then halt, that high-level charge through the primary winding of a step-down transformer. This switching device is the heart of the entire DC-DC conversion process. Physically, it is an integrated circuit, packaged in a case that looks like the popular TO-220 power-transistor package; however this one has six leads coming out its bottom. This part is the "TOP246YN"[Rev.070107-1], 'Integrated Off-Line Switcher'. A visit to the manufacturer's site is recommended. Much superior information about this device, including sample DC-DC-circuits will be found at Power Integrations, Inc. Cutting away from the sales-pitch, and getting back to SPSU, the TOP246YN does that ON / OFF switching at a rate of 132kHz! Running at that very high frequency (compared to a measly 60Hz) results in cost and weight savings for the SPSU. The item within the TOP246YN doing the power-switching is a power MOSFET. However, it is the balance of the integrated circuit within that package that keeps everything under control; 'things', such as output voltage regulation (zero-amps to full-load), and output-regulation under input-line voltage-fluctuations.
  4. Mixed-Section, Voltage Step-Down: A step-down transformer partially occupies HV-space with its primary-side inputs, and LV-space with its output terminals. It is of very low weight since iron and copper masses may be greatly reduced by virtue of its high-frequency operation. The core is one of the ferrite compositions, and much less copper is needed to wind such a small transformer frame, while also meeting power requirements.
  5. Mixed-Section, Regulation Feedback: Also located in this in-between zone is the body of an optocoupler; which feeds LV-regulation control back to the TOP246YN using only an optical-connection to HV components. The HV-side of the optocoupler contains a photo-transistor that obtains its bias from a tertiary transformer winding and a rectifying diode located in the HV-section.
  6. LV-Block, Rectification & Filtering: The high-frequency, quasi-rectangular waveform out of the transformer's secondary winding is then single-wave rectified and smoothed with a filter capacitor. LV-power heads to an output Balun and to the SPSU's LV-pigtail.
  7. LV-Block, Voltage-Regulation:The smoothed DC output voltage is continuously compared to a reference voltage. Any non-zero error-signal is fed back, via the optocoupler, to the TOP246YN IC, which then adjusts the oscillation pulse width to regulate the LV-level to a well controlled 22.0Vdc.
  8. LV-Block, Over-Current Sensing: Based on finding a current-sensing function built into Discovery's Fast-Charger circuit, it is assumed that an over- current protection circuit also exists on the LV-side of the SPSU. The TOP246YN also establishes maximum output current. The need for both current-controls is not understood.
  9. LV-Block, Pilot-LED Driver: The 22Vdc is also used to power the pilot LED on the SPSU. A few more components are needed to drop the level across the LED to around 1.6Vdc, and to control its junction current.

That should do it! Actually, a bit more was said than was envisioned before writing it all. For anyone not interested in all the detail, just reading the bold, list-element headings, may be adequate!

Gaining Entry

Scooba owners in the USA should have little cause to enter the SPSU; for example, we are not faced with changing a known, under-rated component, as we were with the Fast Charger. However, those Scooba owners with utility mains-voltage higher than the SPSU was designed for, will wish to modify its circuit so it can handle, e.g., a 220/240VAC inputs. Whether that conversion is viable, or not, will be given review in a later section. No matter, there will be some owners who wish to see inside the SPSU's enclosure.

Access is via the four feet of the box. One of them is shown in Figure 2. It is similar to those used on the Fast Charger, tee-headed, rubber base-feet, friction-fitted in a screw's counter-bore, and not glued in place -- in the unit sampled -- could be loosened using a 0.375-in. O.D., thin-walled brass tube fitted over the rubber-head, then rotated, +&-, to get an enveloping grip on a foot-pad, then simultaneously lifted and side-thrusted to accomplish dislodging the foot.

Figure 2. Typical Foot, 9-mm Diameter (One of Four)

The base-screws are of the thread-cutting (has tapping flute, so be careful when putting them back, they will cut a new thread-lead more easily than the TFS-type; it would be wise to blunt those cutting teeth). (Base-screws are of size: STM3x*x11LG, #2 Phillips, pan-head. Note: screw-pitch was not measured).

High-Voltage Circuit

Once the Cover has been separated from the Base, an interior view such as that shown in Figure 3 will be available. Referring to the HV and LV-blocks, every thing to the left of the transformer's left/right middle is considered high-voltage -- dangerously high and hazardous to living creatures. All items to the right-side of the transformer are considered low-voltage. You will see two slots cut clear through the PWB on each side of the transformer -- presumably to interrupt surface-conductance.

