Composite video with Atmega328p

I started to get some composite video generated with a arduino for my 6502 project.

UPDATE: 20221021

It is based on Grant Searle’s design, and yesterday I had some signals on my scope which looked like a screen with a character. But my monitor would not recognize a usable signal.

Today I tried a second version and another set of chips and crystals.

It looks like a signal, but I can’t see a clock pulse from the crystal?! So .. how?

Maybe I used a bad power supply. And killed something?

UPDATE: 20221021

After switching to another power supply, and checking the atmega328p fuses again (also wrong) .. at least SOME success!

Still a little sync problem, but i’ve got a blinking cursor!
Some minipro info
#Erase
minipro -p ATMEGA328P@DIP28 -E

#Flash hex code
minipro -p ATMEGA328P@DIP28 -w SBCVideo.hex

#Flash fuses
minipro -p ATMEGA328P@DIP28 -e -c config -w fuses

#Used fuses file
lfuse = 0xf7
hfuse = 0xd9
efuse = 0xfc
lock = 0xff
lfuse = 0xff
hfuse = 0xff
efuse = 0xff
user_id0 = 0xff
user_id1 = 0xff
user_id2 = 0xff
user_id3 = 0xff
user_id4 = 0x37
user_id5 = 0x37
user_id6 = 0x39
user_id7 = 0x36


#Dump all from atmega328p
minipro -p ATMEGA328P@DIP28 -r dump -f ihex


Some info about the fuses:
https://www.allaboutcircuits.com/projects/atmega328p-fuse-bits-and-an-external-crystal-oscillator

Shelly Flood Sensor

I’ve had this Shelly sensor for a long time. But never posted anything about this.
Last weekend we had a -situation- in our kitchen, so what better time to test this device again!

This little disc shaped device has three metal points on its bottom side, those are the flood (water) sensors.
It stay’s in sleep mode when all’s good.
It does several things when it detects water.

  • Emits a alarm signal
  • Wakes-up wifi
    • Sends a MQTT message (when not connected to the cloud like i have)
      MQTT is a alarm message AND it wil send the temperature of the device!
  • After a while (when dry) goes back to sleep

There are connection point on the print you can use .. happy hacking!

My node-red configuration

Above is the part where the mqtt messages gets processed by Node-Red
Sending it to PushOver and my little MqttLcdNotifier

Above is the MqttLcdNotifer .. there are several parts to this

  • Top line is from shelly flood and other notifications
  • Text input puts text from the NR GUI on my TV and the LCDDisplay
    • same parts are being used by my 3D printer when the print tool is getting TO hot, or printing is finished
  • Trigger at work WAS a notification for work .. nonfunc
  • mqttlcd-button is the mqtt message send from the display (the one that i was pushing) to stop the beeping and clears the display
  • Bash notify, is as previously posted a flow which i can control from my linux machines notify “compiling complete” for example.
    This is also being broadcast from my livingroom using speakers.
    (See separate post about this)

Wellll, put this in place 2 years ago, never looked at it again .. still works

Volume is low, due to alarm sounds 🙂

7 segment displays/keyboard decoders in SDK’s

There are a lot of old develop boards for all kinds for cpu’s.

These where build to learn machine code programming. Mostly made in the 80’s, and based on populair cpu’s at that time.

I own a some of these SDK’s (System Design Kits)

8085 – SDK85 i bought recently 8085 CPU
Microprofessor-1 (MPF-1) Z80 CPU
And my own 680x based computer

Most of these use a keyboard scanner which is also connected to 7 segment displays.

The way they work is practically the same. There is a VIA or PIA. Versitile interface adaptor, or Peripheral interface adaptor. These have two times 8 bits to control devices.
When using 4 bits and convert these to 16 lines by using a 75ls145 for example. If you put a counter on those 4 bits, you sequently activate 1 of 16 lines. These lines you can use to scan a keyboard matrix OR display a character on a 7 segment display. These display’s won’t hold the data (and show the character) when not activated. The trick is to update de display fast enough so you don’t see the flickering on/off.

