Thursday, November 11, 2021

Bugs in Complex Software

In the picture: history's first bug.

Complex software will be guaranteed to have bugs in it. And most of the bugs are caused by state changes.

The human mind has limited capacity of keeping track of state (data, variables.) Hence, there is a discrepancy between what the state looks like in the programmer's mind, and what it actually looks like.

How does state change?

Where in the code does the state change?

When does the state change?

In what order does the state change?

And even... why doest the state change?

The mental picture of those, formed by the programmer, is very poor.

So.. bugs!

There is of course a paradigm where state-change plays a zero roll: functional programming. But despite decades of trying, we have not been able to practically use that in complex software. Mainly because the real world deals with state. Try making a video game that carries no state!

John Carmack once did an impromptu talk on functional programming and video games, and made in interesting observation that you could write an AI/sim that takes the entire world as function input, and produces the entire world as its output.

About John's example of a one-person hallway... I think the solution is simple. Input for hall-entering needs more than the current entity. It needs a list of all other entities that want to enter. Then use a heuristic to choose which one can actually enter. Done.

So yeah... complex state changes. As long as we have not harnessed the complexity of state, I find it pretty futile to crusade against C's unsafe memory referencing, e.g. A bug caused by following a dangling pointer is so shallow, compared to a bug caused by asynchronous state changes racing to completion. Throw in multiple threads, multi-player, and it will be too hard to reason about.

Sunday, November 7, 2021


I've been learning a lot about transistors, lately. This is because I found myself having to switch a load that would be too large for an IO pin. Here I summerize my newly gathered knowledge on them.

A BJT transistor is switched ON/OFF using current at the base. This current is then multiplied at Collector/Emitter.

To tune the current flowing through the base, we need a bias resistor. This makes PCB design bulkier, but you can get pre-biased transistors. I find I need to get them with a low Ohm bias, else my transistor will not switch fully on.

BJT comes in two variants, NPN and PNP. The schematic symbol differs in which way the arrow points. Mnemonic: NPN has the arrow "Not Pointing iN."

PNP transistors can be used to switch a load at its high-side: You cut the load's connection to V+ ON/OFF.

NPN transistors can be used to switch a load at its low-side: You cut the load's connection to GND ON/OFF.

MOSFET transistors are often a better choice, unlike of BJT's current based switching, the switch based on voltage at the gate.

MOSFET comes in two variants, P-CHANNEL and N-CHANNEL.

P-CHANNEL transistors can be used to switch a load at its high-side: You cut the load's connection to V+ ON/OFF. A 0V at the gate will switch on the transistor. Source is connected to V+ and Drain is connected to the load.

N-CHANNEL transistors can be used to switch a load at its low-side: You cut the load's connection to GND ON/OFF. A V+ at the gate will switch on the transistor. Source is connected to GROUND and Drain is connected to the load.

To simplify a PCB design, you can get two transistors that share its housing in an "Array."

Wednesday, October 27, 2021

Bi-Colour LED bars.

I am evaluating bi-colour LEDs that are RED/GRN, and can be mixed to ORANGE/YELLOW. The two candidates are SunLED GMDKVGX10D and KingBright DC10EGWA.

I've found that the SunLED has a bright GREEN with an even brighter RED. You need to drive the green channel with at least double the current of the red, to match them. I find that 1kΩ and 2kΩ work for my application.

The KingBright, on the other hand, doesn't show much difference between the RED and GRN brightness. But both are far less bright than the SunLED. Even though the KingBright has the term "High Efficiency" in the data sheet? When I limit both RED and GRN with 470Ω I find it is still quite a bit dimmer than the SunLED at 1kΩ/2kΩ.

What both models do the same: they have the GRN/RED LEDs share their cathode, not their anode. A common-anode design would have been so much more convenient. You would have been able to drive it with a constant current LED driver that is a current sink. All constant current LED drivers that I could find are current sinks, so none of them can be used with these LEDs. So not Texas Instruments TLC591x, nor TLC592x and neither a Maxim6966 would work. I consider this a strange design decision.

Another feature I miss, is an 8-segment version of these bar-graphs. It's just more convenient to work with 8-bit quantities. Or in the case of bi-colour bar-graphs: 16-bit would be preferable over 10-segment 20-bit units. Monochromatic, and fixed-colour bar-graphs come in more varied sizes, like 8-segment, 10-segment and even 12-segments sometimes.

Thursday, October 21, 2021


I am using the TLC5928 constant current sink driver to light up my LED bar graphs.

The neat thing about this part, is that you can set a current (for all 16 outputs) using a reference resistor. The higher the resistor, the less current the chip will output. So far so good.

If you have a LED bar graph that has both GRN and RED LEDs in it, you have a problem, though. The RED is a lot brighter, and does not need nearly as much current as the GRN one. I see typically a factor 4x more current for green. However, the reference resistor sets the current for ALL channels, whether they have a red or a green LED on it.