Upon comparing this high-voltage section to Roomba's Fast-Charger's, several differences are visually obvious; and it will be some of these items that must be changed in the conversion.

  1. A 271KD10 metal-oxide varistor now shunts the Mains (placed just after the 3.15- ampere fuse). In Figure 3, this part is the blue, disc-form component located at extreme left and a little below center.
  2. High-voltage filter-cap C1 has its capacitance doubled (was 47µF), providing a 100µF part. While that may be a good change, no increase in working voltage was given -- its still 200Vdc. That is fair, but gives only a 30-volt margin for the 170Vdc level; safer, would have been a 350WV capacitor. C1 is located near the transformer's upper, left corner.
  3. A larger transformer is needed to handle the 100 * ((1.75 - 1.25) / (1.25)) = 40% increase in charging current. SPSU's transformer appears to contain approximately three times as much ferrite as used in the FC's transformer.
  4. Chip components: A few chips are used on the PWB's foil-side of this HV-section, whereas Fast-Charger's have none.

Figure 3. PSU's PWB-Assembly, All of H-V End and 2/3 of L-V End

Low-Voltage Circuit

Upon comparing this low-voltage section to Roomba's Fast-Charger's, numerous changes are noted:

  1. Rectification of the battery-charging voltage is done by a device that will handle higher power dissipation; in fact, a pair of power-diodes are used in parallel. They are packaged in a TO-220AB case, which uses three pins; thus, it is a 'common-cathode-module', which has its anodes hooked in common by a PWB-trace. Figure 4 illustrates this diode package at location "D16". The installed device is marked "LT 5221", with "MBR20100CT" on a second line. Searches for MBR20100CT turned up a match at Vishay Semiconductors.
  2. Smoothing the rectified, low-voltage, (22Vdc), level, is accomplished by a 1200µF, 50WV capacitor. Note this part's PCB-label is not the same "C4", as in the FC's circuit. In the SPSU it is now "C10".
  3. Note the very nice difference in voltage-rating of that 1200µF capacitor! Fast-Chargers were shipped, for many months, in which only a 25Vdc rated capacitor had been installed at C4 -- giving only a three-volt margin! Many failed.
  4. Three wires (green, blue, and brown, seen in Figure 4) go into the output-pig-tail, but only two are required to deliver charging power! Examination suggests the green wire carries analog feedback, i.e., a control-line, to the low-voltage-regulator section. That change could relieve Scooba of the hazardous (when sµFficient volume for components inhibits adequate power-handling) chore of throttling charging power within the robot.
  5. Chip components: About 20 chip parts are used on the foil-side of the PWB, whereas FCs have none.

Figure 4. Low-Voltage Bridge-Rectifier, D16, Handles Higher Current

PWB Has Components on Both Sides

Upon first looking over the PWB-assembly, and mentally comparing all in view with the memory of a Fast-Chargers PWB, there was the impression that the SPSU was somehow simpler than the FC's! Sure, the SPSU is a little longer, and perhaps that permitted the engineer to spread out parts a bit, but it was not until the PWB was removed from the Base, and turned over that the concept of "simplicity" vanished -- the foil side of the PWB has chip-form parts mounted in both blocks! Figure 5 shows them; the HV-section is to the right.

Figure 5. Chip-Components Populate PWB's Underside

Conversion to 220/240VAC Operation

After reviewing applicable switching-mode, power-supply design requirements and construction methods, one may come to the conclusion that operating a switcher at a higher input voltage than for which it was designed, is not as simple as substituting higher-voltage components. Limiting such 'upgrading of voltage capabilities' to only a subset of components in the HV-block will almost certainly degrade the circuit's EMI (electro-magnetic interference) suppression capability, and may very well create a safety hazard for the user's health and/or property. Conversion of the SPSU cannot be recommended. Let us examine the threats.

Arguments Against Conversion to Higher Input-Voltage Operation

Two issues will dominate, safety and EMI suppression. We will proceed as the power goes, beginning with the first capacitor across the mains, and ending with the step-down transformer; but exclude the three obvious candidates for upgrade (fuse, varistor, HV-filter-capacitor) from this particular discussion list, unless there is a 'safety' or 'EMI' impact relative to any of those three items. References will be made to specific switching power-supply application notes by their "AN-nn" identifications. These publications are all available in pdf format from Power Integrations, Inc. At the 'Power Integrations' site the key-phrase: "TOPSwitch-GX Flyback ", can prove useful when deciding which other Power Integrations application-notes should be studied.

