After disassembling my retired Philips Sonicare (HX6530) enough to access its internal circuitry, I paused to take a look at it under my oscilloscope. After completing that detour, the oscilloscope probes were put away, and the mechanical tools came back out to resume disassembly.

First to be freed was the circuit board. In addition to a few plastic clips (and one melted plastic rivet) it was held with six soldered connections: Two wires for the actuator coil, two for the battery, and two for the charging coil.

From left to the right, here are the components I recognized on the board:

• A single green surface-mount LED, the only user visual feedback on this device
• A single button, the only user input.
• Six large pads. There’s solder on Rx and Tx because I wanted to see if it was transmitting serial data. (It did not.) Two of the pads are labeled Gnd and they both electrically connected to ground. The remaining two pads are labeled Vdd and Vpp, and I don’t understand how they differ. They both seem to show the current battery voltage.
• Two resistors R1 and R3, then the two Alpha & Omega field effect transistors Q1 and Q2.
• Two more resistors R5 and R4, than the two pads leading to actuator coil.
• A through-hole for the battery positive terminal, leading to F1 and an unoccupied pair of pads. The component sitting at F1 has a small “P” printed on it and it doesn’t look like a resistor. Thinking of electronic things that start with “F” and would make sense at the battery terminal, I think this is a safety fuse. The unoccupied pair of pads has a thin trace connecting them. I guess this is a provision to put something in line with battery power: we can cut that trace then solder something on those pads.
• U1 is the Microchip PIC16F726 microcontroller.

I got lost after this point. There are many unpopulated pads, I presume features for higher end models. SW2 would be for another switch, but what might sit at BZ1? Some of these components would deal with power coming in from the charging coil (connected at the holes on either side of the CR6 lettering) and maybe for battery management. The positive battery terminal connection point was mentioned earlier, the negative terminal connects to the upper-right hole adjacent to lettering JP1.

Speaking of the battery, unsoldering its connection to the circuit board freed it from the chassis. Now we can read the lettering on its side telling us it is a 14500 form factor (14mm diameter, 50.0mm length) cylindrical cell with a capacity of 680mAh (milli-Amp hours).

The charging coil stayed on the black rigid plastic chassis, but now that it is freed from the circuit board I hooked it up to the oscilloscope again for another look.

Left plot was taken while the charging coil was connected to the circuit board, right plot is the coil by itself. Note the vertical scale has changed between these two plots: 1V/div on the left and 5V/div on the right. The most obvious difference is the shape: it is now a sine wave instead of a square wave. The amplitude is much larger at about 30V peak-to-peak or 15V above and below the center point. This is roughly triple the rectified 5V I saw earlier. I don’t know enough about analog electronics to know what would accomplish this on the now-removed circuit board, maybe in the future.

With the electronics disassembled, my attention returned to the electromechanical actuator assembly.

I’m partway through disassembling a retired Sonicare electric toothbrush, far enough to see all the internal components. The likelihood of irreversible damage rises sharply from this point. Before I do anything destructive, I wanted to check out some electrical characteristics under my oscilloscope.

The biggest chip on this circuit board is a Microchip PIC16F726 microcontroller.

Next largest are a pair of chips with the logo of Alpha & Omega Semiconductor. A brand I’ve encountered in a few previous teardowns as a provider of field-effect transistors. Their 8801 is a pair of P-channel FET and the 8808 is a pair of N-channel FET, sitting close to the actuator coil wires.

I didn’t see any other significant-looking chips on this board. I thought there might be a lithium-ion battery management chip like a 4056, but I found no likely candidates. Perhaps such tasks are handled in software running on that PIC16F726.

First target: big test pads labeled “Tx” and “Rx”. These names usually denotes serial communication and I was curious if there’d be any sort of diagnostic messages during runtime. I don’t know any of its parameters (voltage, baud rate, etc) so the first pass is to look at them under the oscilloscope. As a user, there were only a few events I can trigger: start/stop brushing, and start/stop charging. I was disappointed but not surprised to see no electrical activity on these pads. The chip is staying quiet until it sees the right access code, which I don’t have.

Second target: the inductive charging coil. Channels 3 and 4 were connected to either end of the coil and plotted relative to battery negative as ground. The plot on the left has both channels 3 and 4, the plot on the right has only channel 3.