So, how to proceed? Well, I decided to do add pulse-width-modulation for the red LEDs. I will still drive all channels with the same constant current, but the red LEDs will have their anode voltage (3.3V) interrupted at a few hundred hertz, with a specify duty-cycle. Whereas the green leds will keep seeing their uninterrupted voltage.

A GPIO pin can never deliver the current required to switch a whole bunch of red LEDs. So we introduce a little switch, called a Bipolar Junction Transistor of the PNP form.

When adding a discrete transistor to do perform this PWM job, I also need to add resistors, which makes the design more cluttered. Fortunately, you can buy pre-biased transistors that already house those resistors along with the transistor. I have evaluated 3 such pre-biased transistors for this job.

DTB114EK (R1 and R2 are 10kΩ)

DTB123TK (R1 is 2.2kΩ)

DTB113ZC (R1 is 1kΩ and R2 is 10kΩ)

Note that in my case, I have 3.3V at the emitter (pin1) and not GND.

I've found that a high R1 resistor at the base of the transistor causes the EMITTER->COLLECTOR current to be too limited when the transistor is switched on. Even when I have a continuous '0' at the base, to switch it on, the current does not get high enough to drive many LEDs. Hence, between these three, the DTB113ZC with a 1kΩ resistor at the base does the best job for my application. I find that if I have a bunch of LEDs drawing 109mA in total, and switch that current with the transistor, 90mA makes it through with the DTB113ZC, 79mA makes it through with the DTB123TK and 47mA makes it though with the DTB114EK.

Tuesday, October 19, 2021

The state of Electronic Circuit Simulation Software under Ubuntu.

Today, I was in need of electronic circuit simulation. Let's see what is available for Ubuntu.

First I tried ngspice, but that seems to be text based.

$ ngspice 
** ngspice-34 : Circuit level simulation program
** The U. C. Berkeley CAD Group
** Copyright 1985-1994, Regents of the University of California.
** Copyright 2001-2020, The ngspice team.
** Please get your ngspice manual from
** Please file your bug-reports at
** Creation Date: Sat Mar  6 03:53:35 UTC 2021
ngspice 1 ->

Ok, install the GUI for it then...

$ sudo apt install gspiceui
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
Package gspiceui is not available, but is referred to by another package.
This may mean that the package is missing, has been obsoleted, or
is only available from another source

Next up: Oregano. I managed to compose a circuit, but when I tried to simulate:

$ oregano 
Segmentation fault

Next up: xschem. Again, I managed to draw a schematic, which is done in an awkward way. When I wanted to simulate, I could find no such option in the menus. It seems it cannot?

Ok, let's move to QUCS. No hit on a package search in Ubuntu, which does not bode well. Ok, let's grab a deb from the archive.

# dpkg -i qucs_0.0.20.201022-2_amd64.deb 
Selecting previously unselected package qucs.
(Reading database ... 281811 files and directories currently installed.)
Preparing to unpack qucs_0.0.20.201022-2_amd64.deb ...
Unpacking qucs ( ...
dpkg: dependency problems prevent configuration of qucs:
 qucs depends on libqt4-qt3support (>= 4:4.5.3); however:
  Package libqt4-qt3support is not installed.
 qucs depends on libqt4-script (>= 4:4.5.3); however:
  Package libqt4-script is not installed.
 qucs depends on libqt4-svg (>= 4:4.5.3); however:
  Package libqt4-svg is not installed.
 qucs depends on libqtcore4 (>= 4:4.7.0~beta1); however:
  Package libqtcore4 is not installed.
 qucs depends on libqtgui4 (>= 4:4.8.0); however:
  Package libqtgui4 is not installed.

dpkg: error processing package qucs (--install):
 dependency problems - leaving unconfigured
Processing triggers for libc-bin (2.33-0ubuntu5) ...
Processing triggers for desktop-file-utils (0.26-1ubuntu1) ...
Processing triggers for mailcap (3.68ubuntu1) ...
Processing triggers for gnome-menus (3.36.0-1ubuntu1) ...
Processing triggers for hicolor-icon-theme (0.17-2) ...
Processing triggers for man-db (2.9.4-2) ...
Processing triggers for menu (2.1.47ubuntu4) ...
Errors were encountered while processing:
root@Workstation:/home/bram/Downloads# apt install libqt4-script libqt4-svg libqtcore4 libqtgui4
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
Package libqtcore4 is not available, but is referred to by another package.
This may mean that the package is missing, has been obsoleted, or
is only available from another source

Maybe GEDA?

# apt install geda
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
Package geda is not available, but is referred to by another package.
This may mean that the package is missing, has been obsoleted, or
is only available from another source

E: Package 'geda' has no installation candidate

Oh dear... I am about to give up now. I can see now, why CircuitLab manages to charge $399,- per year.

Friday, October 8, 2021

The SMD Hurdle

I found that programming microcontrollers and designing electronic circuits is so much more fun than modern software development for PC. I have really taken to the hobby. So much so, that I'm beginning to think a pivot is in order.

