Midnight ProwlerBot

3D Scan it and 3D Print

2022-11-25T17:13:00.008-08:00

The availability of 3d printing to hobbyists has provided an existing and interesting pastime to hobbyists. For instance, electronics enthusiasts used to buy electronics enclosures for projects, but the enclosure was always a compromise. There was rarely ever an enclosure available that fit a project perfectly. It was possible to create such a design, but it required electronics hobbyists to be amateur product designers, and then electronics designers if they wanted a good looking result.

With 3d printing, an electronics hobbyist may design their PCB and then create the enclosure exactly as needed. However, this requires 3d design skills that are difficult for some to learn.

So, many new 3d printer owners resort to downloading designs from 3d printing repositories but this is a compromise. It's really no different from buying an electronics enclosure that "kinda" works other than the fact that the download may be edited if you have the skills. What if there isn't an enclosure available for download that is, "close enough?"

This is where 3d scanning can help, but until lately it has been expensive and unapproachable. Usually some type of LIDAR based handheld gadget costing thousands of dollars is needed. There are DIY possibilities, such as using a Microsoft Kinnect to do the scanning, but getting good scanning results can be challenging.

Latest versions of the Apple iPhone make 3d scanning easy as they use in-built LIDAR sensors to effectively replicate a multi-thousand dollar handheld 3d scanner. Not everyone can afford the latest iPhone and not everyone wants an iPhone but what other options do you have for an affordable 3d scanner?

Photogrammetry is a 3d scanning technique that uses a collection of photographs to create a 3d representation. Unlike the traditional portrayal of a 3d scanner, wherein there's a laser line reflecting from an object on a rotatating table, photogrammetry needs no line - no LIDAR, no laser.

There are photogrammetry based 3d scanning apps available on Android phones. The better your phone camera resolution, the better your results will be. However, for best results, you will need to take A LOT of photos. More photos helps the photogrammetry algorithm create a better representation of your object.

My first 3d scanning attempt was with an Android app called Kiri. Kiri is free and your scans can be exported individually. This is an important point since some apps require a monthly subscription to be able to export your scans. With Kiri, you can export the object that you just scanned in and pay only for that object.

I also tried an app for Android phones called WIDAR. The app is nice and unlike Kiri, Widar provides an editing environment on your phone that is impressive. Kiri provides basic transforms like rotation. Widar provides much more - you may not even need to edit your exported scan in another program.

Widar charges $4.99 a month if you want to do exports. There's an option to access your export for free through a generated link, but I was unable to figure out how to get the export from the link. From what I see Widar is $4.99 a month if you want to export your scans. If you plan on doing a lot of scanning and exporting, $4.99 a month may not be a bad deal.

With Kiri, you pay for what you export, well not quite. Kiri also provides free exports, so, you get so many exports per week for free. If you need to do an export and you're out of free ones, you can either buy an export, which is very affordable, or earn a free one by referring someone to Kiri.

My first attempt with Kiri was terrible. I tried scanning in a Hot Wheels Dodge Challenger. I took 27 photos from all sorts of angles. Kiri processed the results on their servers, not my phone, and then notified me that my results were ready. I had scanned a flat multi-colored blob. The car color was recognizable, but not the shape.

I tried again but this time with 16 inch tall Christmas tree. It worked! Fifty photos and I had an exact duplicate of the tree. However, it's not that simple.

There's a lot of artifacting created by the photogrammetry process; so, you will need to edited your exported 3d scan. This is intimidating and foreign at first but easy once the process is understood.

You'll need something to edit the STL/OBJ file. Tinkercad is an obvious choice since it imports STL files and makes editing them easy. Unfortunately, it is limited to 20000 triangles; so, a large complex scanand provides tools fr editing will be rejected by Tinkercad.

Meshmixer is often mentioned as an option for editing STL files. I'm not a fan of it. It's a bit like using Blender or a sculpting program to do edits.

Fusion360 will import STL/obj files and provides tools for editing STL files but it takes a bit of getting used to. Product Design Online is an excellent Youtube channel that can help with learning how to edit STL files.

After your scan has been cleaned up and looks great, you can 3d print it! Your object can be scaled or changed as you wish and then created on your 3d printer. Even if you are not adept with 3d editing, you can now scan in an object, change it to your needs, and then 3d print it.

What's Wrong with Octoprint?

2022-01-17T05:15:00.007-08:00

I started a 3d print running from Octoprint, as I often do, and after 30 minutes of printing, the printer halted.

Octoprint had tried to send messages a number of times and they went unacknowledged. I didn't have the serial.log enabled, so that was all the information that I got. I enabled the serial.log and tried 3d printing again. Thirty minutes later, the Creality Ender 3 Pro halted and there was a notification on Octoprint of too many communications timeouts.

At least this time I had a log file. I scanned the log file and found six failed attempts to communicate. No other helpful information was in the file. Now what?

I came across an odd recommendation suggesting that it could be the Ender 3 LCD cable position causing communications issues. I moved the cable and tried 3d printing again. A 2 hour print ran successfully! Problem solved! Not quite.

My next 3d print, 3.5 hours long, failed after 2.5 hours. Same issue, too many failed communications attempts.

I tried a new USB cable. I tried using a different microUSB card. I tried using a different USB port on the Raspberry Pi 4 running Octoprint. I tried uninstalling and reinstalling the serial port driver. I tried a whole new install of Octoprint.

It must be the Raspberry Pi 4, it does run a bit warm. Nope, it wasn't the Raspberry Pi 4. That left only the Creality Ender 3 Pro as the culprit.

I checked wire connections and especially ground connections on my Ender 3 Pro. I even considered making my own noise shielding with aluminum foil wrapped around the LCD cable and grounded but fortunately, I tried something on a whim first.

I removed the 3d printed plastic cover for the back of the LCD that I'd 3d printed 2.5 years ago.

As I began to remove the screws that held the cover to the back of the LCD, I noticed that one screw was very loose and the other three srews were not much better. Oh well, cover removed and LCD screws tightened down, I ran a 7 hour 3d print. Because I was tired of wasting filament, I ran the Ender 3 Pro dry. No filament. I used the same settings as always but I wasn't wasting any more plastic.

It worked!

I examined the back of the LCD a little more closely and realized that the LCD ground connection to the frame was dependent on those 4 mounting screws. Loose screws could jiggle and create a ground make/break cycle that could worsen with vibration.

I got rid of that LCD back cover. I may find or make another one that doesn't use the LCD PCB screws for mounting. Then again, maybe not.

If you're having communications trouble with Octoprint to your 3d printer, always check every single ground connection and path. It never occurred to me that grounding to the LCD could affect serial communications, but only because I haven't looked at that area of the printer for over 2 years.

It is Alive! Alive!

2021-11-16T21:47:00.004-08:00

I've created a Boston Dynamics Spot inspired quadbot and finally gotten a servo to move as commanded. Moving a servo may seem trivial at first, but consider that it's a Raspberry Pi 3a+ communicating with a Lynxmotion SSC-32.

The first issue is that the SSC-32 is a 5vdc unit but the Raspberry Pi GPIO is 3.3vdc.

A level shifter circuit is easy to make with an N-channel MOSFET and two 10k resistors; however, three 10k resistors as a voltage divider is even easier. I'll change it later to a proper level shifting circuit but for testing it's fine.

A voltage divider may cause damage if ground gets broken from the voltage divider: all 5 volts then end up going directly to Raspberry Pi GPIO instead of only 3.3vdc.

The second issue was configuration related. The hardware UART used with Bluetooth can get tricky which can cause issues at 115200 baud or higher. So, I disabled Bluetooth and set comms at 38400 baud. Now to write the Python code.

First, let's get the SSC-32 to tell us the version number to confirm that comms is working correctly. The issue that I had was that I couldn't just send a string through the UART. It had to be encoded first which meant assigning the string and return command to a variable and then encoding the variable.

If "VER/r" is sent to the SSC-32, the value can be read back using readLine() and then displayed using print().

Next, we finally move a servo. Assign "#31 P1200/r" to a variable, encode it, and send it over UART to send a 1.2ms move command to the servo on port 31. It took me a while to realize that #1 servo was not nearest the RS232 port on the SSC-32 - I had my servo connected to #31, not #1, so double-check the port your servo is connected to.

Also, remember to set the jumpers on the SSC-32 for 38400 baud, and the VS servo power jumpers to provide power to both servo banks, and lastly remember to set the VL jumper for logic power or power it from a separate battery.

Now, I can now begin coding in walking sequences!