I had expected to see a 60Hz sine wave on this coil when sitting on the charging base, which runs on household power ~120V AC @ 60Hz. What I saw had a period of around 11us, which translates to about 90 kilohertz. That was a surprise, as well as the amplitude. I had expected it to be symmetric about the ground plane but it’s actually going from about 0V to 5V. And finally, it is a square wave and not a sine wave. In short, I expected none of this and I don’t understand what I’m looking at.

I left the toothbrush on the charging base connected to the oscilloscope until the status LED went from blinking (“charging”) to steady on (“charge complete”). I measured the battery voltage at 4.1V at this point. I thought perhaps the inductive charging coil would look different when it’s not actively charging, but it looks pretty much the same.

Third target: the brush actuator coil driven by those Alpha & Omega FETs. Again channels 3 and 4 were connected to either end of the coil and plotted relative to battery negative. The plot on the left has both channels 3 and 4, the one on the right only has channel 3.

When actively brushing, the coil is activated in one direction (~5V on channel 3, 0V on channel 4.) for roughly 1.25ms. Then both ends are grounded for roughly 0.7ms. Then the coil is energized the other way (0V on channel 3, ~5V on channel 4) for roughly the same duration.

Every 30 seconds the brushing pauses and the toothbrush issues a “beep’ to remind us to switch to a different set of teeth. I had been curious if this was implemented as reduced voltage level or reduced coil energizing cycle. The answer is the latter: all of the time periods described above shortens by roughly 25%, that was enough to stop the coil from brushing but enough to cause an audible beep.

After watching the brush actuator waveform for a few cycles, the toothbrush fell silent and the waveform collapsed to a single phase. Oh no, what happened?

One of the coil wires had broken. It was designed to handle stress of brush vibration, but not with the additional weight of an oscilloscope probe. That was too much and metal fatigue took care of the rest. Well, I’ve seen what I wanted to see so this is a good place to stop my oscilloscope examination and resume mechanical disassembly.

After wrapping up a teardown (and unexpected repair) of a Philips product, I decided that was a good teardown theme and proceeded to another retired product from a different Philips subsidiary: Sonicare.

This Sonicare HX6530 could still vibrate the toothbrush head, but it could no longer clean effectively. Its degradation was long and gradual enough I didn’t notice until my dental hygienist commented on my worsening dental buildup. I tried a new Sonicare and realized I should have retired this HX6530 long ago.

Some of the newer Sonicare have been designed to be easily opened for battery recycling, we just have to push on the metal toothbrush stem. This older model is locked up tight and not so easily opened. I thought maybe clamping the plastic will flex and break some glue joints, but nothing interesting happened.

Using my also-much-degraded Wondercutter (a topic for another post) I cut an entry slot and started prying away with pliers. I quickly got far enough to confirm there’s a black rubber O-ring as water barrier.

More prying later, I reached a piece of black plastic that looked like a latch. Once all its surrounding plastic has been pried awayI tried pushing on the metal stem again.

Cutting one latch away weakened the holding power of the remaining latch, so this time a firm push on the metal stem was enough to push everything out of the enclosure.

There are definite signs of water intrusion inside, such as these dried deposits near the tip.

There was also rust near the interface between the coil and the armature. Unlike hair trimmers, reciprocating motion is not created here with a spinning motor shaft turning a crank. Instead it appears to be using a pair of coils facing a magnet.

I didn’t see an obvious candidate to explain why this toothbrush degraded. The signs of water intrusion didn’t seem sufficient to damage the electromechanical assembly, and all components appear intact on the circuit board. The battery voltage measured 3.8V DC, within nominal range.

I had expected the battery to be a single lithium ion cell in the commodity 18650 form factor. It turned out to be a much smaller cylindrical cell made by Sanyo. Here it is next to a 18650 cell for size comparison. Also visible to the left is the coil for interacting with the charging base.

I could peel three large pieces off the core. They are all made of soft vibration-absorbing material, the midships piece incorporated a spring for even more vibration absorption. It makes sense that keeping vibration under control would be an important mechanical engineering consideration for this device. It’s not just for user comfort, they have to make sure it doesn’t shake itself apart!

I was encouraged by the few Philips-head screws I could see at the base of the electromechanical assembly, thinking I might be able to disassemble it non-destructively.

That hope was dashed when I saw the other end of these two metal pieces were welded together. Oh well! Going further will be a destructive act. Before I do that, though, I want to probe the electrical aspects of this device.

