Posted on

Cheap DIY Flashing LED Wooden Signs

This idea came from a few different places. I saw a wooden sign with LEDs on it at a craft sale, and thought it looked amazing, and simple to make. A few weeks later, I found Julian Ilett’s videos on ring oscillators. Putting the two together seemed to make sense, so I did. Over the next few days, the first sign started to come together.

I know its not quite the time of year for signs like this, but it sure does feel like winter again – in April 2018 – so I’ll post this now.

You can see a video of it in action here.

Assembling the Wood

The wood needed for these signs is pretty simple. A few pieces of wood, all the same length, get lined up beside each other. Then get 2 more pieces of wood, the same length as the width of the other boards, and hot glue the widthwise across the back to hold everything together. Then put 2 screws into each of the longer boards, 1 through each of the shorter boards along the back. If you flip the boards over, you should have a flat surface of multiple boards lined together. Sanding it is possible, but not necessary. It would make it look better, but unless you have a belt sander, I would not bother with it. As for the size of the sign, the Ho Ho Ho one is probably around 50 x 30cm, while the Joy To the World one is roughly 30 x 20cm.

Painting your Design

Regular acrylic paint from a dollar store is the easiest way to paint the wood. A clear acrylic spray sealant will go over it once finished, so that the paint will not chip either. I start with a solid color for the background, then add all the details. All the ones in the pictures, I painted by hand, and they didn’t turn out to bad – I definitely would not consider myself an artist.

Drill holes for LEDs and gluing them in

Next up is drilling the holes for the LEDs. I used 5mm white LEDs, so I used a 5mm drill bit to make holes for them. One of the special things about a ring oscillator (the circuit we will be using to flash them), we need an odd number of LEDs, so make sure to drill an odd number of holes. Also, make sure to avoid the braces on the back or else you’ll have a hard time soldering wires to the legs of the LEDs. Once the holes were drilled, I countersunk them from the front to be able to see that LEDs a little better, then repainted the holes. The LEDs went in from the back and were glued in place with some hot glue.

Soldering the Circuits on the LEDs

For every LED on the sign, a 2N3904 transistor, a 5-10uF capacitor, a 1K resistor (for current limiting the LEDs), and a 100K resistor (for passing the signal from one LED to the other) is also needed. An extra 100K resistor is also needed to make the LEDs flash a little more randomly. This extra resistor will go from one OUT from one LED circuit, to an IN connection on another LED an odd number of steps away. Connect the circuit as shown in the pictures below.

Powering Your Sign

First off, the LEDs need roughly  2.5V to shine, so we must supply it with greater than 2.5V. The other consideration is the higher the voltage, the faster the LEDs will flash, but you cannot have a voltage that will give too much of a current through the LEDs. With a 1K current limiting resistors, don’t put more than 9V across the circuit. This can be powered with a USB power bank, but I think 3AA batteries is the best. Just be sure to add a switch in the circuit as well. For the main power wires, I run 2 wires all along the outside of the circuit, and strip them where they need to be connected. Insulated wire is harder to work with here, but is better because it will prevent shorts – sometimes you do have to cross the wires.

Final Touches

I finished them up by covering them with a  layer of clear acrylic spray paint so that the paint would not chip or come off.

Posted on

How I Process and Test my 18650 Cells

Old laptop batteries are a great source of Li-ion batteries, as long as you know how to properly test them to make sure they are safe to use.

In a typical laptop battery, there are 6pcs of 18650 lithium-ion cells. An 18650 cell is just a cylindrical cell with a diameter of 18mm and height of 65mm (approximately). If the laptop battery no longer works, there is usually just 1 group of cells that died, and the other 4 are still perfect, but you have to reliably test them all to make sure they work.

All my cells are tested with my 18650 testing station shown here.

Removing the Cells

To get the cells out of the laptop battery, all you need to do is break off the plastic casing. There are various methods that work here. Make sure to wear gloves and safety glasses – parts of the plastic casing can fly off, and are pretty sharp. The nickel tabs connecting the cells together are very sharp and can cut you very easily, as I have found out too many times.

