The Ware for October 2021 is shown below:

This one should be much easier to guess than last month’s ware; it’s strategically cropped to add a tiny bit of challenge to it. I’ll add some more views of the ware once we’ve got a correct answer.

I was really struck by the quality of the photography of this ware. jackw01 submitted a series of wares (this one included), which I’ll be sharing over the coming months. I have a lot to learn about photography, and I thought it might be of general interest to hear his answer on how this ware was photographed:

[The ware] was shot with a Fujifilm X-T30 using a 40-year-old Canon FD mount 50mm prime lens with a set of 10+16mm macro extension tubes. If you’re unfamiliar, extension tubes are basically a way of achieving macro photography without having a dedicated macro lens. They decrease the minimum focusing distance and increase the magnification of a lens according to the ratio between the length of the extension tube and the focal length of the lens, producing 1:1 magnification when the extension tube length equals the focal length. When using both the 10 and 16mm extension tubes together with the 18-55mm lens, my usable field of view is anywhere from 25-50mm wide depending on the focal length and focus distance with the subject ~0-20mm away from the lens, and with the 50mm prime, the field of view is 30-45mm wide with the subject 70-90mm away from the lens. When using extension tubes, I would highly recommend using manual focus, even though most extension tubes have the electrical connections to support autofocus – the autofocus algorithms on most cameras are just not tuned to work with extension tubes and often won’t focus at all or won’t produce as sharp of an image.

Thanks again to jackw01 for sharing this ware, and for sharing your know-how on photographing them!

Update Nov 5: the ware has been guessed as much as it can be (there is no expectation to guess the specific panel) so I thought I’d share these other images of the ware!

I think one of them might be be a great addition to the entries on TFTs or LCDs on Wikipedia.

I mentioned it to jackw01, and he said he’d be OK with sharing the photos using a Public Domain license and having them added — I’ll give that a try when I get a spare moment, if someone doesn’t get to it before I do!

Wow! The ware for September 2021 was a real stumper. To be honest, when Marcan showed me the wares, I had similar instincts to most of those who entered guesses — I was entirely thrown off by the spendy choice of components combined with the huge array of multimedia connectors. Before seeing this ware, I never associated “karaoke” with “expensive electronics”; well, maybe this is a datapoint on how lucrative a business karaoke must be.

TL;DR: The Ware is a Joysound F1, and the winner is Thorkell, who finally managed to piece together the puzzle the day before the contest was scheduled to end. Calvin was actually the first to guess the correct genre of the machine, but Thorkell came as close as possible given the provided images to correctly identifying the make and model (not enough info was in the photos to reveal if it was the f1 or the fR model). Congrats, email me for your well-earned prize!

Marcan, who you should definitely follow if you have any interest in Linux, reverse engineering, and/or the M1 from Apple (he will be live-streaming an Asahi Linux bring-up on Nov 1!), also kindly provided this very detailed write-up on the ware:

Marcan’s Insights on the Joysound f1

This is a Joysound f1 (codename “Ken”), one of the more popular karaoke machines used in shops across Japan. It was released in 2012. You’re likely to have used one if you’ve been around more than once or twice. There are, to my knowledge, no previous teardowns of these machines on the internet, so I was intrigued by how they worked. I picked one up in an auction, and I will say I was not expecting what I saw when I tore it apart!

These machines are “networked karaoke” and periodically call home to update the song database (and require an ongoing subscription to work), but they are designed to work on anything from FTTH to periodic dial-up connections, so they need to have all the data locally. To that end, there’s a big 3TB hard disk with (almost) the entire song database (it is subsetted differently depending on your network speed, e.g. you won’t get many background videos on dial-up, and songs published as user submissions are always streamed on-demand and available only on broadband configurations). The HDD also contains firmware, updates, and anything else that needs pushing out to machines. As of the August update they seem to be using 2.5TB of the storage capacity, so it’s pretty tight already!

The architecture is bizarre. The Tegra 2 SoM is the main processor of the system, running Linux4Tegra (Ubuntu 10.10 ARM32) with good old Xorg; it is in charge of the main karaoke playback, networking, updates, remote control service, etc. Interestingly, it can also peer with another machine to serve its data over the network, which is useful when an HDD dies, to avoid having to take the machine out of service entirely. It boots off of a ramdisk loaded off of the HDD, and quite impressively, the entire rootfs is less than 150MB uncompressed. Control is usually via external touchscreen or tablet remotes, that connect typically via an external Wi-Fi access point and network, but can also use the internal Wi-Fi card in ad-hoc mode.

The entire audio subsystem is offloaded to the Roland board, which has a full MIDI synthesizer (for the karaoke; most songs are MIDI, although AAC is also supported and decoded by the Tegra before being pushed to the Roland as PCM) and DSP engine for Mic effects (reverb, voice changer, anti-howling, etc). In fact, it even has a fancy system for using an external microphone to measure the acoustic characteristics of the room and automatically compute a DSP profile. On top of the core MIDI patches, the Tegra also uploads an extra set of very high quality bass and drum samples to the Roland via the USB connection. Talk about high-end MIDI!

All the I/O is for things like the mics, external background music/video sources, instrument inputs (e.g. you can add a guitar preamp frontend), and auxiliary outputs. This is the first machine in its range to have HDMI, so it only has a single output; a newer revision called F1v added an HDMI input and dual HDMI outputs, to allow for HDMI idle/background video feeds.