I’ve managed to disassemble the exterior components of a retired Philips Norelco Multigroom (MG7790) which made only a moderate mess with its hair fragments trapped within. In earlier hair clipper teardowns, that was enough for me to access all internal components. However, this unit is IPX7 waterproof which meant everything is inside a clear plastic watertight capsule.

The control circuitry is much more sophisticated than those in the two earlier hair clippers, though I’m not sure why. At the end of the day, this just needs to turn a motor on or off. Perhaps there’s battery management as well? The little of blue peeking through various gaps in this enclosure hints at a single lithium-ion battery in a commodity 18650 cylindrical cell form factor.

The motor is genuine Mabuchi! A pioneering giant for small DC motors, I usually encounter knockoffs of popular Mabuchi designs instead of a genuine Mabuchi. Either that, or it’s a counterfeiter brazen enough to print the Mabuchi logo on their wares. I’d like to think Philips/Norelco uses the real thing.

The bottom of the clear plastic capsule is sealed with a black plastic plug at the bottom, held in place with four beefy clips. I had hoped to release those clips peacefully, but I shattered some of the clear plastic. The cracks extended past the watertight O-rings, meaning I’ve managed to destroy its waterproof worthiness.

Not much to lose now! Feeling liberated because I don’t have to worry about preserving its watertight seals, I tore apart the retaining mechanism. This clear plastic was interesting: most of it is tough and ductile under stress like a polycarbonate, except infrequent times where it shatters like brittle acrylic. I wish I knew more plastic engineering to better understand this material’s behavior.

I thought the black plastic plug at the end was a separate piece of plastic from the structural chassis for all internal components. It turns out to be a single piece, so everything slid out together.

The chip that seems to be in charge is stamped with 10368A F15 H3G7A. A web search found a likely candidate in the Renesas RL78/G12 family of microcontrollers. The R5F10368 variation has 8KB of code flash, no onboard data flash, 768 bytes of RAM. It comes in a 20-pin TSSOP package which matches what’s on this board.

If I’ve correctly identified the controller, it means this is a very software-centric minimalist application because this MCU lacks onboard peripherals befitting the hardware. (Comparison: the MCU in this smoke detector is optimized to be a turnkey smoke detector solution.) There’s no obvious motor control peripheral like hardware PWM, though perhaps an offboard MOSFET would suffice as it only needs to turn the motor on/off. There’s no explicit lithium-ion battery management capability like a 4056 chip, but there is an A/D converter that would be suitable for monitoring battery voltage.

Speaking of the battery, separating the circuit board from the chassis exposed the battery. The label says:

LG Li-Ion Cylindrical
INR18650S3
2200mAH 3.6V~4.2V
G15 2016.08.12

I retired this trimmer because the motor would stop in the middle of a session, even when fully charged, so I decided the battery has degraded. I thought it might be fun to replace the battery cell and see if that gets it back up and running, hence the concern about trying to preserve the watertight barrier. But as it turned out, I had misdiagnosed the problem.

After my Remington hair clipper‘s batteries failed, I had to shop for a replacement. Costco had the Philips Norelco Multigroom (MG7790) on sale and I thought I’d give it a shot. Rather than a single wide fixed-size cutting blade, the Multigroom came with multiple cutting elements to serve different hair cutting/trimming purposes. The tradeoff for this versatility is that the widest flat cutting element is around 2/3 the width of the Remington hair clipper blade. Part of being a smaller and lighter weight device, which I learned to appreciate during its few years of use. I didn’t end up using the other cutting elements very much, and the narrower width wasn’t a significant detriment in my usage pattern.

After a few years of use, the motor would stop running in the middle of a session even if I fully charge the battery immediately beforehand. The manual claimed a fully charged battery can deliver up to six hours of use, but I couldn’t even get ten minutes! I bought a replacement and placed this unit in the teardown waiting list alongside the Conair and Remington hair clippers, where they waited until now.

The two old hair clippers both had a sticker telling us to properly recycle their nickel-cadmium batteries within. There was no such label here telling us of the type of rechargeable batteries used. The power adapter operates at 15V DC, far above the voltage range of any battery pack I would expect to see within. This implies a voltage buck converter in the charging circuit, a level of sophistication I associate with lithium chemistry battery management systems.

