Just another electronics blog

I have been using an Aneng AN8008 multi-meter for a while now and I quite like it. It’s small, accurate enough and has some interesting measurement options for embedded electronics. Moreover, it’s very affordable. Looking at what multi-meters Aneng also sells, I spotted a multi-meter with Bluetooth for a low price as well. Bluetooth in a multi-meter, that’s a new one for me. But thinking about it, for logging data, why not. It’ll be very galvanicly isolated at least :)

After receiving the meter and trying it out for a bit, the app left me a little disappointed. There is just a phone app that can show the values and log them, but sadly storing the logged data to a file is not possible. This is be useful if you want to analyze the data at a later time or put graphs in a report or such. So time to look at what makes the multi-meter tick and see if I can make my own data logger with it.

As I am still waiting on parts for the lab power supply, thanks 2021, time to look at something completely different. I picked up this giant box a few months ago. It contains a good old Motorola 68000 CPU board and a so called Syscon board all in a VMEbus backplane with plenty of empty slots. I could not find much documentation, apart from some Ebay seller selling the CPU board for a lot of money, as it was apparently used in an ASML system. So let’s dive into VMEbus, 68K assembly and some good old reverse engineering.

First things first, what is all on that CPU board.

A while back I was looking at service manuals for a few lab power supplies, and I remembered the elegant approach of the HP E361xA series of power supplies. I wanted to see if it is possible to add digital controls, perhaps even making it able to control it via a computer. Ideally, I also wanted to see the current limit when adjusting it, a feature too many power supplies lack.

So let’s do just that, recreate the lab power supply with some modern goodies added to it.

Before making one, let’s have a look at the schematic from the original and see what makes it tick.

Recently, the Raspberry Pi foundation launched the Raspberry Pi 400, a keyboard with a build in Raspberry Pi 4. It features a CPU very similar to the Raspberry Pi 4, 4GB RAM and it’s priced 15$ higher then the comparable Raspberry Pi.

As it’s all nicely stuck in a keyboard, it looks like it’s meant to be used as a desktop computer, more then a Raspberry Pi 4 at least. I ordered one and decided to use it as my desktop computer for a week to see how it fares.

As I got the personal computer kit, I also got the mouse, power supply and such included. I did use a bigger 64GB microSD card as 16GB is quite small for desktop use in my opinion.

Like many other electronics hobbyists I am enjoying the video’s Ben Eater is making on getting a 6502 CPU up and running on a breadboard. His video’s are very informative, going in depth how the CPU works and slowly buiding up to hello world. If you haven’t watched them, give them a try!

Another great 6502 (and other 8 bit CPU’s) project is Grant’s simple 6502 computer. With a minimum part count he creates 6502 computer that runs Microsoft Basic.

The problem for me is that I have had quite some issues building big things on breadboards, broken wires and other hard to debug issues. Instead I decided to make a circuit board to solve this problem. I also wanted the board to work with Ben Eater’s video’s as well as run Basic. The circuit board turned out like this:

So let’s get into some detail how it works and what is different.

I like CPU architectures, especially weird, interesting and unusual ones. For example, the Intel iAPX 432 is still something I would love to play around with. Recently, I realized that a working CPU can be made with just a simple Move instruction. For this to work, everything needs to be memory mapped. The ALU, program counter, everything.

Of course, this idea is nothing new and this idea is called the Transport Triggered Architecture. I decided to have a look into this, how it works and make a simple TTA CPU.

How does a TTA CPU work

Before I can make a CPU, let’s look into what is so different in a TTA CPU. In most CPU’s, calculations are done using registers and some arithmetic logic unit. For example, to add 2 numbers, the assembly code could be:

LOAD VARIABLE1, REGISTER0
LOAD VARIABLE2, REGISTER1
ADD REGISTER0, REGISTER1, REGISTER2
STORE REGISTER2, VARIABLE3

In a TTA CPU, there is no ALU or registers in the CPU itself. Instead, they exist somewhere in memory. In order to add 2 numbers, they are moved from memory or registers, to the ALU. The ALU result is then moved back to memory/registers. Or in code:

MOVE VARIABLE1,  ALU_A
MOVE VARIABLE2,  ALU_B
MOVE ALU_ADD_RESULT, VARIABLE3

In the simplest form, the CPU in a TTA CPU only needs to move data from one memory address to another. All calculations are done as a result of data being moved around. If you think, how would you jump to a different section of code, it’s easy. If the program counter is also in memory, a jump is as simple as moving a new address to the program counter.

A few TTA computers have been made, and even commercially sold. But in general, it’s a niche architecture that never has gotten popular. So let’s build one :D

In the last post I promised to talk about the custom hardware to drive the 64*64 LED matrix. I wanted custom hardware because my test setup looked a bit like this:

This is not ideal of course, and I want to use my dev board for other things as well. Hence the need for a custom PCB. In this blog I’ll be going over the design of the custom FPGA board to replace the test setup.

Browsing Ali-express is dangerous business. Before you know it you end up ordering strange things like a 64×64 pixel RGB LED matrix. These matrices (called HUB75 or HUB75E) are meant for use for the big outside LED displays.

There are several projects already involving these displays, but I wanted to do more FPGA stuff and this seems like a great excuse. A fast microcontroller can drive these displays. However, an FPGA is much better suited for this. The reason is that these displays do not have on board memory but need to be constantly refreshed to display an image. But when you get them up and running with an FPGA, the results are mighty fun

Recently I got an old keyboard from the 1980s, it has a 20 pin ribbon cable connector, weights almost 3kg’s and has no information on Google. This should be fun.

The keyboard in question is a Computervision branded one, though the PCB says it’s an Amkey MPNK-68. Looking at the insides, the date codes on the circuit board are from 1984 and 1985, meaning this keyboard is probably from the end of the 1980s. Taking a look at the keycaps and all the special functions, it was most likely used for a terminal.

Now to reverse engineer it. A few tools are required to do so:

1. A multimeter.
2. A logic analyzer, like a Saleae Logic.
3. Some screwdrivers to take the keyboard apart.
4. A power suply, most likely 5V

Very simply put, the steps are as follows:

1. Open the keyboard
2. See what the pinout of the connector is.
3. Power the keyboard.
4. Hook up a logic analyzer to all signal pins.
5. Figure out how it works, is it a serial protocol, parallel, something else.
6. Build a converter using a microcontroller board like an Arduino.