Tools

Roger Cheng

New USB Serial Adapter Show Minor Updates

I want to listen in on both directions of communication between main board and control panel of a Canon Pixma MX340 multi-function inkjet. I thought I had the hardware on hand to do this: one USB serial adapter that I’ve been using successfully for years, and a second adapter I bought but haven’t needed until now. The second unit turned out to be a dud, wasting some of my time before I realized it was junk. I went back to my earlier purchase and ordered another unit from the same Amazon listing. (*)

Even though I ordered from the same listing, I did not expect to receive an identical unit. It’s very common for Amazon vendors to switch products without switching the listing. This can be done maliciously, selling a listing with high star rating to another vendor who then offload junk. Sometimes this scam is obvious as the reviews discuss a completely different product, but some buyers don’t even bother to read those reviews to pick up on the fraud. Even without fraudulent intent, a vendor may switch to shipping a different but similar product or upgrade to an updated version. I think I have the latter scenario. Here’s a picture of the new arrival (top) and my veteran workhorse (bottom)

The two devices are substantially similar, with the same size, I/O locations, and number of surface-mount components laid out in the same locations. But the circuit board underneath is not identical with small differences such as the width of traces, taking turns at different angles, or small differences in printed labels.

The most visible difference upon power-up: LED colors. My veteran uses three red LEDs, the new adapter use three different colors. Yellow for power, green for receive activity, and red for transmit activity.

The product listing advertises FTDI chip. Is it a genuine FTDI chip or one of the many fakes riding on FTDI’s success? I don’t know how to tell. I do know, however, that this new adapter passed its first test making it better than the dud. I wired it to listen to traffic from MX340 main board to control panel. I pushed a button on the control panel to trigger a LCD screen update, and this new adapter reported 1020 bytes of data was transmitted. This was the expected value and good enough for me to get back to work.

This teardown ran far longer than I originally thought it would. Click here to rewind back to where this adventure started.

(*) Disclosure: As an Amazon Associate I earn from qualifying purchases

Roger Cheng

Time Wasted By Faulty USB Serial Adapter

I have deciphered how the main board of a Canon Pixma MX340 multi-function inkjet commands its control panel to toggle two onboard LEDs, one more step towards understanding how those two components work together. Now I want to incorporate that new knowledge to my serial data filter program and add a second channel of serial data.

To sit alongside my trusty and proven USB serial adapter, I dug up an inexpensive USB serial adapter I bought years ago. It was still unopened in its package. I don’t remember what I originally bought it for, but apparently I didn’t end up using it. The device is a blue USB plug with four wires coming out of it, a very generic form factor that still comes up for searches on “TTL Serial USB adapter”.

I plugged it into the USB hub and it was recognized as a USB serial device. I wired it to receive MX340 main board commands in parallel with the control panel. I wrote a few lines of Python to read from this adapter and it was a mess. A LCD screen update — which I had established to be 1020 bytes long — was received as 65 bytes. Not only did the majority of data go missing, the bytes that came through didn’t resemble the expected data at all.

I first thought I made a wiring error and doubled checked my connections. Then I thought there was a coding error. When neither turned up a plausible explanation, I flipped the channels connected to my two serial adapters. Now my trusty USB serial adapter reported the expected 1020 bytes of data in a LCD screen update, and this new-from-package adapter reported gibberish instead of proper control panel button scan codes.

This USB serial adapter sucked straight out of the box. Searching on my Amazon order history, I found my order for this adapter in September 2019. I was curious to see if this adapter caused problems for other people so I clicked on the product link, and that went to an error page. This product was so excellent its listing had been pulled! It was probably filled with negative reviews. I would have written one if I had opened the package as soon as I got it and realized it was junk. It had cost $7. Less than $12 I paid for my working adapter (purchased June 2018) but $7 is obviously too much for garbage. Since it’s far too late for me to return this device for a refund, I shrugged and chalked it up to lesson learned. I went back to my order history, found the invoice for the USB serial adapter (*) with a proven history and ordered another unit.

This teardown ran far longer than I originally thought it would. Click here to rewind back to where this adventure started.

(*) Disclosure: As an Amazon Associate I earn from qualifying purchases.

Roger Cheng

Polarizing Filter Photography is Magic

Starting with the recent Philips Sonicare teardown investigation, I’ve reintroduced the use of my Canon EOS M100 camera to take pictures for project documentation. Its excellent sensor paired with a macro lens for close-up photography work quite well to capture small surface-mount components and related circuitry detail. Late in the investigation, I added polarizers to the setup and was amazed at how much of a difference it makes under the right circumstances.

My current understanding is: if I use a light source that is polarized in a direction, then take a photograph with a filter polarized in an orthogonal direction, most direct reflections are eliminated. This means the resulting pictures will not be blown out by reflective glare, even though the subject can be very brightly lit Letting us make out details we could not otherwise see.

Here’s the Sonicare HX6530 circuit board again, taken without polarizers. All the solder pads are very shiny, as are some of the circuit board surfaces. I tell the camera to focus on the markings on the surface of the PIC16F726 microcontroller, but we can barely make it out.

And now a picture taken with polarization. All the shiny reflections are gone, letting us see far more detail that were previously obscured. PIC16F726’s markings are now clearly legible.

This is a great tool to add to the project documentation photography toolbox, but like all tools there are right and wrong times to use them. Even though I can see details with the polarizer in place, some of those details may be misleading. Example: actuator coil pads 1 and 2 show up as red in the polarized picture above. I don’t know why but it tells me to view colors in polarized pictures with skepticism.

But sometimes polarizers are just magic. There’s clear packing tape protecting this solder joint. Normally the tape would directly reflect light into the camera sensor, causing a bright glare on the surface making this solder joint impossible to photograph. But a polarizer can filter out that glare and let the camera focus on the solder joint almost as if the packing tape isn’t there. The downside of removing bright reflection is that it also removes a lot of visual cues for dimensions. Look at the red wire: all visual cues it is cylindrical are gone leaving it looking like a flat red thing.

Polarizer filters are sold for camera lenses for use in this and other scenarios. The Canon EF-M macro lens takes 43mm filters so I started with the cheapest option on Amazon (*) to see how well it worked. The outer ring is assembled from two parts: one part threads onto the lens and stays put, the other part holding the filter rotates freely letting me adjust the polarization angle. These initial tests look promising so I’ll use the cheap filter until I can articulate a reason to move upscale.

The camera lens polarizer is cheap to buy and easy to use, but getting these results also require a source of polarized light. I think the easy solution is to buy sheets of polarizer film (*) and put it over your existing light sources, but that’s not what I did. I wanted to put some old stuff to work.

(*) Disclosure: As an Amazon Associate I earn from qualifying purchases.

Roger Cheng

Dusting Off Canon EOS M100

Recent changes to Chrome browser motivated me switch to Firefox, but that wasn’t the only recent Google decision that irked me. Google also decided to be a lot less generous with Google Drive. Google Pixel camera images received unlimited Google Drive backup storage, but that perk ended with the Pixel 5. Images from Pixel phones newer than Pixel 5 count against Drive limit just like any other data. I guess Google is no longer funding the Drive division for supporting Google hardware adoption.