An important detail for teardown purposes is at the bottom of multigroom label: “IPX7” followed by a faucet icon. This device is waterproof enough to be rinsed clean under running water and can tolerate being briefly submerged in water if we accidentally drop it in the tub. Waterproofing makes teardowns difficult because the interior may be glued shut and/or buried under epoxy resin. Even if they weren’t, a teardown may irreversibly damage the water barrier.

The AC adapter uses this vaguely B-shaped plug that I haven’t seen anywhere else, possibly proprietary to Philips/Norelco. Next to the charging port at the base is a Torx-headed screw hidden under a rubber cover. That looks like a good place to start.

The base plate came free easily. Loose fit and no seals meant the waterproof barrier must be further within. Disappointingly, nothing else budged after this plate was removed, so I started looking at the other end.

Different cutting elements can be easily snapped on and off the top. When removed, we can see the motor-driven crank. The manually aptly referred to this as the hair chamber. It’s easy to dump out loose hair, but the many small corners meant it takes a lot of work to clear out every little piece. Going in with a cotton swab, I got enough of the corners cleared out to see three small fasteners (combination Torx and flat head) buried underneath.

Releasing those three small self-tapping plastic screws allowed hair chamber disassembly, releasing a lot of trapped hair along the way. Now the guts of the device can slide back and forth around 3-5mm, but not quite freed yet.

The last piece of the interlocking puzzle was the power button, which was held by multiple clips that also served as tabs holding multiple internal elements together. I damaged one clip removing it.

The internals slid out as a single piece enclosed in clear plastic. The motor pokes out the top, a rubber dome on the side pushes on the power switch, and a plug of dark gray plastic sealed the bottom. Numerous rubber O-rings made it clear this was the water barrier. Getting inside non-destructively proved to be a challenge.

I used my Conair HC318R hair clipper for several years. Until its batteries couldn’t hold enough charge to last a single cutting session, at which point the blades were pretty worn as well. It was replaced by a Remington HC-920 which was itself retired after several years of service. Tearing down a retired hair clipper will scatter tiny bits of hair all over my workbench, so I’m working through my backlog of three retired clippers all at once.

Going into this second teardown, I expect to find minor execution detail differences between these two broadly similar devices. Previous compare-and-contrast teardowns included two inexpensive coffee makers (Black and Decker DCM600B vs. Mr. Coffee BVMC-SC05BL2-1) and two chop-type coffee grinders. (Hamilton Beach CM04 80344 vs Bodum 11160-3)

I picked up the first major difference from the product label: the Conair listed 4.5V but this Remington only lists its voltage at 3V. This implies two nickel-cadmium battery cells in series instead of three as in the Conair. Theoretically it meant the Remington was less powerful, but if so, the difference wasn’t enough for me to notice. This may be because I didn’t have a fair back-to-back comparison. I would have been comparing the Conair with worn-out blades and batteries against a new Remington.

Both clippers had two prominent Philips-head screws on the cutting head, allowing easy removal of the outer fixed blade less-easy removal of the inner moving blade. This allows us to clean out hair that may have accumulated between the two blades or against the motor-driven crank seen here. This Remington used machine screws threaded into a piece of metal, which should be more durable and longer-lasting than Conair’s approach of using screws that self-tap into plastic. In practice, the longevity was not a problem, though that may be more a commentary on my cleaning frequency (not very) than on mechanical design.

There were two less-prominent screws holding the thing together. One at the base of the handle, and one in the length-adjustment lever. They were recessed but otherwise exposed on the Conair, the Remington added cosmetic covers over them for better aesthetics. Once removed, a metal bracket at the front (what the cutting head machine screws held into) could be slid off, then we can open the enclosure.

Unlike the Conair, Remington has this additional metal strap to hold the motor in place. There’s a piece of foam between the strap and the motor to dampen vibration. The strap are held by a pair of Philips-head screws which had this orange material on top. It may just be a thread locking mechanism, but having orange stuff in the fastener head made things more difficult to remove.

As inferred by the 3V DC listed outside, there is a pair of Sanyo-made AA-sized nickel cadmium battery cells inside.

As I lifted the innards, I could see some external signs of failure on this battery cell. Not sure if this is corrosion from leaking chemicals or something else, all I know is I should avoid touching it.

The Remington circuit board is even simpler than Conair’s board, showing a slightly different evolution. Here R1 was deemed unnecessary and replaced with a solder bridge.

