The ADXL345 is a MEMS accelerometer made by Analog Devices. It’s a popular device among hobbyists because of its low cost, easy availability and rich feature set. But apparently you should be careful about where you buy them: one of my readers ended up with a bunch of ADXL345s that had significant offsets, measurement axes that didn’t work at all, and an inoperative freefall detection mode. After spending lots of time trying to get them to work, he decided to send them to me instead and hopefully find out what was wrong with them.
Today we’ll have a look at another chip that has absolutely earned its place in the history of electronics. Designed in 1986, the MAX232 represented a small revolution in IC design and put Maxim, then still an upstart, in the spotlight for the first time. Dave Bingham was the driving force behind the MAX232’s design, in a team led by Dave Fullagar (of UA741 fame).
The MAX232 became popular because it was very easy to use and it solved a common problem: many computer systems running on a single +5V supply used RS-232 standard serial interfaces, which required +/- 15 V signal levels. The MAX232 was a transmitter/receiver chip with a built-in charge pump that generated the required signal voltages directly from a +5 V supply, removing the need for two additional supply rails.
Clever marketing also made the MAX232 a huge success for Maxim. Putting a big “MAX” in front of its part number made it immediately clear who made the chip, unlike TI’s “SN” or National’s “LM”. The fact that the ‘232 became the most popular among Maxim’s MAX2xx series (which ran at least from MAX220 to MAX249 back in the day) could be either because that number matched its RS-232 functionality, or simply because it had the most commonly used combination of channels (two transmitters and two receivers).
Maxim has since expanded its RS-232 transceiver range to include dozens of different models with varying numbers of channels, different supply voltage ranges (like the 3V MAX3232), power-down options (MAX242), and even integrated capacitors (MAX233) and isolation transformers (MAX252). But the classic MAX232 seems to remain its most popular model. I dug around in my parts bin and found this MAX232CPE from 1989. It’s housed in a 16-pin plastic DIP package.
If we open it up, we find the chip shown below. Maxim’s original datasheet includes an (extremely grainy) chip photo that matches what we see here. We’ve got the charge pump on the left, the receivers and transmitters on the right, and control circuits in the middle. This chip was fabricated in Maxim’s 3 micron CMOS process (called S3 or SG3).
A small 1986 copyright message in the middle confirms we’re looking at the very first generation of this chip.
I recently came across this funky looking thing. Left behind by a dishwasher repairman, it’s a pressure sensor from a dishwasher made by Marquardt, a German manufacturer of sensors, switches, pumps, lights, and various other components for industrial and automotive applications.
The Apple TV is a media streaming device first introduced in 2007. It’s essentially a miniature Mac, running a special version of iOS optimised for playing music and video on a TV, and operated using a remote control. Although I personally have no need for such a device, I couldn’t resist when I found one for just €2.50 in a second-hand shop. It wasn’t that useful anyway since it was missing its remote control, not to mention the fact it was an old and unsupported third-generation model, sold between 2012 and 2015. But it presented a very good opportunity to tear down some modern Apple hardware and get an up-and-close look at their custom silicon.
Although the 555 timer is one of the most iconic chips ever made, and the original version is still sold in huge numbers, it actually makes little sense nowadays to use the classic chip anymore. That’s because an improved version has been around for a long time: the CMOS 555 timer. Most manufacturers that produced the original bipolar 555 timer also make a CMOS version, with typically the letter “C” somewhere in the full product name. Today we’ll have a look at a couple of these CMOS timers and see how they differ from the bipolar model.
Most electronic systems that need an accurate clock, which is to say most microprocessor-based systems, use a quartz oscillator. You’ll typically see a metal package somewhere near your chip that contains a slice of quartz which resonates at a certain frequency thanks to the piezoelectric effect.
Quartz crystals are cheap and provide a very accurate clock frequency, but they take up quite a bit of space and are sensitive to shocks. To deal with those two problems, fully on-chip oscillator systems have been available since about 2010. These use micro-electro-mechanical systems (MEMS) technology, which involves the manufacture of tiny moving structures on a chip. Their price is typically higher than that of a quartz crystal however, and their frequency stability and phase noise performance are often a bit worse. Today we’ll have a look at a few different MEMS oscillator chips and see what they look like inside.
First up is the Si501 by Silicon Labs. It’s an 8 MHz oscillator built using what Silicon Labs call CMEMS technology, which means that they integrate the MEMS bit on the same chip as the rest of their circuits. The package looks rather anonymous, with just a cryptic part number and no manufacturer’s logo. Silicon Labs have since sold their MEMS oscillator business to Skyworks, so future versions of this chip might have a different marking.
I recently read a forum thread where someone showed how a set of LME49710s that he bought online didn’t function the way they should. Although the chips apparently contained an op amp, they were unable to amplify a 60 kHz square wave and output a triangle wave instead. This means that the op amps’ slew rate is too low: the LME49710 is specified to reach 20 V/us, but these chips only managed 0.5 V/us or so.
The thread’s author asked if anyone could help identify his chips, and I offered to examine them for him. A few days later I received the op amps in the post. They were clearly marked with the National Semiconductor logo and “49710” as a model number:
This is your chance to prove that you’re a real SMD soldering expert: to assemble it, you need to place 43 components onto a 10×10 mm2 PCB, pick and place 01005 size resistors (0.4×0.2 mm2), and solder DFN-1006-3 packages (1.0×0.6 mm2). A microscope, sharp tweezers and a fine-tipped soldering iron are essential tools to complete this project.
I bought this smoke and carbon monoxide detector several years ago, and it’s been doing its job just fine ever since: mostly being silent, screaming when I burn my toast, and beeping every two years or so when its batteries run out. Recently however it began beeping for no reason, and I couldn’t get it to stop other than by removing the batteries. So I bought a new one and decided to tear down the old one.
Back in 2016, TI introduced a line of what they called “nanopower op amps”. Where older op amps like the 741 use around 2 mA, and more modern ones might reduce that to perhaps 100 uA or so, TI’s ultra-low power devices consume just a few hundred nA. This enables the design of things like smoke alarms and temperature monitors that can work for a decade on a single battery charge.
This is the LPV801, a single channel op amp that uses just 450 nA. It’s not very fast: with just 8 kHz of unity-gain bandwidth it’s useless for audio, but ideal for slow-moving things like temperature sensors. A dual version (LPV802) is also available, as are single and dual versions with reduced offset voltage (the ‘811 and ‘812 respectively).
Inside we find this neat little design. Five bond pads are bonded to the five pins on the package; two additional ones on the top row are used for testing. In the top-right corner is an L-shaped alignment marker, which is used during laser trimming.