1 – If you can twist the plastic casing and break it apart, then that is the best way to do it. This does not work on all batteries, and I am usually only able to do it on 3 cell Dell packs.

2 – Get a durable pair of wire cutters and/or pliers and try to break the corners off, or split it at the seam.

3 – Hitting the pack against the ground is a pretty good way to get the cells out. You might damage some of the cells, but this is one of the quickest methods of removing the cells.

Once the cells are freed from the plastic casing, you can get to work separating them into individual cells. They are usually spot-welded together in a 3S2P configuration (for a 6 cell pack). Cut all the wires going to the PCB one at a time to avoid shorts. The best way to get the spot-welded nickel tabs off the cells is to twist them off. Grab it with a pair of pliers or flush cutters,  and sort of roll it up. Be careful not to make any short circuits with the metal tools – the entire casing of the battery is the negative terminal, so if the heatshrink around it is broken, it can be easier to create a short circuit.

Initial Voltage Check

The first thing I do when all the cells are freed, is to do a quick voltage test. If the cells are over 2V, then they can go straight to charging in TP4056 chargers, or Liitokala Lii-500 testers. If the cells are under 2V, I mark them with a ‘V’, then charge them up with TP4056 chargers.

Self Discharge Test

Once the cells are fully charged, I let them sit for 24h, then measure the voltage again. If any cells discharge themselves just by sitting there, they will be weeded out here. Some people would recommend a week, others up to a month before testing them again, but for me, 24h is a fairly good amount of time. If any of the cells are under 4V at this point, then they are considered self-discharging, and are discarded.

Capacity Test

Any cells that passed the first two tests are now tested for capacity in Liitokala Lii-500 testers. OPUS BTC3100 are another common tester, but are more expensive than the Liitokala Lii-500, with the same functionality. They are charged, then discharged while measuring the capacity, and finally charged back up again. I write the capacity on the cells, and then sort them based on capacity. Under 1000mAh are discarded, and the rest are separated into 1000-1600mAh, 1600-1800mAh, 1800mAh-2000mAh, 2000-2200mAh, and 2200mAh+. I would recommend only using cells over 1800mAh in final projects, and using discarded cells as practice for soldering.

Sometimes IR Test

The last thing to determine the health of a cell is the Internal resistance. The Liitokala Lii-500 tests the internal resistance of a battery each time you put it in, but I sometimes do another test with my homemade Arduino IR Tester. This test is not really that important if you are using cells in low power applications (<1A per cell), but in higher power applications (1A+ draw per cell) it is more important. The higher the internal resistance of your cells, the more they will heat up as you charge or discharge them. The extreme cases can be caught just by monitoring temperature during the charge and discharge processes.


Throughout all these tests (particularly charging and discharging), I monitor the temperature of the cells. If any cells get over 40 degrees Celsius, they are marked as with an ‘H’, as heaters, and are brought back to the computer recyclers. Red Sanyo cells have a high tendency to heat up.

I have recovered over 2000 cells following these guidelines, and have been fairly successful in determining which ones are good. One word of caution though – Any cells that do no come from a reputable manufacturer – Samsung, LG, Panasonic, Sanyo – are more likely to fail even if they test good. Of all the cells I have used, only a handful of knockoff Chinese brands – SZN, CJ – have failed.

Have a look at my battery testing station here.

This method is by no means the best, most complete and accurate way of testing 18650 Li-ion cells, but it is just my take on it.

If you want to see more resources or other similar ways of testing cells, check out these links:



Posted on

Eco-Charge Portable USB Power Banks

One of the projects that I have been working on for the past year or so are portable USB power banks made out of recycled laptop cells. These power banks are ideal for hobbyists and for powering Arduino projects because they are controlled on and off by a switch. Most commercial power banks have an auto-on function, that turns on when a USB cable is plugged in, and  stays on if is is drawing more than a few hundred mA of current. For most Arduino projects, this does not work because the power bank will keep turning off after 10-20 seconds. With this power bank, if the switch is on, then there will always be 5V on the output (unless the batteries are depleted). I use these all the time on my workbench to power and test small projects. They are very simple to make, and the total cost is under $3 in components (except testing equipment).