It was very convenient to document my projects with my phone camera as it is usually nearby. But being generous with the shutter also meant I was quickly filling up my Google Drive quota. With the flood of nag mail demanding I pay money to increase my quota, I was motivated to migrate from cloud-based storage back to local storage on my TrueNAS array. And if I’m doing that anyway, I might as well dust off my Canon EOS M100 camera. Shown here with a lens for up-close photography and a wrist strap (*) far less bulky than the standard neck strap.

Convenience of cloud backup was a major reason why my M100 had been gathering dust. Now that Google has made it less enticing, I’m returning to the superior photography capability I’ve already paid for. The M200 is the current successor but I don’t feel a need to upgrade yet. Even this older M100 has sensor and lenses that easily outperform a phone camera, even more so when dealing with projects like my Sonicare HX6530 circuit board.

Credit to the Pixel 7 camera design team, it performs well with subjects at normal ranges. But I would not call my usage pattern normal, and I’ve highlighted problems before.

Above is a serviceable picture taken by the Pixel 7, below is from the M100.

When scaled to blog post size, they’re pretty close. But when I want to crop, the M100 sensor delivers more pixels to work with. Here are 1:1 pixel crops from original resolution of above images:

The M100 gives me more pixels to crop closer to the subject. The Pixel 7 is quite competitive here looking at the center of the image, but it falls behind when we start looking towards the edge:

Both lenses degraded as we move away from the center, but the EF-M Macro lens does a better job. And while the phone camera couldn’t focus much closer than these pictures here, the Canon lens can focus far closer. So close, in fact, that the lens itself casts a shadow blocking surrounding light. Which is why it has a built-in ring light to handle those cases.

So a dedicated camera has a better sensor. In front of that better sensor is a better lens, which can be swapped out for different lenses optimized for specific scenarios. And in front of that lens is a finely threaded mount for lens filters, which adds even more capability. For my electronics project photography, adding a polarized filter is a magical transformation.

(*) Disclosure: As an Amazon Associate I earn from qualifying purchases.

Roger Cheng

Miniware Soldering Iron (TS80P)

Following successful initial tests with Miniware’s little hot plate MHP30, I went ahead and bought another product that I had been curious about. This is the Miniware TS80P soldering iron, which I bought through Adafruit as item #4244. I’ve read positive comments on this little USB-C powered wonder through Hackaday and other sources. And I like the idea of having a small soldering iron I could take to events like LayerOne to join in electronics fun.

User interface for TS80P closely resembled that for the MHP30 hot plate, running on a small OLED with two buttons. Since they’re both small USB-C powered heating devices, I suppose there’s a chance they’re running mostly (or possibly exactly) the same firmware code. I’m not a huge fan of their interface, as there are only two buttons whose functions vary depending on context making it hard for me to remember what each button would do at any given time. With time I suppose I will get used to it, and most of the time I just want to turn heat on/off so I doubt the interface would be a long term hinderance.

I had also hoped that the nature of USB-C power meant I could use this soldering iron away from a power socket, which would be useful if I need to perform field repairs on my Sawppy rover. Unfortunately, I can’t do that just yet. I own a few battery power banks with USB-C output, but none of them could support the voltage and amperage demanded by this little soldering iron. If I want to add this to my Sawppy field repair kit, I will have to buy a power bank capable of 12V 3A via USB Power Delivery.

A quick test soldering a 7-pin 0.1″ header went smoothly with neither good or bad surprises. Using this little pen-sized iron felt strange because I’ve become accustomed to holding a much larger iron when soldering. But aside from heat insulation, I don’t know of any fundamental reasons for a soldering iron to be big. Some people prefer the light weight and compact size maneuverability of a small iron, switching to these little guys as their primary soldering iron. I don’t know if I would make such a change, time will tell.

Roger Cheng

Micro-Mark 2″ Self-Centering Machinist’s Vise

For my workholding needs, I have been using the most affordable vise from Harbor Freight: Item #30999 4-inch Drill Press Vise. It was a huge improvement over holding things by hand, but it wasn’t great, and I’m ready for an upgrade. Now I have Micro-Mark’s Item #87469 2″ Self Centering Machinist’s Vise with Swivel Base and first impressions are promising.

The Harbor Freight vise was sufficient for most tasks. The most important feature were removable jaws: I designed and 3D printed many custom jaws to hold pieces of various projects. Once I’ve clamped something down, it stays put, fulfilling the point of a vise. My problem was that the sliding jaw had quite a bit of play along its motion. This caused two problems. First: the tiny amount of play in its motion means it can be difficult to clamp something exactly where I want it. As I tighten the vise and take up slack in its play, those jaws might shift a tiny bit which also shift position of my workpiece. Second: this vise expects all forces to be downward, which is reasonable for a drill press vise. But sometimes my drill bit would pull on the work piece as it drilled, and that would lift the workpiece and sliding jaw upwards a tiny bit as well. Most of the time this isn’t a problem, but every once in a while, that upwards lift ruins things.

When it comes to tools, precision costs money. The Harbor Freight vise was very cheap so its flaws were reasonable. To gain better precision I would have to move into the realm of vises designed for metal machining work. (And spend more money.) A 6″ vise is the standard baseline machining vise but that is overkill for my needs. (Example: Kurt DX6) I didn’t see anything promising on the Harbor Freight catalog, so I looked into the Micro-Mark catalog for accessories alongside their small metalworking mills. I started by looking at their Item # 82577 Quick-Lock Milling Vise, but that vise didn’t appear to have removable jaws. Then I looked at Item #21134 Toolmaker’s Vise which had removable jaws, but the geometry of the design gave me concern. All the machinist vises I’ve looked at had their leadscrew below the working area, so that a portion of clamping force is also directed downwards to resist upward pulls. I’m not certain #21134 does that. The only vise that seems to meet all of my requirements is Item #87469 2″ Self Centering Machinist’s Vise with Swivel Base. I didn’t particularly care about the self-centering aspect, but everything else looked promising.

A few days later it arrived in a well-padded box. The vise itself was coated in oil, sealed inside a plastic bag to avoid oil leaking into the shipping materials. I can confirm smooth confidence-inspiring movement in the jaws, which are removable as required. Even though both jaws move instead of just one moving, they still had far less free play combined than the singular jaw of the Harbor Freight vise. The resistance to upward lifting forces is still a question mark until I really put it to use, but I’m satisfied with everything else.

The only criticism is the handle, which extends below the level of the base. When I bolt this vise to a surface, I might not be able to crank it all the way around unless I make sure it dangles over an edge. Otherwise I would have to remove the handle and reinstall it at a different angle for every turn, which could get annoying. I’m not sure how serious of a problem this will be in practice, time will tell.

