The Ware for August 2026 is shown below.
Technically, this is an integrated circuit – just from a very different era than last month’s ware!
Archive for the ‘Hacking’ Category
Name that Ware, August 2026
Sunday, August 30th, 2026Winner, Name that Ware July 2026
Sunday, August 30th, 2026The structure in question for last month’s ware are a pair of NPN transistors that form the core of a bandgap voltage reference circuit.
As its name implies, a bandgap voltage reference generates a near-constant voltage for use inside chips. The voltage is intended to be fairly constant over process variations, temperature variations, and power supply fluctuations. The circuit itself is extremely clever, using circuit elements that have offsets that move in opposite directions with temperature to cancel each other out. Without this circuit, chips would have performance characteristics that strongly depend upon its operating temperature. Therefore, almost every system on chip has at least one of these.
If you’re ever looking at a chip micrograph, the circuit will jump out because it’s likely to be the only thing that has two arrays of perfectly square NPN transistors next to each other. I’ve highlighted the active transistors in pink. In this case, the transistor ratio is 1:64. Modern chip designs will always include a ring of extra “dummy” transistors around the active transistors. The reason for the dummies is that they protect the core from process variations. At these insanely small geometries, a well-known problem is that transistors on the edge of an array behave differently from those in the core. This is due to a myriad of problems, but a simple example of the problem for visualization purposes is to imagine the devices going through an etching process: the concentration of the etchant will depend upon the density of the pattern being etched. The devices on the edge of an array will etch at a different rate than those in the center, and thus they will have a slightly different net performance characteristic. As a result, for very sensitive circuits like a bandgap voltage reference, it’s standard practice to “throw away” the edge devices as their characteristics are harder to control.
The orange region above the pair of NPN transistors is a pair of PMOS devices that are current mirrors that drive the NPN transistors. The blue square are a pair of resistors, and the green square is a differential amplifier that forms the active circuit core of the bandgap circuit. These circuits are so commonplace that it’s fairly easy to find a representative schematic just Googling for one:
The above was one of the first hits I got googling for “bandgap circuit“. The actual topology of this schematic is not exactly the same, but very similar to the circuit in the layout above.
When looking at a large, unknown die shot, I will often orient myself by first looking for a bandgap circuit, and then looking for an SRAM. Between the two of these, I get a size standard to which I can use to compare analog and digital circuits against, and work my way from there towards other functional elements on a chip.
While nobody guessed this circuit, asdf’s response gave me a chuckle, so I’ll give asdf the prize. Drop me an email to claim your prize!
Name that Ware, July 2026
Thursday, July 30th, 2026The Ware for July 2026 is shown below.
I’ve got silicon on the brain, so I’m going to give a die shot another go at name that ware. Hopefully this one is a bit easier than the last one. This excerpt is a design pattern I look for and find on almost every SoC. It’s technically showing a bit more than just the part I’m focused on, so as a hint I’m primarily interested in identifying the structure towards the lower left corner of this image.
If you’ve seen one of these before, you’ll know what it is – it’s pretty unmistakable, and a design pattern that’s highly conserved across processes, nodes and foundries. Bonus points to anyone that can explain some of the reasons why things are organized the way they are!
Winner, Name that Ware June 2026
Thursday, July 30th, 2026The Ware for June 2026 is an SGI RM6-64 raster manager, from a KONA/InfiniteReality graphics system. Congrats to Christian for naming it! email me for your prize.
This gem of a board was gifted to me by someone at SGI back when I worked there in…I want to say 1998? I even still have the name plate from my cubicle, from back before the ‘bunnie’ nickname really stuck…
Back then, if you wanted to do graphics, SGI was the only game in town, but I joined the party late. Within six months of joining, most of the division I was in had jumped ship for these then little-known upstarts of nVidia and ATI. Lacking any connections and the office a ghost town (my office building was eventually acquired by another startup that’s become a household name, you might have heard of them: Google. Still really weird visiting friends that work on that campus…), I went back to the familiarity of MIT and got my PhD, and that was the last time I held a job in Silicon Valley. Ironically, my main project today is doing silicon, just not in the Valley. What’s that old saying…you can take the boy out of Silicon Valley, but you can’t take the silicon out of the boy?
It was pretty impressive what they could do at the board level at SGI. That all got eaten up by Moore’s Law and integrated graphics…and now that Moore’s Law has run out of steam, we’re returning with gusto to the huge rack servers with lots and lots of interconnect!
Name that Ware, June 2026
Tuesday, June 30th, 2026The Ware for June 2026 is shown below:
This board is from my personal collection of “favorite boards” that I’ve collected over the years. A lot of old memories associated this one – both figuratively and literally.
I’ve received some feedback that gmail in particular is marking my email updates as spam. Not sure I have a good way to fix this yet, as the blog is also under heavy attack from automated AI spammers at the same time (to the tune of several thousand requests a day).
You’ll notice that the blog has no ads, so I also have no budget for moderation, paid mailers, or fancy anti-spam tools. I’ll noodle on a durable solution to this problem in my copious spare time, but for now, if you’re not receiving the email updates, please try to convince Google this is not spam by marking it as ham in your inbox. Yes, have also received feedback that this is ineffective and that “the way” is to pay a service that specializes in sending spam *ahem* marketing emails, but see my previous comment about this being a lean, ad-free blog. But, maybe if enough readers mark the emails as not spam, “The Algorithm” might be convinced people actually wanted the updates they signed up for?