When I started mobile game development in feb 2009, I was ahead of the curve. So far ahead of the curve, that app-store money was easy to come by. No game engines, no unity available. Everything had to be custom roll-your-own code. The number of people that knew OpenGL and ObjectiveC was near zero. It was good to be at the start-point of the democratization of software development. Fast forward to today, and the market is so crowded, it is impossible to stand out.

Just as the iPhone opened up a cottage industry of app developers, I think there is a new wave that you can ride, that is not very crowded. Chinese PCB factories have made electronics design accessible to all. For $40 dollars or so, you can have your PCB printed manufactured in a small batch. And if you add a few hundred, you can even get the procurement, and placement of the components on your board, for full service fabrication of your design. Not bad!

Hence, the last few weeks, I have been designing electronic circuits using EasyEDA. It's an easy to use, free PCB design tool. I like it a lot, except for the slow cloud operations that it relies on for many of its operations. But kudos to them for providing a Linux client for it.

So far, I have ordered PCB service from both JLCPCB and PCBWay. Those are the two biggest players in the PCB market, both located in China. From JLCPCB I ordered my first design, which was Through-Hole components only, so I could do the assembly myself. From PCBWay I ordered full service PCB with assembly and component procurement included. That was quite a bit more expensive. And that order has not been completed yet, as it still waits for component delivery. The JLCPCB order did arrive already, and is shown below.

I'll go into the actual product that I am designing, at a later time. But currently, the design is getting more and more complex. The second version had SMD components, so I had pcbway do that for me.

Revision 3 had even more and even smaller components, in a QFN-24 (0.5mm spacing) package even! And to make matters worse: at both sides of the PCB. This is because my design is densily integrated and has space constraints, so I needed both sides of the board. As I was preparing my order for assembly at JLCPCB, the site told me that two-sided SMD assembly was not supported. I had to pick one side, which they would do for me. Ugh.

So, I need to reexamine my options. Would SMD mounting be something I could attempt myself? I've always backed away from it, as being too much of a hurdle.

But some makers with SMD DIY videos have convinced me that it must be possible. So I decided to have a go at it, myself! Surface mounting tiny little components on a PCB.

For this, I will be ordering a set of new tools:

  • A re-work station, which is just a temperature controlled heat-gun.
  • Solder paste (a mix of metal, solder, flux.)
  • Scraper.
  • Tweezers.
  • Big-ass magnifying glass.

In addition of all this, I must make sure that I order my PCBs with a stencil plate. This stencil will help me apply the paste to where it is needed.

In a good fortune of events, the manufacturer of the leading soldering paste is from Canada. Nice! No import duties, and fast shipping.

Anyways, wish me luck!

Oh no! That is too small!

Tuesday, September 21, 2021

LED Bars

Two LED bar graphs. You would think they would be much alike, but no.

On the left: LTA-1000Y (2017 Thailand.) On the right: FJB10Y.

The polarity is reversed. The LTA has anode on the side of the first pin. The FBJ has the cathode on the side of the first pin.

The obvious difference: face colour grey vs face colour black.

Light colour is similar, but brightness is much different. I need to use 470 ohm resistor with the LTA, and a 2K ohm resister with the FJB.

I did some extra research on another convention, for multi-coloured bars. Is RED on top or on the bottom of the bar?:

  • AVAGO HDSP-4832 has R/R/R/Y/Y/Y/Y/G/G/G with pin 1 as RED anode.
  • KingBright DC7G3HWA has G/G/G/G/G/G/G/R/R/R with pin 1 as GRN anode.
  • Fairchild MV5A164 (obsolete) has R/R/R/Y/Y/Y/Y/G/G/G with pin 1 as RED anode.
  • Fairchild MV5C164 (obsolete) has G/G/G/G/G/G/Y/Y/Y/R with pin 1 as GRN anode.
  • KYX B10BBGYR has B/G/G/G/G/Y/Y/Y/R/R with pin 1 as BLU anode.
  • MT1025CTF has G/G/G/G/G/Y/Y/Y/R/R with pin 1 as GRN anode. Pin 1 (unmarked) at side that has print.
  • SunLED XGMDKVGX10D has 10+10 bi-colour LEDs to interpolate between GRN to YLW to RED. To use them, 2k Ohm on R and 470 Ohm on G works well. Unfortunately, it is a common-cathode setup, which makes it useless for standard LED drivers.
  • F2510GHR has G/G/G/G/Y/Y/Y/Y/R/R with pin 1 as GRN cathode. It is easy to connect incorrectly. UNLIKE ALL OTHERS, THIS ONE HAS PRODUCT NUMBER PRINTED AT CATHODE SIDE (BUT STILL AT PIN1 SIDE.) Gorgeous colours though, with the yellow being more like amber, A nice looking green that is not very bright, though.
  • F2510HYG has R/R/R/R/Y/Y/Y/G/G/G with pin 1 as RED anode.

Some 8 segment led bars:

  • F2010RG is R/R/G/G/G/G/G/G with pin1 as red anode.
  • KYX-B08GOR is G/G/G/G/G/G/R/R with pin1 as grn anode.