Roger Cheng

Miniware Mini Hot Plate (MHP30)

The latest addition to my toolkit is a mini hot plate designed for electronics work at a small scale. This is the Miniware MHP30, purchased from Adafruit as their product #4948. Emphasis of this product is on “small”, as the actual heating area is a square only 30cm (approx. 1 1/4″) on a side. The hot plate unit itself is dwarfed by its power supply that came in the same box. This is a USB-C supply that, through the magic of USB Power Delivery protocol, can deliver up to 65 Watts. (20V @ 3.25A)

I believe MHP30 was designed for reworking surface mount electronics, focusing energy just on the portion of the board that needed fixing. I’ve wished for this kind of capability in the past, for example when I wanted to remove a voltage-divider resistor. I bought a hot-air rework station which was useful, but I was sometimes stymied by boards with a large heat dissipating ground plane. I found that I could help my hot air gun by heating up circuit boards with an old broken clothes iron. That experiment represented a bluntly crude version of what this little focused heating plate promises to do.

The first test run was to remove an already-ruined LED module from a dead LED bulb. I tried to remove this LED with a hot-air gun. Not the small electronics hot-air rework station kind of hot-air gun, one of those big units sold for paint stripping. Now armed with the proper tools, I could successfully remove what remained of this LED module.

The next test was more challenging. I wanted to see if this hot plate could heat up multiple legs of a through-hole part allowing me to pull it from a circuit board and do so without damaging the more heat-sensitive plastic portions. The test subject was a DC power barrel jack from a dead Ethernet switch.

One reason this is difficult is due to the air gap caused by the through-hole legs, which would let heat escape instead of melting solder. I declared this test a failure when the black plastic started melting before the solder did. I have other means to salvage through-hole parts, but I had hoped this would make things easier. Oh well.

Returning to surface-mount components, I wanted to try removing the CPU on this dead Ethernet switch. This would have been difficult with the hot-air gun alone, because most of the heat would be absorbed and quickly dissipated by the black heat sink glued to the chip.

The bottom of the chip is connected to the ground plane of this circuit board, which is a layer of copper occupying nearly the entire circuit board. If I only used hot air gun from above, heat would never get past the glued-on heat sink to this area.

But if I heat from below with the mini hot plate, things heat up enough for the heat sink glue to degrade. Once the heat sink was pulled off, things moved much more quickly.

The mini hot plate sends heat directly into the large heat conducting pad in the middle of the chip, melting the relatively large square of solder in addition to all little pads around chip perimeter. I don’t think I could have done this without the mini hot plate. This test was successful, but we had the advantage of a singled-sided board with no components on the bottom. If there were, then the air gap problem would return. It isn’t a complete answer, but I’m happy to have added the little hot plate to my toolbox.

Roger Cheng

Up and Running on Monoprice Creator 22

After unpacking a Monoprice Creator 22 graphical pen display and installing its driver software, Windows 10 detected a pen input device and activated a few inking tools. One example was a digital sticky note I could use to jot things down. These tools were enough for me to verify that position and pressure information is getting into the system. I also noticed that whenever there is pen activity, one CPU core is completely consumed with kernel-level tasks. This is a hint the Bosto driver is spinning a CPU polling for input data whenever the pen is active. It is certainly a valid way to maximize pen input responsiveness, but not the most efficient. On the upside, we’re now living in era of multicore processors. So I guess it doesn’t matter too much if a CPU core is entirely occupied with pen input.

Sticky notes are fun, but I wanted to use a more powerful tool. Since I’m unwilling to spend significant money until I have more experience, I will start with the free option: GNU Image Manipulation Program (GIMP). Loading up the current public stable release (2.10.32 as of this writing) I found it did not respond to my pen. Rummaging around the internet eventually found that GIMP only added support for Windows Ink compatible pen input devices about a year ago, in development version 2.99.8. Uninstalling the stable release and installing the development release (2.99.12 of this writing) allowed me to select Windows Ink as GIMP’s input API and draw using my new graphic display.

GIMP is a lot more powerful than a digital sticky note, and its user interface is infamously hostile to beginners. There are official documentation and online forums, of course, but I think a guided tour might be a good idea as well. I considered The Book of GIMP (*) by Lecarme and Delvare, but it was written almost ten years ago in 2013 for GIMP 2.8 so many details will be outdated. I might still skim through this book for the major strokes, or I might find a different book.

Last but not least, I need to put some effort into learning to draw. I’ve been doodling random things since I was old enough to hold a crayon, but I’ve never put any rigorous effort into developing the skill. I’m starting with The Fundamentals of Drawing (*) by Barber. Not because the book is great (I don’t know enough to judge yet) but because Amazon’s “Look Inside” feature got far enough for the first exercise: practice hand control by drawing basic shapes. Can I stay focused enough to practice these drills and get good at them, so I could contemplate actually buying the book for the rest of it? Time will tell.

(*) Disclosure: As an Amazon Associate I earn from qualifying purchases.

Roger Cheng

Monoprice Creator 22 Graphic Pen Display (Item #39945)

Ever since I saw my first Wacom Cintiq, I’ve thought graphic pen displays were a cool technology. However, they were priced for professional artists and I couldn’t justify spending that kind of money just for messing around. But when Monoprice held a clearance sale, I could not resist a Creator 22 (Item #39945) for just $170.

The underlying panel has the good color and viewing angle typical of IPS displays. With a 1920×1080 “Full HD” resolution it is acceptable but not great. Noticeably lagging in today’s world of 4K UHD and Apple’s “Retina” displays. There is a tangle of wires feeding into the back of this unit. A HDMI cable for video, a 12V DC @ 2A power supply, and a USB-A port for graphical input. In theory USB Type-C can handle video, power, and input all in a single cable, but we don’t have that advancement here. Heck, peripherals like this aren’t even supposed to use Type-A ports (peripherals are supposed to be Type-B) but here we are anyway.

According to the enclosed manual, there was supposed to be a USB flash drive in the box with device drivers, but I didn’t find one. Fortunately, there was a download link on the Monoprice product page. While the product has a Monoprice name on it, the driver installer showed a different name: Bosto. Looking over Bosto Tablet’s website, it appears Monoprice Creator 22 is a rebranded variation of the Bosto 22Mini.

Part of Bosto’s driver is a calibration app, where we are to tap the pen at crosshairs drawn in each corner so the driver can map their position to onscreen display. Unfortunately, this calibration app doesn’t work properly on setups with multiple monitors: the crosshairs were drawn on the wrong monitor and I couldn’t figure out how to move it. To work around this issue, I disabled all other monitors to run the calibration routine. Afterwards I could re-enable remaining monitors and thankfully pen input knew to work on the correct display. However, despite running the calibration routine, the cursor position did not always match up with pen position. It seemed fine near the center, but near the edges & corners the position would be off by up to 3mm. I would not be happy if I had paid full list price, but I was willing to tolerate it at discounted price.

Out of the box, there was a clear protective sheet of plastic attached to a tab that told me to remove before use. Under that glossy sheet of plastic is another sheet which is intended to stay. This second sheet has a matte surface that serves two purposes here. In addition to an anti-reflective effect, the texture feels better than a slick glass surface for drawing. Unfortunately, this textured surface sheet is already trying to peel from the monitor, leaving white bubbles all around the edges. I presume this texture sheet will eventually wear out and would need to be replaced, but I found no replacements from either Monoprice or Bosto. This is not a dealbreaker as I might never use it enough to introduce noticeable wear. The other consumable is the pen nib, and here Monoprice was generous. The manual listed six replacements, but I actually had twenty in my box.