This is the work in progress with only one zero connected:

The following code will move a servo back and forth continuously. You'll need to ctrl-Z on the RPi to stop it or stop it in the Python IDE. Here's the source code for simple communication between a Raspberry Pi and an SSC-32:

#START

import serial
import sys
import time

ssc32=serial.Serial('/dev/serial0', 38400, timeout=1.0);
print("start")
tempmsg = "VER\r"
ssc32.write(tempmsg.encode())
time.sleep(1)

if ssc32.inWaiting() > 0:
   data = ssc32.readline()
   print(str(data))

while True:
   time.sleep(1)
   tempmsg = "#31 P1200 \r"
   ssc32.write(tempmsg.encode())
   time.sleep(1)
   tempmsg = "#31 P1800 \r"
   ssc32.write(tempmsg.encode())

#note that the While loop is infinite so we never get here, but
#that may change, so it's good to remember to close the port.
print("end")
ssc32.close()

#END

My First Magnetic Tensegrity Table

2021-10-09T08:58:00.011-07:00

Google defines tensegrity as:

"the characteristic property of a stable three-dimensional structure consisting of members under tension that are contiguous and members under compression that are not."

Sometimes it's just easier to see it than read about it so here is a photo of a popular tensegrity toy.

A tensegrity table uses the same principles to hold two flat surfaces locked at a fixed distance from each other using only strings or some other flexible strand such as ball chain. Some tensegrity designs are fascinating to look at as the top surface looks like it's floating in mid-air.

I wanted something a little different. I wanted a table surface to magnetically levitate using tensegrity principles. How hard could it be? Replace the strings with supermagnets and that's it right?

I started by printing a few simple surfaces to retain the flat disc neodymium magnnets that I had purchased and decided to use the attractive force but use it along the outside edges with a tensioned string in the center to tension the structure. The attractive distance of the magnets was not as great as I expected. The two surfaces could be no more than an inch (2.54cm) or so apart for this to work. That would leave almost no room for the tension structure needed in the center.

So I used magnetic attraction for the center string and three fishing lines around the outside edge. It finally worked!

I created the design to be very sturdy but to look almost wireframe. The wireframe look appeals to me as it's similar to looking at a bridge structure.

One of the problems with working with neodymium magnets is that they are powerful. My first attempt at this circular table used a single arm of less than 10mm width to support each magnet. The support arms were bent by the magnets! Too close, and the magnets snapped together. Too far, and there was no tension.

I like the overhead view of the structure because it reminds me of a dart board or maybe a "pop" filer that would be used on a microphone in the 1940's.

Unfortunately, I did not align the two solid magnet supports so that there was a straight on view without one of the structures overlapping.

As a result, when someone first looks at it they assume the center is how the table is supported. It isn't until they look more closely that they see that the bottom and top half are connected only by strings. Finally they realize that the neomagnets are keeping the table together and then they begin testing the table.

First, they always push down on it to see how strong it is and they're usually a bit surprised. After collapsing the table, they pick it up and the neomagnets snap the table back into shape.The last thing people often do is try to spin the top to see how far it'll turn. They're sometimes a little surprised that it won't turn very far but mesmerized by the jiggling of the table as it snaps back to original position.

I think the tensegrity table looks much better after changing the hole alignments and printing a table section with better alignment.

I will be continuing to investigate other ways to use magnets in tensegrity structures. Download the STL and 3d print your own at Printables.com.

Free HDTV Antenna Hack

2021-09-14T02:54:00.006-07:00

Mohu Leaf antennas are popular for their small size, reception, and ease of mounting, but what do you do about TV stations that are too far away?

Typically, that means that a bigger antenna is needed. Sometimes a TV channel has great reception and yet other times, you try shifting the antenna a little bit or even moving the coax cable. Maybe a new antenna would give better reception, but that can be a gamble since not all antennas provide the same performance.

You may remember the days when "rabbit ears" were on top of TVs or seen old photos of TVs with rabbit ear antennas. Sometimes folks would put a little aluminum foil on the antenna to improve reception.

Well, you can do something similar with a Mohu Leaf antenna to improve reception but instead of attaching aluminum foil to the antenna, you make a reflector.

I scrounged around and found a piece of compressed cardboard that was just a little larger than my antenna. Corrugated plastic, or presentation board would also work since it just needs to be stiff.

I had a metal template for spreading solder paste that was no longer needed that was just the right size. My metal sheet was four inches wider than the antenna and close to the same height. I centered the antenna on the metal and bent each side of the metal to a 45 degree angle or so.

If you don't happen to have an unused metal soldering template laying around then you can use any thin metal, even heavy duty aluminum foil will work. If you use window screen material, make sure that it is actually metal. Do not use chicken wire since the large spaces offer less reflection area.

After bending the metal, I used double-stick tape to secure the metal to the cardboard. Next, I punched two screw holes through the metal and cardboard that aligned with the two holes on the antenna. I guessed on the location of the third and fourth hole and just made sure that the would be under the opposite edge of the antenna.

The most important part of this antenna modification is the distance from the reflector to the antenna. Two inches provided the best results for me. Depending on the channels that are weakest in your area, you may want to adjust the distance slightly. More distance will improve reception for lower frequencies while shorter distance will improve reception for higher frequencies.

I 3d printed two inch standoffs but you could just as easily use wooden dowel or other non-metal materials. For the area of the antenna with no screw holes, I created a standoff with a lip to hold the edge of the antenna. Although it may be tempting, do not pierce the antenna!

This is the final result. I put a sawtooth picture hanger through the cardboard to make it easy to hang the unit. My final cost was zero since I used scraps that I already had.

So the real question is performance. Even if the mod was free, it's useless if there's no increase in performance.

I used a TiVo OTA unit to check the difference in antenna performance. It's not scientific or even in decibels, but it will display antenna signal percentage, which we can compare.

The weakest channel that I could receive measured 35% before the reflector and 60% after the reflector. I can now reliably receive that channel without screen garbling randomly.

Every other channel showed an increase of at least 15% with the stronger channels increasing by 25%. I used to loose some channels at night as their signal was weaker at night, but that is no longer an issue.

Even if you have to buy everything, it should be possible to make this reflector and improve your reception for only a few dollars.

Play the MacGyver theme when your 3d print ends

2021-09-01T12:50:00.005-07:00

If you have owned a 3d printer for a while then you've probably already added code to beep the 3d printers speaker to notify you that your print has completed. Otherwise, your stuck occasionally checking the print progress, which can become tiresome.

Why not play a favorite tune when your 3d print has completed? Maybe a fanfare? Celebration? "Flight of the Valkyries?"

With the tool at the following link, you can convert a midi file directly to GCODE that will output tones on your 3d printers speaker: Midi to M300

The M300 GCODE command is used by 3d printers running Marlin firmware to generate sounds, tones. Many 3d printers use the Marlin codebase so if your 3d printer has a speaker then you should be able to generate music.

The music will not sound like music from your smartphone might sound. It actually sounds more like a Commodore VIC-20 could be playing the music but it is recognizable and guaranteed to arose curiosity from those not expecting it.

You can of course generate your own midi files like Lazy Game Reviewer (LGR) might, but there are also numerous midi songs available on the Internet.

Since I'm no musician but still am a fan of the original MacGyver show, I found the theme song in midi form and ran it through Midi to M300 and tried it on my printer.

It sounded terrible!

So I went back to Midi to M300 and unchecked boxes for some instruments and channels and tried again. I had to repeat this several times before I got something that sounded good to my ears. However, it was too lengthy. I didn't need to hear the entire MacGyver theme at the end of a print, just enough to be recognizable and get my attention.

So I started trimming M300 lines from the generated code. The final result has been shorted to around 30 seconds of time and contains the most recognizable parts of the tune.

If you'd like to just try this out on your 3d printer then install a copy of Pronterface and connect to your 3d printers USB port. Some Creality printers now use the CH340 serial USB driver so you may need to install it to get Pronterface working with your printer - it does not autoinstall on Win10.

Copy the GCODE for the MacGyver theme below and save it using Notepad as a text file. Change the extension of the file to MacGyver.GCODE. After Pronterface is connected to your printer, Load the MacGyver.GCODE file in Pronterface.

Now press the Print button in Pronterface and your 3d printer should play the MacGyver theme. If it is unsuccessful, and your certain that Printer face is connected, double check the MacGyver.GCODE file that you saved and make sure that it only contains the M300 GCODE lines shown below and nothing else.