Junior Achievement

At my school this year, I participated in the Junior Achievement (JA) club – where a bunch of students get together, choose a product, and start a business. I pitched these power banks as the product for the year, and it was enthusiastically accepted. We manufactured and sold over 100 power banks, and re-used over 50kg of lithium-ion batteries.

I was the VP of production, and was responsible for safely designing and creating over 100 power banks. We already had a prototype to prove that the product would work, but it needed lots of refining in the components, design, and instructions. We produced a power bank made from recycled laptop cells, and open sourced the design, schematics, and instructions so that anyone in the world can reproduce our design and save more laptop batteries from landfills.

Cost < $3

One of the main things that kept the production costs down was that we could get the batteries for free from computer shops and other recyclers, so they cost nothing. However, using recycled batteries did pose some potential issues with safety that we had to research and test first. Each battery required over 40 hours of testing before it could be used in our power banks. We explored buying new batteries, which would alleviate the testing time, but they would cost $10 per unit almost tripling our production costs, reducing our profit to under $2 per unit. The first few weeks of production were spent emailing and phoning local computer repair shops, computer stores, and electronics recyclers about getting their batteries for free. We contacted over 50 companies, but only heard back from about 10, and only 5 could give us their batteries. Getting batteries for free was proving to be harder than we thought. One of the electronics recyclers (EDI / Foxy Recycle) that gave us some free batteries had thousands available that they would also sell. Given that this seemed to be one of the only places in Ottawa that had the amounts of batteries that we were looking for, we decided to buy them from there, at a price of $0.30 per unit (30 times cheaper than new cells), and they had all the batteries that we would ever need.

Approval and Testing

Getting this product approved by JA took longer than originally expected. Most of the JA executives were cautious of the idea of a group of high school students making portable phone chargers out of recycled laptop batteries, which are classified as a dangerous good. From my extensive experience working with electronics, I knew that this product was safe and needed to communicate that to the authorities. We had to compile a lot of research on the specifications, charge and discharge rates, and safety features of all the components that we are using, especially the batteries and the protection circuit. Personally, I poured over 25 hours into making a document that outlines all of the safety features, and precautions that we are adding to ensure that nothing goes wrong. The potential issues with the production process, as well as the finished product, were all examined to make sure that there would be no problems in any aspect of making these power banks. In the end, we made sure that our product had overcharge protection, over discharge protection, short circuit protection, and overcurrent protection, which are standard on all commercial power banks. We also went the extra mile and fused the cells, which is not commonly done by manufacturers, to provide an extra layer of protection. Each cell needs to be thoroughly tested for IR, SD, capacity, and heating. This testing takes over 35 hours (3h charge, 24h sit, 8h capacity) to complete per battery. Our off-site testing station can handle up to 76 batteries at once. All this testing was done with my 76 cell 18650 Li-ion battery tester.

Design and Case Manufacturing

The design, made in Fusion 360 – a 3D design program widely used in the industry – went through more than 10 prototypes before getting the final version. These each required more than 3 hours to print, and a few weekends were spent refining the design. Finding a way to manufacture the cases was another challenge that we faced. 3D Printing them during the JA hours was not possible, as we would only be able to print 1 per week during each 3 hour period. We started looking into resin casting the cases, and found that there were high up-front costs. There is also a very steep learning curve and it would have taken $30 and a week of time per prototype for testing. With our 10 prototypes, pre-production costs would have been over $300. Resin is also very bad for the environment compared to 3D printed PLA plastic – made from corn starch – so it didn’t feel right labelling our product Eco-friendly if the casing was harmful to the environment. We started looking at other methods of small scale manufacturing possible on a tight timeline. With 3D printing, the prototyping costs are the exact same as the costs for the finished product, and it takes only 2 hours and $2 make a new prototype instead of a week and $30 per prototype. The common 3D Printing manufacturing services, such as Shapeways, were way out of budget for us, so we started looking at local options. The UOttawa makerspace had lots of printers available for free to the public, but to make something to sell, they would charge almost $5 per case. After some more research, we found a local company,, who was willing to print the cases at $3 each, and could make 50 in a few days, exactly what we needed. For the first set of cases, we ordered them from there, but subsequent ones, we were to get them printed with A2D Electronics, and still keep up with the rest of the production chain.