All disassembled components will meet fates similar to their Conair counterparts. Metal will be recycled, plastic will go into landfill. The battery will be added to a collection of nickel-cadmium batteries that will be turned in for responsible disposal. The charging barrel jack will stay with the AC adapter for potential reuse.

I started removing this shiny brass metal crank from the motor output shaft but changed my mind. First, unlike the Conair motor mount + crank, it didn’t trap much hair and could be cleaned up. Second, it was fastened very tightly and would be a lot of effort to remove. Third, I might want to build a reciprocating motion device at some point, and it’d be handy to have this motor already set up ready to go. Fourth, it looks really nice I wouldn’t know what else to do with it anyway.

And finally, there’s no rush. I already have a nearly identical motor salvaged from the Conair unit, with the plastic crank already removed. I’ll worry about removing that piece of brass onceI have an actual need to do so. Right now, I have a third hair clipper to take apart.

I have a set of retired hair clippers on the teardown to-do list. My hair are like bristles in a stiff wire brush, and wears down cutting blades quickly, fresh clippers only keep their “oh wow, this cuts nice” feeling for the first half-dozen or so sessions. I use one for a few years until its rechargeable battery degrades. By that point, its blades are dulled enough to be tugging and pulling hair while cutting. Between the battery and the blade, it was time for a new one.

I’ve been avoiding these teardowns because I knew it would get messy, given the short fragments of hair clinging to every surface of these clippers via either lubrication oil or static electricity. I knew I will find small pieces of hair everywhere for weeks after each teardown. But the paper liner on my workbench is starting to get dirty enough for replacement, and I saw an opportunity: I can do these messy teardowns and throw away the liner, hopefully easing cleanup.

First up is the Conair HC318R. Information embossed on the back of the device was hard to photograph, but it says:

(C) CONAIR CORPORATION, EAST WINDSOR
NJ 08520 / GLENDALE, AZ 85307
MODEL HC318R 4.5V DC 1000mA
USE WITH ADAPTER CA12
MADE IN CHINA
HOTLINE 1-800-3-CONAIR

Next to this information is a small tag informing me the device contains nickel-cadmium battery and must be recycled or disposed of properly. I expect to find three cells wired in series given the 4.5V DC written on the device.

If somebody has the clipper but not the charger, here are the specifications on its label. The charging port looks to be a pretty standard barrel jack wired to be center-positive.

There were two very large and easily accessible Philips-head screws on the blade. Removing them frees the outer blade, allowing us to clean inside.

The inner blade lifts away to expose the motor-driven crank.

Opening the enclosure allows a look inside.

Behind the motor sat the expected three-cell nickel-cadmium battery pack, each cell is the size of an AA battery.

There were no further fasteners holding the working bits in place. Once the enclosure was opened, the electrical guts are easily lifted out.

The battery pack manufacturer is BYD, the battery company that grew into an automotive manufacturing conglomerate. I measured less than 2V across its terminals, so it’s either severely discharged or there’s a dead cell in the pack. Neither would be a surprise after years of use followed by more years of waiting for teardown.

Thanks to decades of manufacturing at scale, nickel-cadmium batteries are cheap. They are also hardy and tolerant of abuse, which cut costs further because they don’t need sophisticated battery management electronics like lithium-based batteries do. Here we have just a small single-layer circuit board housing the power switch, two resistors, two diodes, and one light-emitting diode. M+ (motor positive) is connected to one side of the switch opposite B+ (battery positive). M- location was unused: it had a short trace to diode D3 which was also absent. Motor negative was wired to battery negative off board then a single wire led to B- on this board, which was electrically connected to the other end of the absent diode D3. Apparently, they decided D3 was unnecessary! Finally, L+/L- lead to the AC adapter barrel jack for charging.

The motor is marked with 222C3V6 3760 but I didn’t find much from that information. The crank is a piece of friction-fit plastic and slid off unexpectedly easily. The motor mount was installed with two small Philips-head screws and easily removed.

The barrel jack will stay with the AC adapter for possible future reuse. The motor will join many other salvaged motors. The battery pack will go in the nickel-cadmium battery recycle bag. The circuit board will go into the electronic recycle box. Metal blade and fasteners will go into metal recycle bin. Remaining plastic enclosure bits will go into landfill along with many bits of hair. I didn’t put too much effort into cleaning the workbench, because I had more hair clippers for teardown.