To play the MacGyver theme at the end of a print, open up your printer configuration in Cura slicer and go to the "End G-code" under the Machine Settings for your printer and paste the following GCODE:

M300 P104 S523

M300 P104 S0
M300 P104 S523
M300 P104 S0
M300 P104 S523
M300 P104 S0
M300 P104 S523
M300 P104 S0
M300 P104 S523
M300 P104 S0
M300 P104 S523
M300 P104 S0
M300 P104 S523
M300 P104 S0
M300 P104 S523
M300 P104 S0
M300 P208 S494
M300 P208 S392
M300 P208 S196
M300 P208 S392
M300 P208 S370
M300 P208 S440
M300 P208 S196
M300 P208 S392
M300 P208 S392
M300 P208 S392
M300 P208 S196
M300 P208 S392
M300 P208 S523
M300 P208 S523
M300 P208 S196
M300 P208 S494
M300 P208 S196
M300 P208 S523
M300 P208 S494
M300 P208 S440
M300 P208 S392
M300 P208 S330
M300 P208 S659
M300 P208 S330
M300 P208 S440
M300 P208 S262
M300 P208 S131
M300 P208 S262
M300 P208 S523
M300 P208 S523
M300 P208 S147
M300 P208 S494
M300 P208 S196
M300 P208 S392
M300 P208 S196
M300 P208 S392
M300 P208 S370
M300 P208 S440
M300 P208 S196
M300 P208 S392
M300 P208 S196
M300 P208 S392
M300 P208 S196
M300 P208 S392
M300 P208 S523
M300 P208 S523
M300 P208 S196
M300 P208 S494
M300 P208 S196
M300 P208 S523
M300 P208 S494
M300 P208 S440
M300 P208 S392
M300 P208 S330
M300 P208 S659
M300 P208 S330
M300 P208 S440
M300 P208 S262
M300 P208 S131
M300 P208 S262
M300 P208 S494
M300 P208 S247
M300 P208 S123
M300 P208 S247
M300 P208 S523
M300 P208 S494
M300 P208 S440
M300 P208 S523
M300 P208 S185
M300 P208 S494
M300 P208 S440
M300 P208 S587
M300 P208 S123
M300 P208 S523
M300 P208 S494
M300 P208 S587
M300 P208 S123
M300 P208 S523
M300 P208 S494
M300 P208 S659
M300 P208 S131
M300 P208 S587
M300 P208 S659
M300 P208 S740
M300 P208 S185
M300 P208 S494
M300 P208 S123
M300 P208 S784
M300 P208 S165
M300 P208 S330
M300 P208 S740
M300 P208 S311
M300 P208 S698
M300 P208 S294
M300 P208 S494
M300 P208 S392
M300 P208 S784
M300 P208 S659
M300 P208 S494
M300 P208 S740
M300 P208 S587
M300 P208 S440
M300 P208 S659
M300 P208 S523
M300 P208 S392
M300 P208 S587
M300 P208 S494
M300 P208 S392
M300 P208 S523
M300 P208 S330
M300 P208 S494
M300 P208 S294
M300 P208 S523
M300 P208 S494
M300 P208 S440
M300 P208 S392
M300 P417 S466
M300 P417 S440
M300 P208 S784
M300 P208 S392
M300 P208 S587
M300 P208 S392
M300 P208 S622
M300 P208 S311
M300 P208 S466
M300 P208 S440
M300 P208 S392
M300 P208 S196
M300 P208 S294
M300 P208 S196
M300 P208 S311
M300 P208 S196
M300 P208 S233
M300 P208 S220
M300 P208 S587
M300 P208 S587
M300 P208 S587
M300 P208 S587
M300 P208 S587
M300 P208 S587
M300 P208 S587

Make your 3d printer play music - "How It's Made"

2021-08-25T11:42:00.008-07:00

I found a midi file of the intro music from the TV show, "How It's Made" and found an online midi to GCODE converter. Perfect!

If you want your 3d printer to start playing the "How It's Made" tune through the printers speaker when starting a print then add this GCODE to the end of your printers start code:

;How its Made tune

M300 P103 S123

M300 P635 S0

M300 P117 S123

M300 P337 S0

M300 P137 S123

M300 P190 S0

M300 P107 S123

M300 P249 S0

M300 P166 S110

M300 P1724 S0

M300 P93 S156

M300 P620 S0

M300 P93 S156

M300 P347 S0

M300 P156 S156

M300 P200 S0

M300 P117 S156

M300 P234 S0

M300 P107 S165

M300 P1299 S0

M300 P464 S185

M300 P5 S311

M300 P103 S123

M300 P371 S0

M300 P210 S311

M300 P34 S0

M300 P20 S370

M300 P117 S123

M300 P293 S0

M300 P44 S277

M300 P137 S123

M300 P190 S0

M300 P127 S123

M300 P200 S311

M300 P29 S0

M300 P166 S110

M300 P68 S0

M300 P132 S311

M300 P337 S0

M300 P117 S311

M300 P117 S0

M300 P117 S370

M300 P352 S0

M300 P176 S277

M300 P288 S0

M300 P20 S311

M300 P93 S156

M300 P366 S0

M300 P210 S311

M300 P34 S0

M300 P10 S370

M300 P93 S156

M300 P327 S0

M300 P20 S277

M300 P156 S156

M300 P200 S0

M300 P122 S156

M300 P200 S311

M300 P29 S0

M300 P107 S165

M300 P127 S0

M300 P132 S311

M300 P337 S0

M300 P117 S311

M300 P117 S0

M300 P117 S370

M300 P352 S0

M300 P176 S440

M300 P59 S0

M300 P176 S494

M300 P54 S0

M300 P5 S622

M300 P103 S123

M300 P371 S0

M300 P210 S622

M300 P34 S0

M300 P20 S740

M300 P117 S123

M300 P293 S0

M300 P44 S554

M300 P137 S123

M300 P190 S0

M300 P127 S123

M300 P200 S622

M300 P29 S0

M300 P166 S110

M300 P68 S0

M300 P132 S622

M300 P337 S0

M300 P117 S622

M300 P117 S0

M300 P117 S740

M300 P352 S0

M300 P176 S554

M300 P288 S0

M300 P20 S622

M300 P93 S156

M300 P366 S0

M300 P210 S622

M300 P34 S0

M300 P10 S740

M300 P93 S156

M300 P327 S0

M300 P20 S554

M300 P156 S156

M300 P200 S0

M300 P122 S156

M300 P200 S622

M300 P29 S0

M300 P107 S165

M300 P127 S0

M300 P132 S622

M300 P337 S0

M300 P117 S622

M300 P117 S0

M300 P117 S740

M300 P352 S0

M300 P176 S880

M300 P59 S0

M300 P176 S988

The GameSharp Game Controller is Finally Finished

2020-04-28T11:57:00.000-07:00

I love the idea behind gaming on a smartphone but I detest the touch controls used on smartphones. I'm obviously not alone in that opinion as many people buy clips to attach their PS4 controllers to their smartphones to play games.

I wanted something with a more retro "feel" and really just wanted to create my own controller. After a little research, I found that Microchip's RN42 was often used for interfacing a game controller to a phone; so, I decided to use an RN42.

However, before using the RN42, I first tried a nRF52 Bluefruit LE.

The Bluefruit LE is an excellent device that is easy to use. Just write a little Arduino script and upload it and that's all there is to to it. Configuration of the Bluetooth section can be particularly challenging especially since the device is not Bluetooth. It is Bluetooth LE which follows the Bluetooth Low Energy specification.

Unfortunately the Bluetooth LE specification does not include any config for a game controller. It includes configuration for a keyboard and mouse but no game controller. If I wanted to emulate a mouse and keyboard, as is sometimes used in RetroArch, it could easily be done. I wanted two analog joysticks with my game controller, otherwise I'd have used the BlueFruit module for this project.

I used a PIC16F886 to process the inputs since I no longer had a microprocessor to use when I gave up on the Bluefruit module. Nowadays the Arduino is the popular option but I already had a Pickit2 programmer from Microchip and several PIC chips to choose from. Besides, I was also more familiar with the PIC chips than Arduino.

The 3.3 volt power supply for the the RN42 Bluetooth module was a simple matter of putting a voltage regulator on the 5 vdc power section. I had to do a little conversion of the 3.3 vdc I/O between the 5 vdc PIC and the 3.3 vdc RN42 using a MOSFET and resistor for each I/O point.

The battery charging circuit was easy enough as it was just a matter of following the directions in the datasheet for the MCP73831 that I used to charge the lithium polymer battery.

The 5 vdc power supply was a bit of a nightmare. At first, I used a TPS61090.

I have a heat wand so I ordered a stencil for my PCB, slathered some solder paste across the stencil, made sure there were no overlaps, placed the chip, and soldered it. After applying power, the chip got so hot that it slid off of the PCB!

I'd made a mistake somewhere in my design.

I reviewed my design and examined my solder job and figured that I must have shorted a few pins. I tried again but this time there was only smoke. I gave up and found a chip that was a little friendlier to the home electronics enthusiast.

I selected the LM2735 from Texas Instruments for the 5 vdc supply to replace my failed efforts with the TPS61090.

First of all, the LM2735 could be soldered with a soldering iron, which I prefer over the heat wand. More importantly, Texas Instruments has an online design aid for many if not all of its' power management chips called WebBench. All I had to do was put in my parameters and a schematic was displayed as well as a PCB layout recommendation and even a bill of materials. They make it so convenient that it's even possible to substitute components in the bill of materials and the design will update. The changes are intended to not change the performance of the design however some item changes do induce slight differences in performance and TI WebBench will display that. What a great application!

I finally got the GameSharp game controller working.