One of the other issues we faced was figuring out if we needed to get the product certified, and what labelling was required. Because it is all low voltage DC, we were not required to get it certified, but still needed proper labelling. We explored adding the labelling to the 3D Printed design, melting it in after it was printed, and settled on printing a label out on a color laser printer. We used mod podge and acrylic sealer to attach it to the bottom so that it will not fall off or get damaged with water.

Strict quality control testing was also necessary. We used a USB tester, which puts a 1A load on the USB ports to simulate charging a phone to ensure that every power bank was able to supply the rated current, and did not have significant voltage droop. Testing the recycled batteries followed a strict procedure as well, testing for internal resistance, self discharge, and capacity, the 3 indicators of a cell’s health.


The design, schematics, 3D printing files, and instructions are all available below. This means that anyone in the world, can replicate our power banks and save lithium-ion batteries from landfills. My design is now being modified by UOttawa graduates to build a solar powered reading light. We re-used over 50kg of batteries, and created a product and testing methodology that is reproducible worldwide.

Stay tuned for more posts regarding my these portable power banks. I plan on making several improvements to these in the future, including complete instructions, and possibly a custom PCB, and larger capacity.


Full instructions and an overview of the components will be coming soon.

All the STLs required for 3D printing the cases can be downloaded here.

If you make one, I would love to know and get some pictures of it!


Posted on

Custom Liquid Cooled Computer

This computer was a long time in the making, and it is a bit old now. I finished it in Feb 2016, after almost 2 years of planning, waiting for good prices on parts, and building it. The only things I paid full price for in this build were a few of the radiators. Several iterations were needed so that I had a computer to use while I was waiting on more parts.

I did not pay full price for many items in this build, and found many parts on kijiji, eBay, and from overseas much cheaper than their list prices. Getting these prices on some parts is not something that you will find everyday. I spent over a year just gathering parts for this build. If you take your time and do your homework, you should be able to find some amazing prices on quality components.


I initially got a 6 core Intel i7 5820K from microcenter. One of my friends was driving through the US and was able to pick this up for me. Even with the exchange rate, it was still a lot cheaper than buying one in Canada at the time. I used this the first iterations of the build. While browsing kijiji a while later, I saw a 12 core Xeon E5-2658 V3 for sale at $300. I could only find information for this on the intel website, and a few Chinese eBay sellers selling it, so it is probably an Intel engineering sample. It works great, and I love seeing how fast Cinebench runs! The only drawback to this beast of a CPU is its low clock speed, giving lower single thread performance, though for $300, it can’t be beat. I even sold the 5820K and got around $350, so I basically got $50 to get a better CPU!


I found an AsRock X99 WS board on eBay for a good price, and jumped on it. This basically chose my color scheme for me as it was blue, and did not fit with many other components. The first iteration had a black and blue color scheme, but it did not turn out as nice as I had hoped. My next board was an Asus X99 Deluxe that I found for $250 on Kijiji. The black and white color scheme would allow me to go for any color if I added some RGB LEDs, so that’s what I did. I designed a custom cover that goes around the waterblock to hide some of the silver capacitors around the CPU socket.