Given the above, “acceptable but not great” is my general summary of the Monoprice Creator 22 at its clearance $170 price point. I wouldn’t have been happy if I had paid full price, probably telling myself “shoulda sprung for a Wacom One” if I had. But at clearance price it is a great entry point for me to mess around. If the novelty of pen input fails to capture my long-term attention (quite likely, as I know myself) I still have a serviceable ~22″ IPS 1080p HDMI screen. And if playing with pen input turns into something with actual utility, I may justify moving up to a Wacom Cintiq.

In the meantime, my first impressions here were based on some minimal time with my new toy. Roughly on par with the amount of time I got to play with a Surface Pen on a Surface tablet or an Apple Pencil on an iPad. So relative to a tablet-based drawing solutions, my initial highlights are:

Pro:

  • Will never have palm/finger interference problems.
  • Much larger surface area.
  • Textured drawing surface feels better than glass.
  • Affordable (due to clearance sale.)

Con:

  • Requires a computer.
  • Tangle of cables
  • Lower pixel density.
  • Dead-end discontinued product.
Roger Cheng

Monoprice Graphical Pen Display Clearance

For decades I’ve been interested in graphical pen displays that integrate a computer monitor with a graphics tablet into a single unit. There’s something very satisfying about drawing directly on screen and see it respond like real paper, yet with all the flexibility of working digitally. This technology started with professional gear costing thousands of dollars, and now we can get very close with commodity touchscreen tablets. Companies like Apple and Microsoft sell pencil accessories to turn their respective tablets into drawing sketchpads. Even though they haven’t worked well for me due to touch input interfering with stylus input, I was close to buying an Apple Pencil for my iPad. But then I saw Monoprice clearing out their inventory of graphical pen display.

As a company, Monoprice finds markets where incumbents thrive on huge profit margins. They find a contract manufacturer to build products under the Monoprice name and sell at a lower price. Monoprice rise to fame came from their HDMI cables that are a fraction of the price of Monster Cable (& friends). With that success, Monoprice used the same tactic to enter a wide range of markets ranging from 3D printers to camping equipment. But not all of their experiments were successful. Occasionally Monoprice decides a particular market is not worth further investment and it appears graphical peripherals have become the latest example.

For the past few years, Monoprice offered a line of graphical input devices. There were several graphical pen displays for several hundred dollars, and a few affordable drawing tablets (without integrated displays) for tens of dollars. These products undercut pricing of Wacom equivalents by at least 30%, but that still left them more expensive than I could justify buying. Especially when the software situation is an unknown given Wacom’s status as de-facto standards in this field: Every single artistic application will be compatible with Wacom devices, the same guarantee could not be made of Monoprice counterparts. Apparently enough people thought the same ad I did because in the past few weeks I noticed one of those tablets listed in a Monoprice clearance sale email. I checked Monoprice site and saw this was what remained of their product line.

Looking at that feature chart, my attention was drawn to the “Touch Screen” row. Given my past struggles trying to draw on a tablet screen that was also sensitive to my finger and hands, I liked the idea of getting a graphical display that was not a touchscreen. I don’t have to struggle with tuning “palm rejection” settings if the screen never cared about my palm to begin with! Unfortunately, the Monoprice clearance sale that brought this to my attention had discounted item #40443, the lone product on this chart that was also a touchscreen. Since that’s not what I wanted, I held off buying hoping one of the other items would be discounted in a later sale. My patience paid off when item #39945 was discounted from its $380 list price down to $170. This was too tempting to pass up so now I have a Monoprice Creator 22 Graphic Pen Display.

Roger Cheng

Disappointments in Cheap Digital Sketching

Despite my minimal artistic skill, I’ve long been fascinated by the possibility of using pen-based screen input for a digital drawing sketchbook. But I could not justify buying the good gear. Which meant a long string of experiments in more affordable approaches filled with frustration and disappointment.

My earliest experience was doodling with a small iPAQ (a Windows CE-based device) which used a sharp pointed stylus on a resistive touchscreen. These early PDAs like PalmPilot and Apple Newton barely sufficed for note taking. Given their low resolution, poor responsiveness, and no pressure level sensitivity, they compared poorly to contemporary Wacom drawing tablets.

Following Apple’s trailblazing iPhone, cell phones moved to higher resolution screens and capacitive touchscreens. Any phones with ambition to be competitive must deliver instant visual response to touch input, which is great for drawing responsiveness. But those capacitive touchscreens also eliminated the ability to use stylus for fine point accuracy. There existed “capacitive touchscreen stylus” but every unit I had tried merely delivered a wide touch surface as inaccurate as a finger. And finally, capacitive touchscreens introduced a new problem: I could no longer rest my hand on the screen for stability, as that would be treated as additional touch input messing things up.

I had high hopes for my next step: a Samsung 500T Windows 8 tablet with integrated pen digitizer using technology licensed from Wacom. The screen was much larger than a phone, and the Wacom digitizer delivered pressure data. Unfortunately, overall system responsiveness (not just to pen input) was intolerably slow. Just one of many problems which made that tablet into a big sack of sadness.

After that disappointment, I went upscale and upgraded my laptop to a Microsoft Surface Pro. It also had a Wacom-licensed pen input digitizer, and much faster hardware for better responsiveness. I could mostly use it as the digital sketchbook I always thought would be neat to have. The biggest complaint was that I never mastered a posture that would keep my hand off the screen. The Surface Pro had a basic form of “palm rejection” like the 500T: touch input could be disabled when the pen is active. But this required keeping the pen in range of screen digitizer. If I lifted the pen more than a few millimeters, it would fall out of range. Which meant the tablet reverted to touchscreen input and immediately engulfed in chaos caused by my palm.

After that era, consumer level tablets moved away from Wacom technology. Microsoft’s Surface Pen has since switched to a different technology, and Apple launched their own take with the Apple Pencil. In addition to pressure sensing, both of these technologies added the ability to sense tilt angle, literally opening up a new dimension in digital sketching. When I tried floor demonstration units at Best Buy and Apple Store, I was quite impressed. But with my drawing style and posture, they both still suffered from unwanted hand input. This will be an ongoing problem on anything designed primarily for finger touch input. What I wanted is a product primarily designed for pen input and ignores my fleshy parts, but that is a very limited product niche. I didn’t think I could ever justify the cost of that niche until Monoprice decided to clear out some inventory.

Roger Cheng

Computer Pen Input Has Always Been a Novelty to Me

One skill unique to digital artists is the ability to draw on a tablet while looking at a separate screen. I can’t think of any other art medium with such a separation between an artists’ hands and their workpiece. Towards the end of The Line King: The Al Hirschfeld Story(*) documentary was a clip of Hirschfeld trying a drawing tablet. This lifelong master of pen and paper struggled to make a basic sketch, saying “It’s almost impossible to control […] I supposed it is possible to control it, it’d just require another lifetime to do it, that’s all.“

The separation of drawing tablet and scree is a curious evolution of computer input devices, because things didn’t start out that way. Decades ago, specialized computers had light pen input devices that allowed pointing and drawing directly on screen. It made it all the way down to consumer (or at least business) level hardware with provisions on early IBM PC video cards.