There is an LED to indicate Bluetooth module status, an LED to indicate that Bluetooth is connected, an LED to indicate 5vdc power status, and an LED for battery status. Initiatlly, I was using a MOSFET that was controlled by the OFF/ON slider switch so that I could use a small switch that did not need to handle a lot of milliamps. I found that I didn't need the MOSFET after all and now power control is handled solely by the slide switch - that's what the bodge wires are about on the bottom right side of the PCB image.

I wanted something that looked like a Nintendo Gameboy so that's how I designed the 3d printed case to look.

The buttons used are straight from an SNES controller. The low profile joysticks were used in keeping with a Nintendo handheld styled theme. I realize the Gameboy did not have joysticks but later Nintendo handhelds did and they were always low profile.

A door on a hinge opens at the top of the unit and the phone is dropped inside.

That will need to be changed eventually since every phone is not the same size. I may end up doing some sort of spring loaded clip from the back side while maintaining the aesthetic as it is on the front.

Overall, I'm happy with the GameSharp results. I still need to polish the PIC software a little to reduce lag and I need to adjust the 3d printed case front a little bit since my button holes ended up a little offset from the button contacts, but other than that, this was a great project that I can now enjoy playing games with on my phone.

For the curious, the cost of the boards was only $16 for 5 from JLCPCB. The boards are larger than 100mm by 100mm so there's a premium price to since the button contacts need to at least be HASL for reliability there's a premium for that too.

The RN42 module was around $15 when I bought it in December 2019. The rest of the parts cost around $35. Add shipping and the total cost of the project with no mistakes would be around $75.

Setting up a new Creality Ender 3 Pro 3d Printer

2019-11-18T15:31:00.001-08:00

I have caught the 3d printing bug. However, I did not want to spend a lot of money on something right away. Besides, as my wife says, "I am thrifty."
I bought the Creality Ender 3 Pro because I had read about the few improvements included in the pro version that were not in the standard version. First is the magnetic bed, I'm not sure if it justifies the $40 price premium, but the 1 year warranty and support and the improved extruder does justify the price premium for me.

The first bit of advice that I could offer in setting up the Creality Ender 3 Pro is to first use the foldout manual to identify everything and lay it out and then put the foldout manual away. It can be followed by a person determined to interpret the images, as I was, or a person ignorant of the build video that comes with the unit on the micro SD card. Plug in the micro SD card adapter and watch the build video. It is much easier to follow than the multifold manual.
After finishing the build I was unable to level the bed. the left side was lower than the right by almost 1/3 of an inch! I took the bed off to see if the bed frame below was that far off from level. It was! Argh!
So I examined the bed frame closely and concluded that it had to be the X-axis bar carrying the printer head that was not level. I was correct.
In my mind, if you have rollers riding a frame and the holes are aligned then the parts must have been created to be square to each other. This is not true. BTW, I'm an electronics and IT guy not a mechanical engineer as should be evidenced by that last statement.
I watched a few setup videos, there's nothing about this issue from Creality BTW, and I learned about the concentric nuts. Concentric nuts on the Creality Ender 3 printers are used to move specific rollers a range of about +- 1/8 of an inch. They function similarly to a cam on a camshaft.

I moved the X-axis to the top and measured each side to check for square. It was off by 1/3 of an inch.
I tried adjusting the concentric nuts on each side of the X-axis. I had to loosen the lock-down nuts a little of course, which I almost forgot to do. I never did get the X-axis square. Every adjustment resulted in either the X-axis being loose or much too snug to move. I finally compromised with being 4 mm from square with the assumption that the bed could be levelled to that.
As I began levelling the bed, I noticed that one corner of the bed was higher than the rest, even completely cranked down. I examined the bed mounting again and noticed that the spring with the bed connector mount was compressed more than the others. That bed connector mount compressed the that corner spring about 1/8 of an inch. No wonder the one corner was higher than the rest and the other 3 corners felt weak and wobbly.

I added spacers to each of the other 3 bed corners to match the corner with the bed connector mount. Now I could finally align the bed! I adjusted each corner to be just 1/2 millimeter from the bed and at last I was ready to print!
The print for the cat model ran for 5 hours and 43 minutes. The top of his head was cut off just below the ears. The filament that came with the printer, which is not on a spool, had tangled and had not fed through the printer for the last 10% of the print.

To make the best of it, I sanded the hole in the top of the cats head and used the resultant container as a toothpick holder.
I retried the cat model print out after untangling the filament and stringing it out like a telegraph wire - I really did not want to hand my wife another cat with the top of his head cut off. The print went well and the quality is actually impressive.

I must admit that I thought that 3d printing was finally ready for the average person because it is so affordable now. The pre-built models offer less frustration certainly but if anything needs to be replaced or becomes misaligned then it could become frustrating for the non-hobbyist.
Some day 3d printing will be as easy as using a toaster and as cheap as an inkjet printer and as maintenance free. It's too bad that we still have years before we reach that point.

The MyPi is Finished!

2019-07-16T03:38:00.000-07:00

The MyPi kit is now complete! A power supply was sourced that would provide plenty of current to run everything and a comparator circuit module was sourced for monitoring the battery voltage.

Testing was done with a few modules being patched in where needed. The PCB's were redesigned to accomodate the new modules and mosfet. We also discovered some problems with overheating that we addressed by adding ventilation holes to the top and bottom of the MyPi.

After 3d printing a variety of speaker covers, a pair of covers was chosen and for the first time, we had a fully complete, MyPi!

Sure we had other MyPi cases that we had printed but this case was complete. At last, there were no errors and nothing missing. Even the boards inside are completely attached with connectors and aligned nicely with wires being used only where absolutely needed.

If you would like to try building your own MyPi from scratch, you will need access to a 3d printer. You will also need to create your own PC boards. The case will eventually be made available on Thingiverse. The color of your MyPi is up to you as you print it out yourself. Also, I intend to make the kit available in August 2019.

Here's a list of what's needed to build your own MyPi:

1 3d printed case top
1 3d printed case bottom
1 3d printed SD Card access door
2 3d printed speaker covers
1 lipo battery 2500mAh
1 USB connector
1 slide switch
1 mosfet
1 two amp boost supply with built-in charger
1 lm393 comparator module
2 six mm buttons with button caps
2 twelve mm buttons with button caps
1 thumbwheel potentiometer
2 class D amplifiers
2 speakers
1 Raspberry Pi 3b+
1 Five inch HDMI LCD
1 DPad PCB
1 Fire buttons PCB
1 Adapter PCB
2 Trigger buttons PCB's
Carbonized rubber buttons for DPad and fire buttons
DPad
Fire buttons
Connectors and header pins; however, it is possible to use solid wire jumpers

The kit will contain everything needed to build a MyPi except a Raspberry Pi, a USB battery charger, and a battery. Depending on demand, a kit may be made available that will include a Raspberry Pi 3a+, a lipo battery, and a USB charger. A comprehensive assembly manual with photos and schematics as well as configuration and troubleshooting tips will be included with the kit.

After I am confident that there are no more changes needed to the circuitry, the overall schematic will be made available on GitHub.

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

The LiPo Shim in the MyPi isn't Going to Work

2019-07-05T08:01:00.000-07:00

The MyPi current load upon booting has been measured at 1.2 amps or so. When running certain emulators the current demand is around 1.1 amps. With nothing happening, the demand seems to sit around 0.8 amps.

The conclusion that the current measurements lead to is that a Powerboost 1000c by Adafruit will not be adequate for the MyPi as it provides only 1 amp. This conclusion was reached early in development and was the reason that the Pimoroni LiPo Shim was selected.

The Pimoroni LiPo Shim provides 1.5 amps of power which is perfect for the MyPi. Low battery detection is also provided so that we can shut down he MyPi before the battery dies; thus, saving the SD card from damage.

The Pimoroni LiPo Shim worked great with the MyPi although we noticed an odd buzzing and screen powering on/off if the MyPi was not shut off immediately when low battery was detected. It was an odd behaviour but tolerable until we tried adding load sharing to the MyPi.

The Powerboost 1000c has load sharing capabilities. The Pimoroni LiPo Shim does not have load sharing. Oh well, so we add a Schottky diode, a mosfet, and a resistor and we could switch between the battery and USB power. We were very wrong.
The basic load sharing circuit described on Zak's Electronics Blog actually works well; however, not with a Pimoroni LiPo Shim.

We must be able to provide the Pimoroni LiPo Shim with power from a battery or USB. When using USB, everything worked great and the battery charged as we used the MyPi while connected to USB power.

When connected to just the battery through the mosfet, the MyPi would shutdown after 10 minutes! The Pimoroni LiPo Shim would indicate low battery and if the MyPi was not shut down it would buzz irritatingly.

We tried an RC snubber across the mosfet. We tried loading the incoming side of the mosfet and even tried loading the input of the Pimoroni LiPo Shim. We finally tried a variety of capacitances and even inductances in various configurations. Nothing worked.