Activate a line and read a byte with the VIA = Reading keyboard row
Activate a line and write a byte with the VIA = Display on a segment

These VIA/PIA’s where made with specific timings to match the CPU.
6522/6820/8255

Below you see some different implementations of these keyboard/display combo’s

Thaler 6502 kit
Microprofessor MPF-1 kit (ignore red circle)
SDK85 kit
Eltec 6800
My version using darlington arrays (ULN2003)

When looking at the 8085 version you see transistors being a ULN2003 is a chip with those transistors/amplification enclosed.
It doesn´t draw much current from the bus, and diodes protect the way the current flows.

Quick n easy media viewer in a browser

Februari 2021 i made a website to view images and movies in a browser to do some quick sorting. (borrowed some code from a codepen page i recall correctly)
At the time i didn´t have a good way to view webp webm media.
I wanted to view multiple files at the same time, and make it short and simple.

BTW no webserver needed, just open the file from a directory!
jpg’s png’s webm webp mp4 svg and animated gifs work. (maybe more, didn’t test more, whatever your browser supports)

With recent updates of the chrome browser the video attributes to mute is broken, i so made a workaround.
Also everything is in one file now.
Except for one issue .. i couldn´t create one file for images AND videos.

There is a piece of javascript i could not fix … yet
I have to do execute a document.createElement which is different for images and videos.
Also the attributes of video are mute,autoplay,loop,playinline

Examples:

index.html

<a href="video.html">Video version</a>
<div id="drop-area">
  <form class="my-form">
    <label class="button" for="fileElem">Drop some images</label>
  </form>
  <div id="gallery" /></div>
</div>
<style>
body {
  font-family: sans-serif;
}
#drop-area {
  border: 2px solid #ccc;
}
.my-form {
  margin-bottom: 10px;
}
#gallery {
  margin-top: 10px;
}
#gallery img {
  width: 300px;
}
}
#fileElem {
  display: none;
}
</style>
<script>
let dropArea = document.getElementById("drop-area")
;['dragenter', 'dragover', 'dragleave', 'drop'].forEach(eventName => {
  dropArea.addEventListener(eventName, preventDefaults, false)   
  document.body.addEventListener(eventName, preventDefaults, false)
})
;['dragenter', 'dragover'].forEach(eventName => {
  dropArea.addEventListener(eventName, highlight, false)
})
;['dragleave', 'drop'].forEach(eventName => {
  dropArea.addEventListener(eventName, unhighlight, false)
})
dropArea.addEventListener('drop', handleDrop, false)
function preventDefaults (e) {
  e.preventDefault()
  e.stopPropagation()
}
function highlight(e) {
  dropArea.classList.add('highlight')
}
function unhighlight(e) {
  dropArea.classList.remove('active')
}
function handleDrop(e) {
  var dt = e.dataTransfer
  var files = dt.files
  handleFiles(files)
}
function handleFiles(files) {
  files = [...files]
  files.forEach(previewFile)
}
function previewFile(file) {
  let reader = new FileReader()
  reader.readAsDataURL(file)
  reader.onloadend = function() {
    let img = document.createElement('img')
    img.src = reader.result
    document.getElementById('gallery').appendChild(img)
  }
}
</script>