I guess early product designers assumed people wouldn’t put up with trying to manipulate input that was separate from its screen representation. If so, the computer mouse proved that assumption wrong. People were indeed willing to learn to manipulate with physically separated input devices when there is enough productivity at low enough cost to be worth learning a new skill. The same basic arguments applied to artistic creations. Given the advantages of digital creation over physical media (ease of edit, archive, copy, transmit, etc.) some artists were willing to learn to draw on a desktop tablet while looking at a screen.

But human beings still prefer to have visual feedback physically corresponding to manipulation input. And thankfully technology has advanced enough to give that back to us, at least in some contexts. We interact with modern slab-faced smartphones with its touchscreen instead of a mouse, and artists can get graphical pen displays that integrate drawing tablet capability with a computer screen. I remember when the technology became a practical product and watched an artist demonstrate the first Wacom Cintiq. While being well aware that most of the skill is in the artist, I was nevertheless blown away by how immediate, responsive, and thereby intuitive drawing on a graphical pen display was. And I was not alone, as that first Cintiq was successful enough to launch a full and still evolving product line of graphical displays. It was extremely expensive, but it meant an artist no longer had to draw on one device while looking at another device. A capability well worth the cost for artistic creators.

As a software code jockey with no artistic skill to speak of, I had no justification to spend thousands of dollars on a graphical pen input display. It was just a novelty that I loved to play with whenever I had the chance. I got close with a few Windows and iOS tablets, but it wasn’t quite the same.

(*) Disclosure: As an Amazon Associate I earn from qualifying purchases.

Roger Cheng

Notes on Siglent SDS1104X-E Oscilloscope User’s Guide

I have some basic ideas on how to use an oscilloscope, but I’ve never had one of my own until I bought one during this year’s Amazon Prime Day sale. Given its complexity and cost, I thought it would be a good idea to invest some time into Reading The Fine Manual. This did not start well, as there was only a Quick Start Guide in the box, featuring this particular gem:

These symbols may appear on the product: (But we won’t tell you what they mean!) Thanks, guys. Despite such minor mistakes, the quick start guide seems fine if perfunctory. I was moderately annoyed that they used the same manual for two-channel and four-channel versions of this scope, so I would occasionally look at something that made no sense until I realize it was about the two-channel. That annoyance aside, I learned valuable things like adjusting probe compensation as part of unpacking and initial setup (they were all slightly under-compensated but easily resolved with the procedure) but most of the other descriptions assumed I already knew how to use an oscilloscope. I was worried until I saw a note saying I could find more information in the User Manual.

Okay! A real User Manual exists, even if it isn’t in the box. I went hunting online and found my answer on Siglent NA (North America?) document repository where I could find the User Manual (and many other guides) in PDF format under the SDS1000X-E-Series section. It has the same annoyance of using one manual for both 2- and 4-channel versions, but now with a lot more useful detail.

  • One valuable thing I learned and need to keep in mind is that most knobs on this oscilloscope are like the quadrature encoder knob I took apart: there is a button press in addition to rotation. If I’m poking around looking for a feature, it might be a knob press.
  • I like the idea of the “Auto Setup” button. It is advertised to looks at the channel’s signal and choose an appropriate vertical and horizontal scaling. Sounds like a counterpart to “auto ranging” capability on a multimeter, I hope it will turn out to be as useful as it sounds.
  • These scopes came with probes that have a switch to toggle between 1X and 10X attenuation. It appears the probe has no way to communicate its current setting to the scope, I have to tell the scope. Something to keep in mind and check when things make no sense.
  • When I zoom out to a longer timescale, there’s a threshold where the cheap DSO-138 would automatically switch to showing data in a horizontal scrolling display. After reading this user’s guide I know it is called “Roll Mode” (Page 34) here and it’s something I can choose to toggle on/off with a button, independent of timescale.
  • I frequently try to adjust display timescale on the DSO-138 so I could zoom in and out to look at various features. Now, I have an actual zoom function (page 35) so I can keep the longer timescale waveform on screen simultaneously with a short timescale subset of the same wave.
  • DSO-138 would frequently fail to show fast blips. If I need to see peaks of very brief signals, I can choose to display “Peak Detection” mode. (page 46).
  • Typically having multiple channels mean multiple lines all graphed against time, but setting “Acquire” to “XY” (page 49) allows graphing one channel versus another instead of time. There will be some vector graphics fun with Lissajous curves in the near future.
  • It seems like half of the manual goes into depth on what each of the trigger modes do. I will need to re-read this section several times. Eventually I should be able to recognize which situations are best fit for certain trigger modes.
  • I was very excited to read about Video Trigger: it sounds like the oscilloscope knows what NTSC composite video signals should look like and can trigger on specific parameters or fields. Once I master this mode, I foresee it becoming extremely valuable for debugging my ESP32 composite video output library.
  • I had no idea “Measurements” (page 130) are something oscilloscopes can do now. So instead of reading the screen to see how much time is represented by an on-screen grid division, and calculating the period of a waveform, and from there calculating the frequency… now the scope has measurement tools to do all that math for us. Wow, fancy!

Judging by what I’ve learned from this User’s Guide, I’m very happy with the potential usefulness of my oscilloscope purchase. I hope it will prove to be actually useful as I learn to harness its abilities.

Roger Cheng

Finally Bought a Real Oscilloscope

An oscilloscope has been on my workbench wish list for years. I had been limping along with a degraded DSO-138 kit, occasionally wishing for something with more channels, or more bandwidth, or just the ability to measure voltage levels accurately. The key word being occasionally. I haven’t felt that I would use an oscilloscope enough to justify the expense. But when this year’s Amazon Prime Day rolled around, the memory of deciphering multi-channel signals was fresh on my mind, and I clicked “Buy” on a Siglent Technologies SDS1104X-E Oscilloscope. (*)

I had actually been eyeing a Rigol DS1054z(*), which had become a very popular entry-level oscilloscope for hobbyists. It is sold far and wide including my favorite vendor Adafruit, and its popularity meant plenty of online resources. From basic beginner’s “Getting Started” guides to hacks for unlocking features. Ah yes, those features. They were a big part of why I hadn’t bought the Rigol: it really sours me on a company when they would hold features for ransom even though all of the hardware is already present. Sure, I could visit questionable websites and generate codes to unlock those features without paying for them, but just the idea of buying from a company that would do such a thing turned me off.