A load sharing supply was found that could provide 2 amps. Perfect! Unfortunately there's no low battery detection included or even an on/off switch.

So we added a mosfet controlled by a slide switch for turning power to the MyPi on/off. That was one issue solved.

However the MyPi really needed some type of battery level detection and response as it could be very frustrating to experience a shutdown right in the middle of a game of Outrun.

A comparator circuit is relatively simple: it's basically an op amp with no feedback so the output snaps from rail to rail. It would be relatively easy to put a LM393 chip and resistors into the design, but it goes against the goal of the MyPi.

The goal of the MyPi is to create a kit built from modules thus making it easy to solder together with minimal wiring and minimal component soldering. Wiring issues account for a large percentage of problems with projects so if wiring can be reduced, so can the potential for problems.

We could have build boards with the LM393 comparator circuit but we were fortunate to find an LM393 comparator module. There'd be no need to do any more component soldering.

Testing has just been completed and the power circuit works, the load sharing works, and low battery detection works.

Now, we just need to design it all together to eliminate as much wiring as possible and the MyPi should be ready to play again!

MyPi - doubling as a Plug and Play system

2019-06-12T06:54:00.000-07:00

One of the less commonly used features of the Raspberry Pi 3b+ is the composite video output. Although many HDTVs have composite inputs, the results do not look good compared to HDMI input.

So why bother with a composite output?

By ordering a special cable, mine cost less than $7, and commenting out two lines of the config.txt file, the video signal of the MyPi is routed from HDMI LCD to the composite video connector on the adapter cable.

This feature effectively means that the MyPi can function similarly to an older Plug and Play gaming unit like the one shown below:

However, instead of just 6 games or so, the MyPi can play thousands of games from Atari 2600 games to Nintendo and even Playstation games on an older CRT TV looking just the way things looked years ago.

Imagine playing two player retrogames with a MyPi. How?

Just add a Bluetooth controller to the MyPi, I added a Steelseries Stratus XL, and it becomes a two-player Plug and Play game unit - you can relive games as you remember them on a CRT TV!

Play games like Rampage with a friend on that old CRT TV or maybe play Double Dragon together.

When you are finished, just change the line comments on two lines in config.txt, reboot, and unplug the composite cable. Your games will display on the 5 inch HDMI LCD. Now you can play games anywhere again.

Building a MyPi

2019-06-06T14:51:00.000-07:00

To test the latest changes made to create the MyPi project, I needed to build a MyPi. The instruction manual to build a MyPi is almost complete and is awaiting final photos of the MyPi. A video detailing how to build a MyPi will be released as soon as we can finish it. Below are details and color photos of building the MyPi. Please keep in mind that there will be changes made leading up to the release of the MyPi as a project kit.

I started off with the latest updated PCB's.

The first step was to solder the buttons to the small trigger buttons boards in each of the corners in the image above. The next step was to solder the adapter board shown in the center above to the Raspberry Pi 3b+ and attach the 5 inch LCD to the Raspberry Pi.

Note the two black header connectors. They will later slide together with the DPad board and Fire Buttons board assembly. Below is an image of the adapter board, Raspberry Pi 3b+, and 5 inch LCD assembled and ready to be installed.

The next steps are to solder the amplifier modules, the volume control, the LiPo Shim, and LiPo battery charger to the underside of the DPad board and Fire Buttons board. For the sake of simplicity, every effort has been made to use circuit modules. Modules such as LiPo battery chargers and audio amplifiers are plentiful and affordable. Besides, it's easier to solder in a module rather than to solder a group of SMD components for most people. Rather than detailing every step of every module connection, here is the completed DPad board - Fire Buttons board assembly.

The open rectangle just above the 'M' in the image above is to allow access to the micro SD card. However, the position and size of the opening is not quite right yet; so, we need to work on it.

The next step is to plug the LCD assembly into the completed DPad board - Fire Buttons board assembly shown below.

The case color has been changed to an off-white color. We decided that the off-white looks better and would look much better later when we apply stickers, speaker covers, or other decorations to the assembly later. Here's a peak at the inside of the case with the back cover removed.

I found that a button did not work on GPIO25 for some reason. Shorting across the button sensing pad with a piece of metal worked fine but using a carbon contact button would not work. So, I changed the button to a different GPIO. I don't believe that there's a problem with my Raspberry Pi, being that the button worked if a direct short was applied, but while troubleshooting the problem, I realized another problem - I would not be able to use the resistive touch screen since I was using inputs that were also used by the screen for touch sensing.

It was at this point, I decided that since there were 40 pins available on the Raspberry Pi 3b+ that it made little sense to ignore the additional GPIO pins. The next PCB version that is now on order, will not have GPIO conflicts with the resistive touch screen inputs - so, I'll configure the touch screen when the new PCB's arrive and are installed.

We all make mistakes. Careful observers will also note that there is no way to plug in the battery into the battery connector. Naturally, the next PCB version will have a correction for the battery plug.

Here it is. The latest version of the MyPi.

I selected a brown sticker with a rainbow as a tribute to the computer that started it all for me - the Commodore Vic-20.

We have a few details to work out before we can make the MyPi kit available. The first issue is verifying that my battery load sharing circuit works correctly. I wanted a handheld gaming unit to play retro games that would have a large screen to improve visibility. I also wanted stereo sound. The result is a power demand that is just over 1 amp at times, which means that using modules such as the Adafruit Powerboost 1000c will not work well. Before replacing the Powerboost 1000c, it was quite common to see a little lightening bolt in the upper right corner of the screen.

We are working on the speaker cover appearance and additional stickers for identifying controls and improving the overall appearance. I'm not sure if we will be selling a MyPi kit with the white case and brown highlights similar to a Vic20; however, there will always be at least one MyPi with a Vic20 influenced color scheme sitting on my shelf if I'm not playing a game on it.

Finally a Handheld Retrogame with a Large Screen and Stereo Audio

2019-04-04T05:41:00.000-07:00

After a lot of trial and error, I finally have a 5 inch screen based handheld for playing retrogames. It's a Raspberry Pi 3b+ running Retropie, which is nothing new, but it is packaged in a handheld form similar to a large Super Nintendo controller or maybe a Playstation controller. With the assistance of a good friend, here is what we created.

It took quite a lot of effort to reach the point of a completed case to contain the Raspberry Pi, 5 inch LCD, buttons, speakers, battery charger, and LiPo battery. This is a photo of our first attempt:

At first, using little PCBs to put things exactly where they needed to be seemed like a good idea. He trouble is of course all of the wiring. The more wiring there is, the more likely there will be a mistake made when building the unit. In fact, the first time that I wired everything together, I had swapped the 5vdc and 3.3vdc lines - the Raspberry Pi survived; I had to buy a replacement LCD

Since I did not want to make the same wiring mistake again, I created PCBs to replace as much wiring as possible:

As a result, everything simply connects together - the audio amps, speakers, switches, all of it practically falls into place.

Retropie makes it easy to enjoy classics such as Nightstalker on the Intellivision:

Of course there are always the classic arcade games such as 1942 running on Mame:

Anyone that grew up in the 1980's and visited arcades would remember Outrun. Outrun is a true classic and it can be played using Mame or Cannonball. Playing Outrun using Cannonball program offers a collection of new features to Outrun such as new camera views, foreign game versions, cheats and more! Here is Outrun running in Cannonball on our gaming handheld:

There are other games that have been ported to the Raspberry Pi and will run on our handheld such as Wolfenstein 3d, Duke Nukem, Quake, and of course Doom:

Of course our gaming handheld can run numerous games through the magic of Retropie - the 5 inch screen makes veiwing much more comfortable than similar handhelds with tiny screens. Tiny screen game units look cute, and are a great novelty, but would you really want to play Playstation One games like Bugs & Taz Timebusters on such a small screen? Here it is on our gaming unit:

Naturally, since our handheld is powered by a Raspberry Pi 3b+ it is more than a gaming device. It is actually a fully capable Linux based computer. This means that not only can you play games from Atari, Intellivision, Nintendo, Commodore, Playstation, and more, but you can actually type in programs into a Commodore 64 and run them.

Just attach a wireless or Bluetooth keyboard and you now have a handheld retrocomputer! Try typing in a Commodore Vic20 program and enjoy computing from the golden age of computing. Maybe you prefer Atari 800 or have always wondered what a Sinclair computer was really like.

There is hours of entertainment available playing games, trying out computers that you never got a chance to try, and even writing your own native Raspberry Pi games in Python or the language of your choice.

I look forward to playing hours of cool retrogames and reliving my Vic20 days with the handheld gaming unit that my friend and I have created. We intend to make this available as a kit that you could easily build for yourself. Follow @JessLGames on Twitter to stay posted on our progress or just check this blog regularly.