Another qualifying statement is needed here: To lend credence to portions of the following points, or arguments, references to quoted Standards, such as a Safety Standard, will be given in the various application notes. We cannot know to what standard(s) the SPSU was designed and built (other than getting a hint from the icons of those certifying agencies that are printed on the SPSU's label), therefore such references in the ANs, are to be considered only as indications of preferred-practice in this low-power, DC-to-DC conversion field.

  1. X2 Capacitor's AC Working-Voltage Margin: On page-eight of AN-15, those readers unfamiliar with the "X" and "Y" designations of "safety" capacitors, can get educated. The (yellow, rectangular-bodied) capacitor at position "C2", is part of the SPSU's input, differential-mode, EMI-suppression network, and is classed as an "X2" type capacitor. This one has a 280VAC rating, which provides a nominal voltage-margin of 57% for the design-point, 120VAC mains voltage. If a 240VAC mains were to be applied here, the nominal margin would drop to only 14%. AN-15 (p.6) advises designing for a +6% high-side on mains voltage; that condition would worsen the threat to: 100 * (280 - (1.06 * 240)) / 280 = 9% margin. Not much! BTW, this capacitor as found in one FC assembly, was only rated at 275VAC! Could the C2 in SPSU have been misread? If so, that would move worst-case margin down to 7%. Hardly any margin at all; and temperature effects have not even been considered!
  2. Fusing-Current, Bridge-diodes' vs. Fuse-F1's:This issue tends to be off-topic because it questions the suitability of the 3.15 ampere rating of F1, as specified by the circuit-designer. That fusing current, and the i,t (current, time) characteristic of the particular fuse-design, is hereby compared to the probable failure point of a pair of bridge diodes. First note that the maximum average rectified forward-current rating is one-ampere, for the 1N4007 HV-bridge diodes. Then consider an over-current condition of five amperes input current, and decide whether the fuse opens, or diodes fail. We do that by looking at the i,t curves for a Bel Fuse MRT 3.15. The 3.15A-curve suggests that fuse, conducting 5A, will blow in about 30-seconds. Meanwhile, the bridge diodes could have handled 3600 unidirectional pulses, if they remain "diodes" during that stressful period. So, we may ask: what is being protected by this fuse? Clearly, the diodes will fail a continuous over-current condition before the fuse opens, hence, the SPSU-hardware is not being protected, and it may be that not even the output-driven charging-circuit gains any protection from an F1 rated a 3.15A. BTW, the same fuse and bridge diodes are used in the Fast Charger too! The opening line hinted that this is an off-topic issue, however, it is on-topic in the sense that this fuse may be one of the components that should be changed during a conversion to higher mains-voltage operation. The argument is to halve the stock value (if going to 240VAC mains) because the input-power required to operate the SPSU remains unchanged, so the current, when applying twice the voltage, ought to be half of that seen when using 120VAC mains. That remains to be proven. Certainly it is not the case when power is first applied to a SPSU that has been off long enough to discharge its HV-filter capacitor. That initial, high-amplitude pulse, or two, will easily be twice that seen for a 120VAC input voltage. In fact, that initial, power-up surge, may be one of the factors which forced the designer to use a 3A fuse. Determination of the proper fuse-rating, for either 120VAC or 240VAC mains, is beyond the scope of this document; but, attention is drawn to the necessity of re-engineering this fusing-current specification before any changes are made to the value.
  3. Transformer: Core-saturation: One of the many transformer design parameters is the maximum flux density of core magnetization. AN-17, (p.9) indicates the components of transformer design must be iterated to control maximum flux density within the limits of 2000 to 3000 Gauss. Of course one of the factors that determines the core's flux-density, is primary (winding) current. Holding other parameters fixed (by virtue of toying with a commercial product), the magnitude of primary-winding current is a function of the voltage applied to the primary winding. So, reassigning a transformer which has been designed to work with 170Vdc (rectified 120VAC) switched through it, to work with twice the input level (339Vdc from rectified 240VAC mains), what do you suppose happens to that carefully trimmed, 2000 - 3000 Gauss, design?
  4. Transformer: Voltage-withstand: Transformer's windings and inter-coil insulations have been based on 120VAC operation, and now you want to expose those materials to twice the intended operating voltage? In AN-15, Safety Principals, starting on page-four. Design requirements begin on page six, to cover such topics as: AN-18 (p.3) describes two construction methods for these high-frequency transformers -- "margin-construction" or "triple-insulation" wire. In a following section, (p.10) advising how to choose which construction to use, this generalization is given: "Applications requiring the lowest transformer cost but not the smallest possible transformer size can use a margin wound transformer". Then, on page three, a hint about the increased cost for higher voltage capability: "The creepage distance required between primary and secondary windings by safety regulations is typically 2.5 to 3 mm for supplies with 115 VAC input, and 5 to 6 mm for 230 VAC or universal input supplies. This creepage distance is maintained by physical barriers called margins. In most practical transformer designs, these margins are built up on each side of the bobbin using electrical tape, with the windings placed between them...".
    Now, let us view those requirements from the Company's viewpoint. If all that is desired is a 120VAC-input SPSU, but the power-supply engineer reports these options: The cost of the transformer will be 1.25X the basic-cost, if '130VAC-reinforced' design is to be built, or 2.1X for the '130/230VAC-basic' construction; forget about the 3000VAC construction! Which do you suppose the bean-counters will authorize for production? BTW, those are hypothetical cost factors.