While the Siglent oscilloscope did have a few paid upgrade features, they all involved additional hardware not already onboard. This made the concept more palatable for me. For reference, they were:

  • WiFi capability. The scope comes with an Ethernet port for network connectivity. Wireless comes at extra cost for the software upgrade in addition to the cost of a supported wireless adapter. I prefer wired Ethernet so I did not care.
  • AWG (arbitrary waveform generator) capability requires extra hardware in the form of Siglent SAG1021I. (~$175 *) So far, my waveform generation needs have been very basic. So basic, in fact, that I wrote a HTML app to cover my needs. I don’t think I’ll miss this feature.
  • MSO (multi-signal oscilloscope) capability requires a Siglent SLA1016 (~$330 *) which adds sixteen additional digital channels for logic analysis. Between the four channels already on board the oscilloscope (which already has logic analyzer functionality without paying to unlock as would a Rigol) and eight channels on my Saleae, I think I’ll be fine without the MSO add-on.

One thing that made me frown was that the AWG and MSO addons connect by something Siglent called “SBus”. Proprietary expansion ports are nothing new, but they chose to use a HDMI connector for the purpose. With a warning that plugging in actual HDMI devices would damage the oscilloscope. Gah! I see the economic advantage of using an existing high bandwidth connector already produced at high volume, but the resulting user experience sucks. Since I don’t plan on making any SBus upgrades, I will try my best to ignore that not-HDMI port.

This oscilloscope cost more than a Rigol DS1054z, though it is technically cheaper because many of Rigol’s paid add-ons were on the Siglent without extra charge. The Prime Day discount closed the price gap enough for me. Once it arrived, I dug into the manual eager to learn about my first real oscilloscope.

(*) Disclosure: As an Amazon Associate I earn from qualifying purchases.

Roger Cheng

Windows SFC (System File Checker) Revived Explorer

On a computer running Microsoft Windows operating system, the executable application explorer.exe is very important. It handles the start bar and is the starting point for almost every activity on the system. Its core position means if Explorer breaks for any reason, it’s very hard to do anything else on that system! Yet its complexity and wide span of responsibilities also means having wide exposure to things that go wrong. Countering this risk, Explorer has recovery measures built-in as well. If it freezes up, there’s a watchdog time to restart the process. If it should crash, there are mechanisms to relaunch it. And if the relaunch immediately leads to a problem, it relaunches with gradually decreasing capability until it finds a configuration that is stable enough for the user to go in and figure out what went wrong.

This happened to my Windows machine. Something went wrong and Explorer went into a failing loop that restarted once every 10-15 seconds. This continued for several minutes (Explorer restarting a few dozen times) until it stabilized in a very reduced configuration that was unfortunately pretty difficult to use. The Start menu is missing, but at least Window+E shortcut key still worked to open File Explorer. That allowed me to launch diagnostics tools, though I had to use my phone to search for their paths on the system so I could find them in File Explorer.

The first stop to diagnose Windows problems is the Event Viewer, which I had to launch from File Explorer by double-clicking C:\Windows\System32\eventvwr.exe. Clicking the root node in the tree “Event Viewer (Local)” will show a summary of events. My system showed over four hundred “Error” events in the past hour, an obvious place to start looking. Expanding that tree took me to a list of those hundreds of Application Error logs, here is one example:

Unfortunately, the details of this Application Error were not scaled for high DPI displays and pretty unreadable in that screenshot, so here is a copy of the text shown inside that “General” tab:

Faulting application name: explorer.exe, version: 10.0.22000.832, time stamp: 0x8947d46c
Faulting module name: wintypes.dll, version: 10.0.22000.778, time stamp: 0xb903efeb
Exception code: 0xc0000005
Fault offset: 0x0000000000022b20
Faulting process id: 0x1f44
Faulting application start time: 0x01d8b13598ecfee1
Faulting application path: C:\Windows\explorer.exe
Faulting module path: C:\Windows\SYSTEM32\wintypes.dll
Report Id: d0fb6df2-0bad-45bd-aff3-dee9c438b3d2
Faulting package full name: 
Faulting package-relative application ID:

Looks like explorer is pointing a finger at its dependency wintypes.dll. Unfortunately, there isn’t enough data here to tell us if the problem is in wintypes itself or a dependency in turn. But at least it is a narrower scope than explorer, whose scope covers damned near everything. A search for “wintypes corrupt” found many websites advertising “Download wintypes to fix your system!” But downloading replacements for Windows system executables off the internet is a recipe for security disaster and I’m not going to do that. There were a few promising diagnostics steps, the one that was eventually successful was this Microsoft community forums thread.

The repair procedure was to first run the Deployment Image Servicing and Management tool (DISM) to ensure my system has a valid copy of the system image. Followed by running the System File Checker tool (SFC) to scan through my Windows system files. SFC will compare what’s on my system against the system image archive. If a mismatch is found, SFC will replace the corrupted file with a clean copy from the system image. These are system-level administrative tools. In order to run them, I had to launch an administrator command prompt from File Explorer. (Right click on C:\Windows\System32\cmd.exe and select “Run as administrator”)

It took several minutes for those tools to complete. After SFC reported scan and repair was complete, I rebooted my system. And this time, Windows explorer started with full functionality. Success! I went back to take a look at the log file (C:\Windows\Logs\CBS\CBS.log) and searched for mentions of “repair”. I found these two lines:

2022-08-15 23:15:56, Info                  DEPLOY [Pnp] Corrupt file: C:\Windows\System32\drivers\bthmodem.sys
2022-08-15 23:15:56, Info                  DEPLOY [Pnp] Repaired file: C:\Windows\System32\drivers\bthmodem.sys

Huh. The Bluetooth communications driver caused me all this grief? It is indeed part of my Windows system, because I’ve been playing with Microsoft Phone Link and it connects to my Android phone over Bluetooth. I didn’t think a problem with this file would bring down Explorer, but now I know it can. I also don’t know how this file got corrupted to begin with, and that might be important to know if it should happen again. For now, I’m happy my computer is back up and running.

Roger Cheng

Cutra Wondercutter S Is Expensive but Also Really Cool

Cutting apart a well-sealed Makita battery cartridge was a big milestone for my Cutra Wondercutter S purchase. It was the first big project that would have been unreasonably difficult to do with any of the other tools I have on hand. With this milestone, I am confident the tool has a unique niche. Does it justify the cost? That’s a harder question whose answer will depend on usage and budget.

I bought my Wondercutter from Micro-Mark after doing a bit of window-shopping research on the item. I was simultaneously optimistic and skeptical that its capability will live up to its high price tag. When I opened up the box, I saw that a small piece of test material was included in the package for us to take our first Wondercutter cut. It is some sort of foamcore material, and it presented the best-case scenario. It was thick and durable enough to take effort cuting with my X-Acto #11 blade and even then, the cut wasn’t very clean. But Wondercutter sliced through like hot knife through butter on this material and also leaving a clean edge.

I then tried it on materials that I expect to work with. A scrap sheet of 3mm acrylic took more effort to cut than the test material, but it did cut just the same. The sight of acrylic melted along the cutting edge was accompanied by the smell of heated acrylic. Some of the cutting action must have been from heat melting the acrylic and not just the ultrasonic cutting action. This sample indicated the Wondercutter is no substitute for a laser cutter for clean sharp edges. But I don’t have a laser cutter, so a Wondercutter will work in a pinch for quick acrylic cuts at home.