End of the Saga of the 5 Inch Display

2018-12-20T13:31:00.002-08:00

For those that haven't read the prior two articles that I've written about cheap China sourced 5 inch displays here's a summary:

If you need to do any soldering on the display PCB do NOT buy the unit.

My first experience was with a Kumar display that I disassembled to figure out what had failed:

Disassembled Kumar LCD

My second experience was with a Waveshare LCD that did not display anything.

Nothing. No display at all no matter what was connected.

My third experience was with a replacement Waveshare LCD that displayed fine until a connector was soldered onto it and the solder flux cleaned.

Note the discoloration or smear, which is internal to the LCD: the light diffuser is damaged.

My fourth experience was with a replacement Waveshare device that had streaking issues upon shutdown which became worse after soldering on a connector - no flux cleaner was used on this LCD PCB.

Streaking just as shutting down and after shutdown.

The fourth display connected to a PC HDMI connection. Notice the odd coloured hue.

So, I was done with trying to find an inexpensive 5 inch display for my handheld Retropie.

I decided to try AliBaba and just source my own display directly from a manufacturer. Now I am but an individual working with my spouse and son to create games and projects so I am certainly no one that a large company would take seriously.

However I found a company that did take my concerns serious and responded to them. I had asked if there was any way to add a 26 pin connector to the LCD build spec so that users did not need to be troubled with soldering a connector onto a display and possibly experience the frustrations that I had experienced.

The conpany suggested a 40 pin connector instead and had a build ready for order in less than a week. This is a display made in China just like the others. However, this display works!

I ordered a sample express and this is what I received:

5 inch LCD with a factory 40 pin connector (2 views)

I finally had a 5 inch display that I could just plug into a Raspberry Pi and access all of the GPIO with no soldering required! Perfect!

I put this together into my Retropie prototype and this is what it looks like for now. I was a little off on the screen position so that will change in the next prototype:

My 5 inch handheld Retropie prototype with the new LCD.

If anyone is interested in obtaining a kit to make your own 5 inch display based handheld or would like a 5 inch display that includes a 40 pin connector for accessing Raspberry GPIO direct message me on Twitter at JessLGames - there's a minimum order quantity so I will need at least 30 interested purchasers before I can place an order.

Follow up to Do Not Solder on China Sourced 5 inch LCD Displays

2018-12-06T06:56:00.000-08:00

After the Kuman 5 inch HDMI LCD display failed, and I found the reason, I bought a 5 inch LCD from Waveshare.

Waveshare has a version 2 of the LCD display and states that this is in response to knockoffs of their version 1 design. I figured that going with a supplier of displays that was not a knockoff would be better than my experience with the Kuman display.

Waveshare customer support was excellent. The first display that I received from Waveshare failed to display any image at all from any source.

After asking for a swap through Amazon, I got my replacement Waveshare 5 inch LCD within a week.

I soldered the 26 pin connector to the Waveshare display, cleaned the flux with 91% isopropyl alcohol, let it dry and tried it. The Waveshare display worked great!

Two days later I noticed a cold solder joint on my 26 pin connector. This can happen when trying to solder quickly and keep LCD heat exposure limited. I cleaned the flux, let it dry, and to my dismay the LCD had a large faded area on it. I knew that it was the LCD diffuser that was distorted from my experience with the Kuman display but I didn't know why.

Now I knew, it was either the heat from the iron or fumes from the alcohol.

Waveshare provided a replacement with mininal hassle. The replacement took 18 days to arrive from China. I plugged in the new display into my Tivo and it worked fine. I did a burn-in for a few hours to test the display as a TV display just to verify reliability. No problems.

The second day I connected my Raspberry Pi 3b to the 5 inch Waveshare LCD. It looked good! However, when I shut it off, the text seemed to linger on-screen and it would slowly fade to a few vertical lines. I tolerated it.

Keep in mind that there has been no soldering done on the display yet and the vertical lines appear only when shutting down after displaying from a Raspberry Pi.

I soldered a 26 pin connector to the 5 inch Waveshare display PCB. NO ALCOHOL WAS USED THIS TIME. No cleaning. I hate leaving flux residue bhind but I didn't dare clean it. I was very careful to minimize any flux flow from the solder - if any spread from a trace, I wiped it immediately. No flux ever got near the dge of the PCB. If there were any problems this time, they had to be heat related.

I let the solder cool for a few minutes and tried the display. Retropie looked great! It looked great until I shut it down. Now, when shutting down, there were more vertical lines.

Waveshare had me try the display on a PC and all I got was a strange pink hue.

On a laptop, the 5 inch Waveshare displayed nothing at all. Oddly, on a Tivo everything seemed to work fine.

Waveshare have graciously offered to refund the purchase of the display, which I accepted. Of course that does not cover the costs of the 26 pin connectors, plus shipping, that I've attached to each 5 inch display.

So far, in an effort to use a cheap LCD from China as a handheld game cinsole display, my expenses are $54. It was about $30 for the Kuman display that I destroyed to find out what was wrong with it. It has been $24 in costs for buying connectors and the shipping to get them. Sure I could have bought more than one 26 pin surface mount connector at one purchase and saved on shipping, but what hobbyist needs more than one SMD 26 pin female header in their stock?

One possible solution would be to use a cold heat solder tool to attach the 26 pin SMD header to the LCD PCB but I don't own one and could not justify using it just for soldering on a header. Besides, I don't know how well the cold heat solder tools work on SMD.

The best solution would be that Waveshare and whoever supplies the cheap clone displays, include a 26 pin female header soldered to the PCB as part of the build.

However, including a 26 pin header on the PCB would not have helped with the first LCD, which did not display at all.

It would have helped with the second displayas it would not have developed display distortion - maybe. The third display arrived with some sort of issue but only when shutting down, but I suspect that the soldering may have made it worse.

BTW, Waveshare have asked that I give them 5 stars in my Amazon review. Would you give them 5 stars?

Do Not Solder on the PCB of China Sourced 5 inch LCD Displays

2018-11-13T02:15:00.001-08:00

I wanted a handheld game console for playing retrogames. I love Tempest, Ms. Pacman, 1942, and numerous other arcade games. However, I also want to play other games such as Atari 5200 games, Commodore 64 games, Intellivision, and more!

The Raspberry Pi 3B is near perfect for retrogaming as it runs numerous emulators except maybe Dreamcast or N64. N64 runs but very slowly.

I wanted something bigger than the Nintendo Gameboy equivalent also known as PIGRRL. I wanted a 5 inch screen with a DPad and joystick rather than just a Dpad.

I bought a Kuman LCD display from Amazon. The display attaches to the Raspberry Pi 3B through 26 pins although it only needs power and HDMI input. The HDMI is input to the display from a custom HDMI PCB jumper from the Pi to the display. This yields a pretty secure mount.

However since 26 pins are used by the display, the display routes those pins to a solder pad arrangement on the display PCB. Since soldering individual wires would be arduous, a female 2.54 mm header was soldered to the pads. 90% isopropyl alcohol was used to remove the flux residue. Oh yeah, I kept the display vertical when using the alcohol to prevent seeping into the display.

After waiting 10 minutes for the alcohol to evaporate completely, I powered on the Retropie handheld game console for the first time. Success!

But wait, there's a a blotch on the display. The display also seems dim. Kuman LCD display again. So I bought a Waveshare LCD display V2 as a replacement.

I assumed that I can't return the unit since Kuman unit since I soldered on the unit. Maybe I'm not supposed to use alcohol near the display or maybe I received it damaged. I was so angry that I decided to just buy a different brand LCD display.

When my new Waveshare LCD display arrived I knew to test first before any soldering. I plugged the unit into our Tivo. There was no display. Nothing. I returned it for a replacement.

The replacement Waveshare LCD display arrived. It did not autoscale so only a partial screen was visible. I plugged the display into my Raspberry Pi 3B and it looked fine. This display apparently relied on drivers to do scaling. At least it was nice and bright!

I soldered a female 2.54mm header onto the Waveshare LCD display. I cleaned the flux with 90% isoprpyl alcohol. After ten minutes, I powered up the Retropie handheld game console with my new screen.

The Waveshare LCD display was dim and had a blotchy shadow just like the Kuman LCD display.

Argh!!!

I decided to disassemble the broken Kuman LCD display LCD display to find out why there was a blotch. I had to know what was wrong.

After some careful manipulation, I got to the problem - behind the LCD at the backlighting layer is a clear piece of plastic with the exact blotch pattern that I was seeing. Apparently this is the light diffuser as the display is dark if power is applied without this plastic.

The actual texture of the plastic is different in spots which causes the shadows. I tried water. No luck. I tried Windex. I tried Dawn. No difference. Now I wanted to know what caused something so tough to be damaged.

I rubbed the plastic with 90% isopropyl alcohol and let it sit for two minutes. No changes occurred. I tried our hair dryer on the hottest setting at 1 inch away - no difference. A heat gun would of course do something but I was drying to simulate the heat that may be caused by a soldering iron. I finally gave in and held my soldering iron within an inch of the plastic and still no changes.