video.html

<a href="index.php">Images version</a>
<div id="drop-area">
  <form class="my-form">
    <label class="button" for="fileElem">Drop some videos</label>
  </form>
  <div id="gallery" /></div>
</div>
<style>
body {
  font-family: sans-serif;
}
#drop-area {
  border: 2px solid #ccc;
}
.my-form {
  margin-bottom: 10px;
}
#gallery {
  margin-top: 10px;
}
#gallery video {
  width: 300px;
}
}
#fileElem {
  display: none;
}
</style>
<script>
let dropArea = document.getElementById("drop-area")
;['dragenter', 'dragover', 'dragleave', 'drop'].forEach(eventName => {
  dropArea.addEventListener(eventName, preventDefaults, false)   
  document.body.addEventListener(eventName, preventDefaults, false)
})
;['dragenter', 'dragover'].forEach(eventName => {
  dropArea.addEventListener(eventName, highlight, false)
})
;['dragleave', 'drop'].forEach(eventName => {
  dropArea.addEventListener(eventName, unhighlight, false)
})
dropArea.addEventListener('drop', handleDrop, false)
function preventDefaults (e) {
  e.preventDefault()
  e.stopPropagation()
}
function highlight(e) {
  dropArea.classList.add('highlight')
}
function unhighlight(e) {
  dropArea.classList.remove('active')
}
function handleDrop(e) {
  var dt = e.dataTransfer
  var files = dt.files
  handleFiles(files)
}
function handleFiles(files) {
  files = [...files]
  files.forEach(previewFile)
}
function previewFile(file) {
  let reader = new FileReader()
  reader.readAsDataURL(file)
  reader.onloadend = function() {
    let video = document.createElement('video')
    video.setAttribute('volume', '0');
    video.setAttribute('height', '300');
    video.setAttribute('autoplay', true);
    video.setAttribute('loop', true);
    video.setAttribute('playsinline', true);
    video.setAttribute('oncanplay', 'this.muted=true');
    video.src = reader.result
    document.getElementById('gallery').appendChild(video)
  }
}
</script>

Management on old ILO instances

We bought some servers a while ago, but these have old ILO versions (2).

To manage these servers via ILO was no problem until modern browsers refuse to connect to these web services, because of TLS 1.0 issues.

So what i did was using a second user account on my workstation with a old (downloaded from a ESR archive) version firefox. To administer the ILO

wget https://ftp.mozilla.org/pub/firefox/releases/50.0/linux-x86_64/en-US/firefox-50.0.tar.bz2
extract in other users homedir

usage:
# ssh with X forwarding and start old version
ssh -X otheruser@localhost firefox/oldfirefox

While this was working for me on a debian based machine it didn’t work for my friend who was using Fedora on Wayland.

So i made a more generic solution which would work always. Also when working from windows.

I downloaded a old Fedora version ISO.
https://archives.fedoraproject.org/pub/archive/fedora/linux/releases/15/Fedora/x86_64/iso/
Using the DVD iso i knew the old JAVA was present.

So i started virt-manager and created a new virtual machine

Select your downloaded Fedora 15 ISO

Where is the thin option?!??!

Create a disk image for the OS, don’t worry about the size we are going to shrink it to a minimum (thin provisioned)

Booting from ISO
Do not forget to tick Customize now

Disable all thats not needed! .. Only Gnome Graphical internet .. and JAVA

Create users, complete the install and reboot.
Test your installation.
Shutdown

sudo qemu-img info /var/lib/libvirt/images/fedora15.qcow2
When above gives you a RAW image, you need to convert from RAW to QCOW first.
Mine showed a 9G qcow2 image .. far to large

sudo apt install guestfs-tools
virt-sparsify -v fedora15.qcow2 fedora15spars.qcow2

9665380352 okt 10 13:05 fedora15.qcow2
2926444544 okt 10 13:21 fedora15spars.qcow2
From 9 down to 3GB
Browser works

Testing the Java applet console

Success!

Generic Matrix Keyboard Print

I only have to check my new design for a generic matrix keyboard once again, and then i’m going to order a few of these prints.