The next tests were done on various failed prints from my 3D printer. Printed PLA cut more easily than acrylic, with similar sight and smell of melting. PETG was more difficult to cut and, thanks to its higher temperature resistance, there was only minimal melting. Another characteristic of cutting PETG is that I could occasionally feel vibration through my fingers, like a dental cleaning session which can be unpleasant. Printed support materials were easier to cut than solid shapes, as they are intentionally printed to be weaker. For this purpose, using a Wondercutter is faster than my X-Acto blade, but by itself is not necessarily enough to justify the expense. I’ve read that the Wondercutter is great for cutting resin support structures, so I look forward to that if I ever get into resin-based 3D printing.

The final test was done on a circuit board. Some Wondercutter vendors claimed it can cut circuit boards, though neither Cutra nor Micro-Mark mentioned it either way. This was a no-go. While the blade could make its way through FR4 fiberglass, cutting progress came to a literal screeching halt as soon as the blade touched a copper trace. It damaged the blade, so I had to replace it. Thankfully there is a pack of 40 replacement blades in the box. I hope it didn’t damage the transducer.

With this bit of testing under my belt, I start pulling out the Wondercutter for various teardown projects. The first big win was cutting the ultrasonic transducer mount free from an oil diffuser. It went much faster than an X-Acto blade and less messy than a Dremel cutting wheel. Some smaller jobs were sprinkled here and there, like cutting the battery tray free from various devices so I could use them in my own projects. Or the plastic-encased bearing assembly of an evaporator fan. The Makita battery project is similar to the oil diffuser transducer: much faster than an X-Acto blade, and less dangerous than a Dremel cutting wheel. The Wondercutter is an expensive investment an order of magnitude more expensive than a Dremel and two orders more than an X-Acto blade. But every once in a while, I find myself in need of its capabilities and glad that I now have it.

Roger Cheng

Sanyo CCB Close Enough to SPI For Logic Analyzer

I’m examining the control signals for a Sanyo LC75853N LCD driver chip, which uses a Sanyo proprietary protocol they call CCB. (Computer Control Bus.) It’s popular enough that I could find CCB reference material online, but it’s not popular enough to be natively understood by Saleae’s Logic Analyzer software. Beyond Saleae’s set is a list of Community Shared Analyzers but Sanyo CCB didn’t make the cut there, either. Those additional analyzers were written using Saleae’s Protocol Analyzer SDK so there is the option to write one for CCB. For the purpose of initial experimentation, though, their default SPI analyzer is close enough.

Before we even try using the SPI analyzer, we can look at the raw data. CCB transmits the peripheral address before raising CE. Here I can see 0x42 hexadecimal, or 0b01000010 binary. (The white binary numbers were not part of Saleae software, I drew it in afterwards.) In an unfortunate bit of coincidence, this binary value is symmetric so it alone couldn’t tell us if CCB transfer least-significant bit first or most-significant-bit first. According to spec, it is least-significant-bit first. Seeing this gave me the confidence I’ve wired up everything correctly for further probing.

The clock pulses measured out to be in the ballpark of 400kHz, which I can probably work with. But more importantly, I was relieved to see that the clock pulse widths varied somewhat between transmitted bits. This is encouraging because it meant the protocol is graceful under irregular clock pulses, making it more likely I can successfully communicate using CCB in software. Which is great because I don’t have dedicated CCB communication peripheral hardware.

The next step was to activate SPI analyzer with the following parameters. The biggest difference between CCB and SPI is the behavior of CE line, and thankfully Saleae’s SPI analyzer can be configured to ignore CE. (“Enable” set to “None”.) I set the SPI analyzer options to the following values to decode all the values regardless of CE status:

SPI Analyzer OptionValue
MOSILCD-DI
MISOLCD-DO
ClockLCD-CLK
EnableNone
Significant BitLeast Significant Bit First
Bits per Transfer8 Bits per Transfer
Clock StateClock is High when inactive
Clock PhaseData is Valid on Clock Trailing Edge
Enable Line(Doesn’t matter when “Enable” is “None”)

Now the software can decode data for us. This time, the decoded values 0x42, etc. in this image was drawn by the software.

This was the start of the very first data transmission after I applied power to the tape deck. Which is why CLK started as low even though it is normally high when inactivity. When Enable is set to None, I see all the data regardless of CE status.

First question to answer: the B in CCB is “Bus”. How many devices are on this bus? Taking advantage of the difference between CCB and SPI, I can tell the SPI analyzer to decode just the CCB address by saying CE is Active Low:

SPI Analyzer OptionValue
EnableLCD-CE
Enable LineEnable is Active Low

The decoded values on LCD-DI were all 0x42, which tells me the LCD control chip is the only peripheral on this bus, which makes things simpler. I won’t have to worry about reading data intended for the wrong device. And once I decided I didn’t have to worry about different addresses anymore, I can switch the SPI Analyzer over to Active High CE. This will cause the analyzer to ignore addresses (since I expect them to all be 0x42) transmitted while CE is low and decode just the data.

SPI Analyzer OptionValue
EnableLCD-CE
Enable LineEnable is Active High

Within the ~3/4 second of the faceplate getting power, something is transmitted repeatedly approximately once every 5 milliseconds. Zooming in, I see they are three consecutive CCB transmissions to address 0x42:

Address1234567
0x420x000x000x000x000x000xC00x10
0x420x000x000x000x000x000x000x80
0x420x000x000x000x000x000x000x40

Next: compare this captured data to the LC75853N datasheet to see if they make sense.

Roger Cheng

Clothes Iron Assistant for Hot Air Rework Station

I’ve been gradually building up my skills and tools working with surface-mount technology components. After the most recent episode of frustration trying to rework components without a hot-air rework station, I went ahead and bought a simple starter unit on Amazon. (*) This is quite similar to the hot-air rework station I got to play with earlier, and I’m operating under the assumption that something simple is far better than nothing at all. It is certainly less destructive than the blunt hammer of a paint removal heat gun. Using my new hot air rework station, I was able to salvage more red LEDs from my pile of salvaged electronics and install them on the Mr. Robot Badge. The job is a lot easier once I had the right tool! The result is not quite perfect, because my salvaged LEDs are slightly different shade of red and a touch dimmer, but it is running and it’s all red.

Then I saw this on Twitter:

Hey, I have one of those “SMT stations” too! In my pile of household appliances awaiting teardown is a retired clothes iron. Something is wrong with its temperature regulation circuit. It would heat up to the requested temperature then turn its heater off. But then it would cool far more than acceptable before it would start heating back up again, and it never heats all the way back up to the requested temperature before turning off again. In short, it was no longer useful as a clothes iron and I thought I’d take it apart eventually. Due to my confusion when looking at a simple coffee maker circuit, I’m not optimistic I could fix the iron, it was just to see if I could learn anything.

But now I have something else I can do with it instead of taking it apart. Reading the rest of the above Twitter thread, I confirmed my understanding that unmodified clothes irons do not get hot enough to melt solder by themselves. What they can do is assist the hot air gun by keeping the working area warm. This helps the hot air rework station work better by reducing the amount of heat dissipated away uselessly.