I even tried baking soda hoping that it was a residue that could be scrubbed. Maybe a fume residue from heated glue?

Finally, I realized that the material had been changed somehow. In desperation, I tried Bar Keepers Friend to try to scratch up the plastic evenly. It scratched the plastic evenly and yet the shadow impression in the plastic remained.

The now scratched plastic just made the display much dimmer. However, the shadow blotch still remains.

In summary, I have no idea what damaged the clear plastic diffuser. It may have been solder fumes. It may have been alcohol fumes. Maybe alcohol seeped in somehow. I hope Waveshare or their supplier figure it out because they are affordable, handy little displays.

Until the issue is corrected, if you're going to use a Waveshare LCD display or a Kuman LCD display. in a project, do NOT solder onto the display. Find another way to access the Raspberry Pi 3B pins or you may damage your display.

Don't Upgrade That Game Engine!

2017-10-31T21:29:00.001-07:00

I use Unity to develop the 2d platformer that I am working on. All other mobile games that I created were based on the libgdx engine.

Although libgdx is an outstanding application framework, it is not as easy to use as a game engine. It isn't supposed to be. An application framework makes the coding easier and makes it easier to access challenging functions, but it is not a visual tool. Using a visual tool makes things easier - supposedly.

I create games as a hobby. I'm not getting paid for anything other than a few dollars of revenue that helps pay for tech toys and software. However, I want to create games quickly so that I can move on to the next great idea.

A game engine helps make this a reality. To do what I have done so far in Unity to create a 2d platformer, would take much longer using libgdx. On the other hand, I would have more direct object control in libgdx and not as susceptible to bugs in the engine.

Libgdx is open source. A bug in the engine and it is addressed. Sure it may take a while for an update, but if I am not willing to wait, I can usually find the recommended repair myself and have a hybrid libgdx until the next version fixes the bug.

Unity is closed source so we will have none of that self repair thing going on. Who knows how long it may be for a repair to be applied? Also, if the needed repairs do not apply to a large enough group of protesting developers then it is not economically viable to repair the issue; thus, that small group of developers must find a workaround for dealing with the bug.

I am not saying that one solution is better than the other. Each has a different purpose and set of goals.

So I updated to Unity 2017 from Unity 5.6. Unity 2017 adds some new 2d tools, such as a tiling engine that I'd like to use. However, I forgot the old adage "Never upgrade software in the middle of a project."

Of course we must upgrade software for various reasons and if you're always working on a project then the adage seems ridiculous. You must upgrade sometime.

I upgraded to get access to the new Unity tools. However, I find that parenting no longer works quite the same as it did in all prior versions. Setting a player to a moving platform as his/her parent usually means that where the parent goes, the player goes. Not quite.

The player could move left to right on a platform in prior Unity versions. In Unity 2017, the player can no longer move back and forth within the parent platforms area - his position is fixed and immovable.

Next, I find that when the player rides a platform down, he bounces. This has never happened before.

So, I rewrite the code to avoid using parenting as a shortcut for moving a player on a platform and instead track and match velocities. This looked better anyways so in the end, success.

However, none of my animations would move. Sometimes, it is easier to animate the position of a character than to write code to move it - such as when leaping in an arc. In prior Unity versions, it was possible to move an animation position in an animation and also in code.

In Unity 2017, any reference to position in an animation seems to override any scripts attempt to set the objects position. Thus, it looks like an animation cannot be moved.

As anyone that has done any coding knows, it is bad practice to control the position of an object in more than one place in your code. At some point, there will likely be conflict. In this case, I should never have controlled position in an animation and in code.

So I fixed the problem by controlling object positions solely from code. This is much cleaner and easier to troubleshoot.

This may sound like a Unity rant. It really isn't. I am truly grateful that Unity is available to everyone and that you pay for it when you start making money or need more features.

This is a rant against inconsistency. The animation character movement issue should never have been possible. It should have always required either commitment to code or animation but not both.

Player object parentage to a moving platform is another story. I like the solution that I've been forced to create, but the player should be able to move left and right while child to a platform. Even if the problem was my code, a software engine upgrade should not change the objects behavior.

Am I unhappy with Unity? No. Frustrated? Not really. Irritated? Yes.

I've learned my lesson - do not upgrade Unity unless you are prepared to correct unexpected issues that may have changed simply for the sake of change.

Atari 2600 Pacman on my RetroPi Project

2017-10-22T07:04:00.003-07:00

Photo of my latest side project - a RaspberryPi running RetroPie enclosed in an old Atari 2600 Pacman cartridge.

I modified the cartridge to fit the RaspberryPi and printed a 2600 cartridge label using http://www.labelmaker2600.com. After laser color printing the label to some Avery label paper, I used packaging tape to laminate the labels - I got the idea from one of "the 8 bit guy" videos.

After cutting and attaching the labels, I turned on the RetroPie and tried out Atari 2600 Pacman - it is worse than I remembered.

I tried out Vice, which is the Commodore emulator, and ran the Vic20 emulator. I wasn't able to get any cartridges to run, but it was nice to see what felt like my old Vic20 of 37 years ago.

I went back to the Atari emulator and ran Asteroids. What a load of fun to trek down memory lane! RetroPie offers emulators for PlayStation, Amiga, and so much more. I will get to them eventually.

It's nice to be able to play my old Amiga games and PlayStation games again.

Managing Object Distribution in Game Levels

2017-10-04T21:13:00.001-07:00

Creating levels for a video game is a challenging task. Of course the first level is relatively easy as we just make a cool looking terrain and put a few platforms in the sky for the player to jump on. However, by the fifth or so level, it becomes very difficult to keep track of everything. This obstacle, and of course others, are what often discourage ambitious game developers from finishing their games.

To simplify this process, I use a spreadsheet on Google Drive. First of all, Google Drive is free and the spreadsheet autosaves. It also saves space on the hard drive since the spreadsheet is on the cloud.

By listing each level in the spreadsheet by number and using a column for every object that the player may interact with on a level, it becomes much easier to create an overall game rather than individual levels.

For instance, a game that has four available weapons to pickup may accidentally result in all of the weapons being clumped in the first or last six levels or so. However, if you can scan a list of X's in spreadsheet columns it becomes easy to see that weapons have been clumped together and as a result take action to spread the weapons out to other levels.

The most obvious benefit to using a level objects tracking spreadsheet is when there are many objects to manage. For instance, a game may have 10 enemies, 4 awards, 4 treasure types, 6 door keys, 3 life related pickups, 6 retrievable weapons, and gold to pickup. Without level object tracking, it is difficult to know difficulty of levels in relation to each other and object distribution.

This is one reason that some games feel easy and then are suddenly impossible to complete - the difficulty was increased but without a visual way to compare the difficulty with the prior level.

Also, stopping development and returning days later can be challenging and even obvious to game players when they notice the dramatic increases in difficulty.

Certainly, there are tools available to game creators for a wide variety of tasks. However, for the hobbyist creating games, a spreadsheet is free, and eliminates the need to learn yet another piece of software. Besides, a quick glance is all that is needed to get the overall image of object distribution and level difficulty.

My Least Favorite Computer

2017-01-29T12:50:00.001-08:00

When William Shattner started doing Commodore Vic20 commercials, I found that I just had to have a Vic20. It had nothing to do with Captain Kirk of course, but those comparison commercials wherein the Vic20 features were compared to other computers was very convincing to this 14 year old.

I saved every penny from my paper route and got $350 to buy the "Wonder computer of the 80's". I got the tape deck unit with the Vic20 and a few books and progs. Without the tape deck, it would have been little better than a gaming console.

Fast forward through the 80's and my intense love of playing and creating games continued. The Commodore 64 was purchased by my wife and I and we used it to create college reports. My first good paying job I got an used Amiga1000. It was a wonderfully frustrating machine as any changes to it were expensive. Besides, the IBM PC era for the home had finally arrived.

We sold the Amiga and bought an Amstrad PC-20. It was affordable, had EGA video and sound and two floppy drives - no hard drive. It did not run Windows but included GEM which is similar to the black and white Apple Macintosh OS.

Improving on the machine meant opening the flip top and plugging in ISA PC cards. There were only 2 slots. Oh yeah, and the lid did not close after a card was plugged in.

It was a strange PC and we didn't have it long before buying a full IBM PC compatible with memory, VGA graphics, Soundblaster, a CD reader and all. Overall, I've enjoyed every computer that I've ever owned - except the Amstrad PC-20.

Should You Buy a Former Police Car?

2016-09-06T10:49:00.001-07:00

Recently my 21 year old son started looking for a new used car to purchase. His old car had become unbearable.