  • 32 keys (2 blocks of 16)
  • 1 key is changeable from generic key to reset signal key
  • jumper header connector VCC, GND, DataAvailable, Reset and 5 data lines

I’ve used the opensource tool freeroute to autoroute the coppertraces

  • From Kicad export PCB in Spectra DNS format
  • Open Freeroute and run autoroute
  • export as Spectra Session File
  • In kicad import this session file again

Autotraced …
Front
Back
Current setup

TODO: Look into charlieplexing ! For input and Led control

Raspberry and SSD

I had a crash recently on one of my raspberry-pi’s ..
SDcard failure, they are not made for a lot of write actions.
In the past i’ve changed some images to read only and with a r/w overlay.
Also tmp filesystems in memory .. all not ideal.

So i’ve started to make every RPi ssd bootable.

I’ve got several ssd already from other projects.
Sata to USB adaptors are cheap, only a few euro’s.

Steps to take:

  • Download Raspberry Pi Imager tool
  • Choose OS > Misc Utility Images > Bootloader > USB Boot
  • Select storage and write to a temporary sd-card (Not needed any more after flashing for normal operations)
  • Boot USB with this Micro-SDcard .. i didn’t have a screen connected .. So i just waited a few minutes
  • While i was waiting i wrote a OS image to the SSD using the same imager tool
  • Choose OS > select sata/ssd drive
  • Change options (cog), enable ssh, choose hostname and set password
  • Write to drive
  • Remove sdcard from RPi attach ssd/sata and boot

My 3D printed sdcard case, luckily there was still one in there (32GB kindda big, but it was only for temporary use .. 16GB was broken ..

So .. without attaching a screen or keyboard, just a network cable. I have a running OS on a RPi from SSD

Motion detection for stupid webcams

Last year i made a script for a friend who wanted to detect visually if his garden sprinkler was on or off. A few days ago i saw someone who wanted to see if things where moving in his house. (didn’t trust his landlord i think)
But he only had a dumb/simple/cheap camera .. so it had no motion detection.

I was thinking of my script, and could easily adapt it for this usage.

Most ipcams have somekind of URL/API you can use to capture a image.
Some examples below

# Reolink
wget "http://ipnumber/cgi-bin/api.cgi?cmd=Snap&channel=0&rs=randomstring&user=user&password=password" -O image.jpg

# Foscam
wget "http://ipnumber:88/CGIProxy.fcgi?cmd=snapPicture2&usr=user&pwd=password&cnt=randomstring" -O image.jpg

# Edimax
wget "http://user:password@ipnumber/jpg/image.jpg" -O image.jpg

# Hikvision
wget "http://user:password@ipnumber/ISAPI/Streaming/channels/101/picture" -O image.jpg

So using below script i can capture a image, compare it to the previous, and when it’s above a certain threshold sends a email.

#!/bin/bash
# Only uses wget and image-magick
treshhold=500
fuzzyness=20%
# CHANGE WEBCAM THINGY TO OWN URL AND CREDENTIALS
wget -q "http://webcamip/cgi-bin/api.cgi?cmd=Snap&channel=0&user=user&password=password" -O previous.jpg
while true;  do
wget -q "http://webcamip/cgi-bin/api.cgi?cmd=Snap&channel=0&user=user&password=password" -O current.jpg
value=$(compare -fuzz $fuzzyness previous.jpg current.jpg -metric mae diff.jpg 2>&1 | cut -f1 -d.)
if [ $value -gt $treshhold ] ; then
echo "ping $treshhold"
echo "Something moved" | mail -s "Movement" user@example.com -A diff.jpg
fi
# Comment below if you want to compare against a base line .. not previous image
cat current.jpg > previous.jpg
sleep 60
done

Example previous picture

Example current picture

I got mailed with result

Hints tips:

Use crop to detect only a part.

copy current.jpg to a second file
Use painting black a part and compair with different treshhold fuzzyness to get different hotspots.

