With innumerable microcontroller boards on the market it’s sure that there will be one for every conceivable application or user. Among them are some seriously tiny ones, but this wasn’t enough for [Alun Morris]. Wanting to see how small he could make an ATtiny board without a custom PCB, he took a SOIC-8 version of the popular minimalist processor and mated it to a 6mm by 8mm piece of 0.05″ prototyping board to create a device that is dwarfed by its connectors.
It’s an extremely simple circuit and hardly something that hasn’t been done before, but the value here is in the tricky soldering to make it rather than its novelty. The ATtiny402 and three passive SMD components are fitted on the smallest possible sliver of prototyping board to contain them, and the female headers and set of programming pins contribute far more to the volume of the device than the board itself. He also tried a side-on design with two smaller slivers of board before settling on the more conventional layout. The demonstration of the system in action seen in the video below the break is a magnetic flux detector, dwarfed by the 40-pin DIP Z80 it is sitting on.
A lot of boards claim to be tiny, but few are this small. This ESP32 is a more usual contender.
12 thoughts on “How Tiny Can A Microcontroller Dev Board Be!”
Wow, tiny indeed, nice post, thanks :-)
Very cool. Although I have used DIPs that were pretty easy to connect up for running or programming without any board at all, just plug in a connector for power and a few I/O.
Even the Parallax Propeller wasn’t too bad in its day, hook up about 8 to 10 pins and you could program it and run something simple. Which was rather handy when were were given bare chips as samples without any nice dev board.
It was a sad day for me when NXP dropped the LPC810 (ARM in an 8-pin DIP). But once you want to put stuff in a package smaller than an Altoids tin, the new fangled surface mount stuff is the way to go.
Thanks for reminder jonmayo,
I looked at this a couple of years back iirc, great idea I thought to offer a small/light package with serious computing capacity ie 32 bit CPU eg for multiple local use for micro manipulators where tight feedback built in to actuators etc as one potential. There are a couple of other ideas somewhat “out there” for distributed DSP if some high speed serial bus between them…
Pity that chip isn’t in production, though I guess it wouldn’t be difficult if a serious mass application arose…
Which begs the question jonmayo, what would a comparative equivalent that’s in production be these days, maybe with fast 16 bit CPU in a 10pin max package or a really cheapie 8 pin one ?
Something to muse over, I like the idea of big arrays of these little blighters doing things not yet explored ;-)
PIC32 (MIPS 32-bit) is still available in PDIP-28 packages. But also in very compact QFN-20 package which could fit on a DIP-like carrier board. (but QFN is a paint to solder)
For 16-bit there is the MSP430 and PIC24 series. But to be honest I don’t find 16-bit to be all that compelling over some of the modern 8-bit architectures (AVR). For me the choice between bits boils down to tools. A good C compiler and a reduced need for assembler and cycle counting is the big draw for 32-bit MCUs.
For a small ARM, NXP makes a Cortex-M0 in QFN-16 package. That’s hard but not impossible to attach to a board yourself. But I would almost argue that if you are going to use a board, then do something more complicated to justify it. For example pay to put a BGA part down and have lots of I/Os. or Maybe get a tiny FPGA and hooked it to a tiny MCU for a really potent data collection or soft radio. iCE40 comes in a 30-ball BGA package, super tiny and you could fit it on a DIP12 PCB with a small voltage regulator and small MCU like a Kinetis in 36BGA. (I sure seem to dig up the NXP product lines, honestly I don’t work for them)
Cool, will consider wider terms of reference. Have been a user of Atmel and silage/cygnal 8051 series for decades, been indoctrinated it’s like a language option. For a small micro seems ideal with good mix of efficiency vs instruction types, would be great to find a comparable paradigm for a 16 or 32 bit upgrade. I suppose the latest Risc series open source best fit for the foreseeable future re embedded variants.
Try for an even smaller board: ATTiny10. The largest chip is a tiny SOT23-6 package, has 3 GPIO (4 if you disable reset) and really tiny 1k flash and 32B SRAM, no EPROM
“With innumerable microcontroller boards on the market it’s sure that there will be one for every conceivable application or user.”
I’m sure that’s a job in itself, finding that one needle in the haystack, hoping it’s in stock.
Modern day development planning:
1. Coding and testing: 1 day
2. Search for microcontroller that can do the job with only 1 day of coding and testing: 5 days
Vs my strategy:
1. Grab micro out of closet
2a. If you’re lucky, coding and testing 1 day.
2b. If unlucky, spend 6 days trying to get the chip to do something that it shouldn’t instead of just ponying up and buying a bigger one…
I’ve built a large number of very similar tiny assemblies, but in all but the simplest circuits, there seems always seems to be one wire that needs changing and so the whole assembly needs to be rebuilt. I avoid doing these now unless it’s absolutely necessary due to space requirements.
Shameless plug, but same. This comes in a 10mm*10mm board and uses the smaller WLCSP package. https://hackaday.io/project/177006-attiny20-bga-breakout-boards Haven’t programmed these yet though.
Was wondering what kind of smaller connector can be used for interfacing when needed… like sections of an LVDS or CSI like inline connector or maybe there is something smaller I’m not aware of? Then make an adapter to save on space.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)