Ever seeking a good value, my son decided to look at used police cars - in particular Ford Crown Vics. They are built like a tank and have more in common with a 1972 deuce and a quarter than any modern car. So, the Ford Crown Vic should be reliable and a bit fun to drive with the big V8. Police departments use it so it can't be that bad is the assumption.

After looking at various car ads online it became obvious that buying a used police car meant either buying from an individual or used car dealership, or from an auction. Buying from an auction can be a great deal as it is possible to get a 10 year old car for under $1000 with only 120,000 miles on it. We found that the car will probably need to be towed or have a jump, which of course means unidentified troubles.

However we noticed that almost no police cars from sources other than an auction went for less than $2000 and most went for $3500. Apparently people buy the police cars from auctions, repair them, and make a grand or two on the sale. Of course the trouble is that buying from anyone other than the auction means that we have no idea of the cars history and we must trust that the seller did a good job on the repairs and thus didn't just use duct tape and bailing wire to hold everything on the car together.

More importantly, we noticed something interesting in the auction listings for police cars. The listings mentioned the number of hours that the police car had idled. At first this seemed odd but then I remembered a scene in The Blues Brothers movie where a pair of cops are sitting in a car idling behind a billboard as the Blues Brothers screech through the turn directly in front of the cops. Naturally the cops slammed the accelerator to the floor to begin pursuit. You will need to rent the movie to find out the rest and it is worth watching. Just don't be tempted to watch The Blues Brothers 2000 as it is a terrible disappointment.

I've digressed, but the point is the abuse that police cars go through and the original Blues Brothers movie exaggerates that fact, but it also displays how cops abuse those poor cars in the course of doing their jobs.

Curious about the significance of vehicle idling hours we decided to do a little mathematics. Assuming that a vehicle has survived 100,000 miles, how many hours has that vehicle actually been operated? Let's assume that the average speed for those 100,000 miles is 35 miles an hour as most hours are on surface roads where speed limits average about 35 miles an hour. Of course the car will go over 70 miles an hour at times and less than 25 miles an hour in most parking lots and residential neighborhoods. So, 100,000 miles divided by 35 miles per hour results in 2857 hours. If you have a car with 100,000 miles on it then that car has operated for about 2857 hours.

To put that into perspective, consider what that means for a car that operates continuously. That is to say that if your car with 100,000 miles on it were to achieve those miles by operating 24 hours a day 7 days a week, your car would have operated for only 120 days. Just under 4 months!

Now consider that police cars are used continuously. Three shifts of eight hours per day or some similar variation. That police car is idling or giving chase for most of that time. So when we see the miles listed for a used police car as 120,000 miles, that means that car has the equivalent of 4 months or more of continuous driving time on it as explained above.

Next we look at the hours idling as we mentioned earlier. This is a very, very important number. In many of the listings that I've seen, the idle hours were listed around 2800 hours or within 500 hours of that number. If we compare those hours to the number of hours that it takes to acquire 100,000 miles then we see that it is nearly the same. In other words, that police vehicle with 2857 idle hours on it has the equivalent wear and tear on the engine of a car with 100,000 mile on it.

Ah but the transmission, suspension and everything else do not have those wear and tear hours on them you may say. Yet when you look at the mileage, the listings almost always list over 100,000 miles or more. The suspension, tranny and other parts have 2800 hours of usage on them or 100,000 miles of usage. So what? We just proved that the car has 100,000 miles on it.

Or did we?

We just showed that everything except the engine has 100,000 miles on it. However the engine also has those same 100,000 miles on it. Let us not forget, that the vehicle has idled for 2800 hours also. We must add those 2800 hours to the life of the police car to know the true age and true value of the vehicle. To make it simple, we view those idling 2800 hours as miles since the engine doesn't care if it is moving or not. As we established, those 2800 hours are about 100,000 miles equivalent so in other words: you would be buying a police vehicle with 200,000 miles on the engine and 100,000 miles on everything else but the engine. The engine is likely junk. Even though the engine was idling unloaded for all of those hours, there is still strain upon bearings and joints as well as the heat surfaces such as the pistons and valves.

The police vehicle is effectively a 100,000 mile car with a 200,000 mile engine in it. Even if the vehicle is purchased at $500 it is still not a good deal. A new engine is well over $3500. A 50,000 mile engine may be available for around $2000 but it may still have problems as the accessories stay with the car and only the engine is replaced - that air conditioning unit has run the equivalent of 200,000 miles and will need to be replaced independently of the engine. Same goes for the alternator, water pump, steering pump, timing chain, etc...

Of course if you intend to replace the engine and assorted parts as needed, you may still come out ahead if your total purchase plus repairs equals less than the blue book value of the car. On a Ford with 200,000 miles on the engine and related parts, I wouldn't count on it.

Flipt Memory Game - Updated Graphics

2016-03-10T21:17:00.001-08:00

Improve your Memory with Pattern Recognition

2016-03-05T22:19:00.002-08:00

As we age, we often find ourselves loosing confidence in our memories. This is usually brought on by event's such as loosing the car keys for the second time this week. As a result, we seek to improve our memories.

An article in The Atlantic speaks of using pattern recognition to improve memory. The article explains the importance of using pattern recognition and how important it is to us humans.

By using pattern recognition more often, we can improve our ability to remember. Not only that, but if we teach our youth to more effectively use pattern recognition then they will likely have an advantage over others as they age and hopefully have less difficulty with memory - they will have a technique at their fingertips to help them better utilize their memories.

The Flipt Memory Game encourages pattern recognition to achieve the best solve times. There is no requirement to use pattern recognition, but the fact that the each puzzle is constructed of abstract geometric shapes encourages the brain to seek familiar patterns. This helps with solving each puzzle level.

Similar to looking at clouds on a summer day, patterns emerge such as bunny rabbits and other objects. In the Flipt Memory Game, a Space Invaders character may emerge. Or maybe seargeants stripes, or even an airplane or bullseye.

Remember the emergent image and recreate it as fast as possible in the Flipt Memory Game. As levels become more difficult and memorization time decreases, you will find your brain more actively seeking patterns which will help you remember the puzzle. Your memory becomes more efficient and productive as it uses pattern recognition to help with memorization.

Download the Flipt Memory Game at https://play.google.com/store/apps/details?id=com.lewisj.flipt

What type of game is it?

2016-02-23T20:36:00.001-08:00

So you've finally finished your latest game. You've tested it and it plays well. It looks good. It has social interaction features and all of those other features that online articles say that must be in an Android game for it to be successful. So why are the downloads barely trickling?

Of course the answer to that question is best answered on a case by case basis. Did you just upload the game and hope that because it is so spectacular that others would find it or that it would be featured in the Google Play Store? Have you advertised at all? Have you let anyone know about the game through any channels other than the Google Play Store?

So let's say you've done all of those things that are mentioned in nearly every article about getting lots of downloads. What's missing? Well maybe it's the category of the game that's the problem.

Most games can be classified as more than one type such as arcade games can usually be classified as either action games also or racing games or even puzzle games. Just uploading a game to the Google Play Store and putting it in the most popular category can contribute to your game being nearly invisible.

My latest game, Flipt Memory Puzzle, is a puzzle game. Or is it a strategy game? Maybe a board game? It certainly isn't an arcade game. So which category is best for my Flipt Memory Puzzle?

If we look at strategy games we see that many of the games are tower defense related or even have an RPG quality about them. However if we look at the Top New Free section of strategy games we find that there are less than 160 games in the Top New Free category.

We could put Flipt Memory Puzzle in strategy and have high visibility right? Sure but the visibility is to the wrong audience so those that see Flipt Memory Puzzle ignore it and go download a tower defense game.

How about the puzzle category? The problem with the puzzle category is the same as the problem with the arcade category - there are thousands of puzzle games. We would be lucky to even make the list of Top New Free games in the puzzle category.

Lastly, what about the board game category? Well Flipt Memory Puzzle looks like it could be a board game - it looks a little like Boggle - but does that mean that board game is our category? Examining the list of games in the board game category shows games that could be considered similar to Flipt Memory Puzzle such as dice games and memory games.

The Top new Free section of board games has less than 125 games listed which will help make it easier to get on the list and at least rank at a visible level. So board game it is.

Flipt Memory Puzzle quickly appeared at number 43 on the list of Top New Free board games. Not bad right? Not quite. Were not talking about an extremely active category here such as arcade. Ranking as high as 43 in Top New Free games in the arcade category resulted in 100's of downloads for one of my prior games. On the other hand, ranking at 43 in the Top New Free category of board games has resulted in a trickling of downloads.

In summary, the category that a game is assigned is very important. It may even be as important as the name. Putting a game in a dead category will surely result in the game dying or require an awful lot of marketing efforts for success. Putting a game in the wrong category can be just as bad as it gets ignored and even if it is noticed, it is assumed to be a mistake.
On the other hand, putting a game in a category that is hot will require extra efforts to get noticed but if it is the correct category for the game it may help increase downloads. Even something as simple as the category that your game is uploaded to can make or break it.