Below detects RED, use above ide with crop to detect red/green/blue leds

compare -verbose -metric mae 1.jpg 2.jpg /tmp/1.diff
1.jpg JPEG 2560x1920 2560x1920+0+0 8-bit sRGB 248819B 0.050u 0:00.057
2.jpg JPEG 2560x1920 2560x1920+0+0 8-bit sRGB 248949B 0.030u 0:00.137
Image: 1.jpg
  Channel distortion: MAE
  Channel distortion: MAE
    red: 12517.5 (0.191005)
    green: 11967.1 (0.182607)
    blue: 12492.8 (0.190628)
    all: 12325.8 (0.18808)
1.jpg=>/tmp/1.diff JPEG 2560x1920 2560x1920+0+0 8-bit sRGB 1.19495MiB 1.470u 0:00.197

Glade and python – network tester

THIS IS A WORK IN PROGRESS ! .. Updates follow

UPDATE: Found a glade project from 2002
https://www.henriaanstoot.nl/2002/02/20/reverse-engineering-a-alpha-ticker-led-scoller/

The goal of this project is to have a raspberry-pi with a screen wich shows network information.
It wil be using a battery, touchscreen .. maybe some status leds.
When debugging network issues we want to have information when/if/how a network port works on our switches.

It should show:

  • dhcp ip
  • gateway
  • can access internet?
  • speedtest
  • detect if vlan tagged network packets are present on the port?
  • icmp test
  • list of detected nearby hosts?

A long time ago i played with glade and C / Perl.

But i’d rather use python so i’m looking into glade/python combi for this little project.

Glade is a gnome/GTK user interface RAD tool. (Rapid Application Development)

i’ve used zenity and yad before to create simple gui’s for bash scripts, these where only for quick and dirty solutions. (See other posts)
Glade is a far better solution, but a little harder to use.

Below is a little framework i started with

Python script

import gi

gi.require_version("Gtk", "3.0")
from gi.repository import Gtk

class Handler:
    def onDestroy(self, *args):
        Gtk.main_quit()

    def on_firstbutton_clicked(self, button):
        print("Ping test")

builder = Gtk.Builder()
builder.add_from_file("mytest.glade")
builder.connect_signals(Handler())

window = builder.get_object("Main")
window.show_all()

Gtk.main()

Glade file

<?xml version="1.0" encoding="UTF-8"?>
<!-- Generated with glade 3.38.2 -->
<interface>
  <requires lib="gtk+" version="3.20"/>
  <object class="GtkWindow" id="Main">
    <property name="can-focus">False</property>
    <property name="title" translatable="yes">Networktool</property>
    <property name="default-width">440</property>
    <property name="default-height">250</property>
    <property name="icon-name">network-wired</property>
    <child>
      <object class="GtkFixed" id="fixed1">
        <property name="visible">True</property>
        <property name="can-focus">False</property>
        <child>
          <object class="GtkButton" id="firstbutton">
            <property name="label" translatable="yes">Ping test</property>
            <property name="width-request">100</property>
            <property name="height-request">16</property>
            <property name="visible">True</property>
            <property name="can-focus">True</property>
            <property name="receives-default">True</property>
            <signal name="clicked" handler="on_firstbutton_clicked" swapped="no"/>
          </object>
          <packing>
            <property name="x">56</property>
            <property name="y">40</property>
          </packing>
        </child>
        <child>
          <object class="GtkButton" id="speedtest">
            <property name="label" translatable="yes">Speed test</property>
            <property name="width-request">100</property>
            <property name="height-request">16</property>
            <property name="visible">True</property>
            <property name="can-focus">True</property>
            <property name="receives-default">True</property>
          </object>
          <packing>
            <property name="x">56</property>
            <property name="y">89</property>
          </packing>
        </child>
        <child>
          <object class="GtkTextView">
            <property name="width-request">179</property>
            <property name="height-request">166</property>
            <property name="visible">True</property>
            <property name="can-focus">True</property>
          </object>
          <packing>
            <property name="x">222</property>
            <property name="y">36</property>
          </packing>
        </child>
      </object>
    </child>
  </object>
</interface>

Which is generated using the Glade designer

When running you get below screen

Old glade program (2003)

"If something is worth doing, it's worth overdoing."