So I set up my own “SMT station” and placed upon it the mainboard from an old retired Roomba Red. The proprietary CPU (or at least customized enough to have an iRobot logo on it) was too big for me to remove with the hot air station alone, but with the iron keeping the entire circuit board warm, I was able to melt the solder and remove the chip intact. There’s a small chance this chip is still usable, not that I plan to do anything with it.

Also, the same temperature regulation problem that made me retire this iron also made it bad at maintaining circuit board temperature. I can work around it by unplugging the iron and plugging it back in to reactivate that initial heating period. Far from ideal, but good enough for today’s learning exercise.

Then I revisited my previous hot air experiment failure: the CPU from the mainboard of an Acer Aspire convertible Windows tablet. This time I was successful! Sort of.

It was far more difficult to remove than the Roomba CPU. I damaged a corner of the module with my tweezers by yanking too hard, it’s been bent upwards so definitely destroyed. But this was great practice and I’ll keep at it as I do my teardowns. Eventually I’d like to reach a point where I can salvage and reuse something more complex than a LED.

Roger Cheng

Window Shopping: Cutra Wondercutter Ultrasonic Knife

During an online video meetup with some makers, I learned that consumer-level ultrasonic knives exist. One member of the meetup was taking theirs apart for some reason I’ve now forgotten, I just remembered his quick demo of the handheld cutting blade cutting through a 3mm thick sheet of acrylic. It wasn’t exactly “hot knife through butter” (maybe cold knife through refrigerated butter) but certainly far superior to what I could do with my X-Acto #11 blade.

I had known of ultrasonic tools in the medical field, specifically dental tools. I also had some idea they existed in the realm of industrial manufacturing equipment. But something I could conceivably buy for my own workbench? That’s news to me. Unfortunately, the person on the video wasn’t able to give me much information to go on, so I started searching for “ultrasonic knife” from my usual list of tool vendors. Unsurprisingly, I got a hit at McMaster-Carr: Item #3415N11 Fast-Cutting Ultrasonic Precision Knife. Visually, this looks like the same device I saw being disassembled at the meetup. But McMaster-Carr didn’t give me very much information on this device, not even some things I consider basic like make and model for further evaluation.

A search for “Ultrasonic Knife” on Amazon would give me several multi-thousand-dollar industrial machines, and also this listing. (*) Titled “The Wondercutter S” it looks right, but this listing felt odd for several reasons. The brand is listed as “MICRO-MAKE” but there’s nothing else by that brand name. There is also a logo on the device absent from the McMaster-Carr listing. It is stylized so I couldn’t quite decipher it to letters, but it is definitely neither “Wondercutter” or “MICRO-MAKE”. This listing didn’t give me the confidence I needed to commit several hundred dollars, despite Amazon Prime guarantees.

Continuing online search, I also got a hit at Home Depot which was a mild surprise. I had not associated the big orange DIY home improvement store with high tech tools. From this listing I got a brand name “CUTRA” which explains the stylized logo I couldn’t read earlier.

Now that I have a brand name, I found its company site and their own product site. It appears to be a Korean company and I finally got the specifications I could sink my teeth into. There were also a lot of demonstrations videos on what this device could do, the one that got my attention was cleaning up supports for 3D printing. I’ve never enjoyed cleaning up supports, and I’ve had a few dangerous close calls with my trusty X-Acto blade doing so. A couple of hundred dollars is a significant investment, but if it saves me a single hospital visit that would make the item worthwhile.

From this site I also learned that Wondercutter was crowdfunded as an Indiegogo project back in 2017. Well, I definitely missed the early bird backer special of $258 USD! I just have to pay retail now. Elsewhere on the site I saw I could order direct from Korea, but they have signed an official distributor for United States: Micro-Mark. Now this is a name I know from my scale plastic model days! I used to be a very good Micro-Mark customer buying Tamiya paints (they’ve all since dried out) and small model-building hand tools (I still use some of them).

Well, at least this explains the mystery branding of “Micro-Make” on that Amazon listing, it was a typo of Micro-Mark. There is actually a Micro-Mark storefront on Amazon (*), but with only a subset of the full catalog. For example, they sell replacement Wondercutter blades (*) but they don’t sell the Wondercutter itself on Amazon. Why would they leave it open for an imposter vendor to take away potential Amazon sales? That’s a curious business decision. Micro-Mark claims to be the exclusive distributor for North America, and I can see they are listed as vendors on sites like Walmart. But I’m not sure what’s the point of going through Walmart (or Home Depot or Amazon) if Micro-Mark is actually the distributor. It seems to make more sense to order one direct from Micro-Mark’s own website.

(*) Disclosure: As an Amazon Associate I earn from qualifying purchases.

Roger Cheng

Disappointing Budget Keyboard Protector

I bought an Apple M1-powered MacBook Air laptop and bought a laptop cover to protect it against scratches and mild impacts. The hinge mechanism of this machine presents a challenge for case makers: there’s not much clearance to work with. Letting the lid open past 90 degrees is a tradeoff between protection and adding stress to the laptop joint. The first case I tried(*) had a very noticeable “bump” that my lid had to push past, and I worried about long term durability of the hinge under that stress. I then tried a different vendor (*), whose back corner (where the hinge is) is thinner and potentially more fragile but it put far less stress on the laptop hinge joint. Even though I could feel a bit of a bump as I open the lid, I decided to stay with this case which has worked very well over the past three months.

The same could not be said of the keyboard protector membrane that came bundled with the case. Historically I had not used such things and its inclusion did not factor in my case purchase decision. But since I got it anyway, I thought I would give it a try.

The good news: it is very good at its primary purpose of protecting the keyboard from small particles, like food crumbs that get dropped when I use the laptop while eating.

The neutral news: I had been worried about how it would change the keyboard feel. I’m a touch typist and particular about how my keyboard feel as I type. (I never got a “butterfly keyboard” MacBook for this reason.) And while the keyboard tactility definitely feels different, I did not find it objectionable. At least, at first.

Now we come to the bad news, part one. After a few months, the membrane material starts exuding some kind of liquid. There doesn’t seem to be any odor and there’s very little of it, but it starts leaving blackened blotches where the liquid touches the keyboard and made the membrane more transparent than other areas. This looks bad and I worry about getting that stuff on my fingers. The membrane never gets tacky enough to stick to my fingers, but it still doesn’t feel good to the touch.

The bad news, part two, is in the flexible nature of the material. After using it for several months, the material has started to stretch. If I try to keep the shape aligned with the keyboard, the stretched portions would bulge up which looks bad. Plus when the laptop lid is closed, these bulges touch the screen and leave behind little spots of the aforementioned liquid.

If I smooth out the bulge so the membrane lies flat, its shape no longer align with the keys. This is especially obvious in the arrow keys in the corner, where the designated bumps for arrow keys are now 2-3mm further to the right than the actual keys themselves.

Due to this unsightly shape, black spots, and sticky sensation, I’m going to throw away this membrane. It was a low-cost test so no big loss. But if I ever get serious about putting protectors on my keyboard I know to either (1) look for higher quality material or (2) budget for frequent replacement.