The above should not be construed as an exhaustive list. With more diligence, additional design infringements can likely be found.

Historical Reference

Modified Roomba Fast Chargers are working OK, what's the problem? Surely there must be several tens of owners who have upgraded their FC-PSU to be powered by 220/240VAC mains power, and to the good side, there have been no reports on the boards claiming the modified FCs have done any of these awful things:

However, the lack of those events should not be construed to say such conversions will always be safe. Lack of catastrophic events does not mean the modified FC-circuit is functioning within its initial design parameters; they may be functioning nearer to the edges of those parametric ranges!

For Those Who Like to Live on the Edge

Taken block by block, the design of Scooba's PSU is essentially identical to that of the Fast Charger. Yet, there are differences in details that might give problems with the SPSU, but have not with the FC. So far as we know, just one SPSU up-conversion to 220/240VAC operation has been done, and it is reported to be working as nicely as the modified Fast-Chargers are. Without doubt there will be those who will change out the obvious set of parts as an effort to make the SPSU work in their locale.

This final section provides only a list of the installed, "obvious set of parts", including their part-numbers, and/or characteristics, along with comments about things to be cautious about. Only limited cross-referencing of component P/Ns, that 'could' be used to support higher-voltage mains operation, is offered. The tasks of confirming the baseline characteristics of the presently installed components, followed by identifying suitable replacement parts and their sources of procurement, will all be left up to the person making the changes -- a person with an adequate skill level in modern electrical engineering.

With exception of the included 'fuse', the following components, at least, must obviously be removed and then replaced by similar components having the capability to safely adapt the high-voltage section of the SPSU to work reliably with a 220 or 240VAC mains voltage:

  1. Line Fuse: In the SPSU which was examined, its F1 case marking is: "CQ MET | T3.15A 250V". This is a 3.15A time-delay fuse. There is an argument that the PSU's input-power remains unchanged as mains voltage increases, hence, a safe change would be to use half the fusing current if input voltage is doubled. That concept, as applied to a switching PS, needs to be verified prior to changing F1. For help in determining whether a fusing-current change is advocated, interested owners might find Littlefuse's document (a pdf file) Fuseology valuable.[Rev.070107-2]
  2. Varistor: On the SPSU board, the 271KD10 metal-oxide varistor, at RV1, must be replaced with a higher-voltage unit. Use of a Fuji Semiconductor VAR360V04W was suggested as a higher-voltage replacement on the newer Fast-Charger PWB-assemblies, by a reliable source in this thread at the Roomba-Review board.
  3. HV-Section Filter Capacitor: Filter capacitor C1 must be replaced with a higher-voltage capacitor. Keeping capacitance at 100 µF, working voltage should be in the order of 400V. No attempt has been made to locate such a part. Do not even think about changing to an axial-lead capacitor, the same, bi-pin "radial-pins", low-inductance, style must be retained.
  4. Line-Cord's Plug-Cap: The US (NEMA) plug cap must be cut off the line cord and discarded. Hints are given in the Power Integrations application notes about the inductance of that power-cable participating in EMI control, thus it would be wise to not change the cable's length more than necessary. The owner must then fit a regionally suitable plug-cap to the cord.

As indicated, these are only the obvious components that need attention; there are others. By accruing a greater understanding of the switch-mode power-supply design process, a greater appreciation of which components are being either over-stressed, or are being pushed out of the 'design window' to, say, reduce efficiency or to increase EMI emissions.

Revisions

If you wish to return to the main (Scooba Technical Comments) document, click Main

Otherwise you may reverse path by pressing the 'Backspace' key.

This page is currently maintained by G. Plews