The front touch panel is driven by the Marvell Armada SoC and is its own system running Android Eclair. It gets an SD feed of the main system’s video to display when idle, and it can composite its menus on top. It is otherwise a completely standalone system, with its own song metadata database updated from the main unit, etc. It communicates with the main system chiefly via USB networking and some of the same APIs that external Wi-Fi remotes would use. This is the first machine from Joysound to have an embedded touchscreen interface, and basically what they did was take the existing JR-300 “Mary” stand-alone touchscreen remote and embed it into the main unit. They call it “pamary” (Panel Mary, presumably). Amusingly, the ad-hoc Wi-Fi dongle is connected to the Armada, not the main SoC, so external remotes connected in this way end up routing through it into the main SoC. No idea why they did it like that.

The next generation (Joysound MAX “Zeus”) is basically an iteration of the same architecture. They ditched the Tegra 2 and replaced it with a Renesas R-Car-H2 (keeping with the automotive SoC theme…) and the distro is now Yocto-based, but the front panel SoC remains separate. The Roland board is much reduced, presumably using newer more integrated technology. The HDD is now 4TB. There is also a newer MAX2 version, which I don’t have, but I don’t expect it to be much different either.

Thanks for playing this Name That Ware! Hope you enjoyed it!

The Ware for September 2021 is shown below.

This ware was kindly contributed by @marcan42. I’m really impressed at the quality of the camera work for the wares!

These are just a subset of the boards from the ware, but I suspect it’s more than enough to get a positive ID. The digital board is a bit more telling, but I find a well-designed analog board to be too attractive to pass up posting. I also love the slightly browned regions evidencing the hard life that the linear regulators experienced, filtering out all that power supply junk to create a clean, smooth power rail suitable for audio. If nobody can figure this out, there’s another board I’ll add to the set which might be helpful.

Bonus points to anyone who can come up with a good theory for the ~3mm 45-degree angle cutouts sprinkled around the digital board’s power plane pours. I can’t come up with any consistent rhyme or reason for them to be there, so maybe the engineer was just going for style points? Nothing wrong with going for style, but perhaps I missed the memo on some sort of black magic with respect to detuning resonances in power planes, or something like that.

Update Oct 7
It’s been interesting to see where the guesses are going, but unfortunately nobody is getting close. So, here’s another important board from the ware:

The Ware for August 2021 is a DirectTV receiver model D10. As I had surmised, it was quite easy to guess the nature of the ware, based on obvious clues such as the smart card slot, tuner circuit, and the general TV-receiver-esque vibe about the board. One does not integrate a line voltage power supply and an RF front end into a single circuit board unless one is sure to sell millions of them; the regulatory overhead is just too great otherwise.

Anyways, to make the ware a bit more challenging I had asked for comments specifically about the unpopulated cut-out section of the board. Participants correctly recognized from the get-go that it should be an accessory to the receiver, likely related to a remote control. GotNoTime even went so far as to call out a part number, the KESRX01 ASK receiver chip (as listed in the schematics that were previously linked by Eben Olson). While the schematics share the chip designator U33101, unfortunately, the KESRX01 is a 24-pin package, but the board has a 28-pin footprint. Thus, it’s quite likely these are closely related schematics, probably made by the same OEM, but they also cannot be an exact match.

This motivated me to poke a bit more; a bit of googling around revealed a good match for the footprint and functional category: the TDA5200. Below is a juxtaposition of the reference schematic from the datasheet along with the layout:

And just for quick reference, here’s the pinout and internal block diagram of the chip:

I got as far as confirming that the power supply lines match up, and that the crystal and RF lines seem to make sense, which means it’s likely to be this chip or at least a closely related one. If I had to build up a functional circuit using this board blank and no schematics, I’d say it’s likely to succeed, modulo a couple patch wires and a bit of antenna tuning.

My propensity to figure out what goes into the blank spots of a circuit board is a quirk that stems from having to, at times, reverse engineer devices that have had the part numbers sanded off or obscured for “security” reasons. That’s probably part of the reason I’m drawn toward empty spots on circuit boards; they are good practice for looking at context and circuit traces to figure out part numbers. It’s like circuit board karaoke!

I guess this month there isn’t a winner, since nobody quite got the answer I was looking for, so I’ll leave it at that (although I am probably partially to blame for not being clearer as to what I was hoping to see as an answer!). However, thanks again to everyone for playing!

The Ware for August 2021 is shown below.

This months ware is probably a pretty easy guess. To make things a bit more interesting, the prize will go to the entry that has the most feasible (or the most entertaining!) theory as to the purpose for the tiny break-away, stand-alone PCB is on the left hand side, as indicated by the red arrow. The cropping just barely obscures the edge of the PCB, but basically there are three mouse bites on the edges that retain the sub-assembly PCB, so it could be sheared off and turned into a separate item. I always pay extra attention to blank spots like this PCBs, because they are riddles into some aspect of a product’s design or supply chain: someone put the effort in to design a thing — but then decided not to use it. This PCB has a lot of blank spots, but this is the only one that could be readily sheared off into a separate assembly.

This ware is also a guest ware, courtesy of “JeffA”. Thanks for the submission!