I started out buying 8GB of Crucial DDR4 when it first came out, and paid around $130 on Newegg for it. Later on when I upgraded my CPU, I also bought another 32gb of RAM in a 2x8GB and a 4x4GB lot for a total of around $120. Trying to get all my sticks recognized was not going to be possible, as they were all different capacities, so I installed the 4x4GB sticks for 16GB, and used the other RAM in other computers. I salvaged some black heatsinks from an old set of Corsair DDR  RAM, and attached them to these sticks that do not have heatsinks.


I started out with a single 500GB Seagate drive from my old computer, but found a listing on kijiji for some 2TB WD4 drives for $50 each. I bought 2 and put them in RAID 1. I designed and 3D Printed some custom brackets to mount behind the motherboard tray in the extremely spacious NZXT Phantom 630 that I have. Temperatures have not been a problem. One of the original drives failed after a year and a bit, so I bought another 2 2TB drives on eBay for around $100 again. Repairing the RAID array was not a problem, just a simple case of firing up Intel Rapid Storage Technology.


I have 2 Avexir 256GB Sata SSDs that I got for around $75 each from NCIX back when it was still around. I designed another custom 3D Printed bracket  for them and put them in RAID 0 for extra speed. They have served me well, and are nowhere near full yet.


The 1300W EVGA G2 power supply that I won on eBay for $125 was definitely a steal. It had previously been used for bitcoin mining, but a beast like this should be able to handle it without a problem. This was one of the best and most powerful power supplies out there, and had a great review by johnnyguru.

Graphics Cards

The 2 GTX780s that I have in this system are getting a bit outdated now, but remember, I built this system in 2016. I got one of them at CanadaComputers on clearance sale, and the other one on Kijiji. The backplates on them are custom-cut on an Epilog laser cutter out of 2mm acrylic. I designed them in inkscape – an amazing and free 2D Graphics software. The letters are vector cut, and the NVidia logo was raster engraved.

Liquid Cooling

The liquid cooling in this was a fairly long process. I ordered components from Aliexpress, and found a lot of EK fittings for an amazing price on Kijiji. This bag of fittings was ultimately what influenced me to go with hardline tubing in this build. The GPU waterblocks are universal blocks that I ordered from Aliexpress, and have served me great. I don’t game at all, so they are not under heavy loads much, if at all, so these are perfectly fine (and a whole lot cheaper than full cover blocks). I added some small aluminum heatsinks on the VRAM chips, and an 80mm fan blowing on the power delivery section of the card. Through a bit of trial and error, I was able to power them from the original fan headers on the boards.

As for radiators I have a 360mm, a 280mm, a 240mm, and a 120mm radiator. I believe all the fan slots on my NZXT Phantom 630 case are taken up with radiators. This is way more than needed to cool all the components in this PC, even if they were overclocked. I went for this overkill setup because I thought I would be upgrading and overclocking within a year, but clearly that did not happen.

The pumps in this build are 3 12V 18W pumps from Aliexpress, controlled with PWM from an Arduino with MOSFETs. This board also controls the LEDs in the computer. All the pumps are mounted on some custom 3D Printed brackets sitting on top of the lower 240mm radiator and fans.

This was not my first liquid cooling adventure, and it certainly won’t be the last.

Custom Cables

Pretty much all the cables in this build are custom made and sleeved. It was quite interesting to find all the required connectors for this build. I had to buy some adapters on eBay just to get the connectors – SATA, and 6/8pin graphics card connectors specifically. To get an 8 pin ATX power connector for the motherboard, I had to buy a lot of 50, but it was only about $6. If anyone needs an 8-pin ATX plastic housing, let me know. I also had to find pinout diagrams for my power supply.

The first time I sleeved power supply cables, I went for paracord sleeving. This time, I went for the more standard plastic sleeving, and I’m glad I did. It was so much easier to work with. I used black and white sleeving that went over black and white 16AWG wire from eBay. Crimping the ATX connectors on took a bit of practice, but I eventually got the hang of it. One of the quirks of the power EVGA 1300 G2 power supply that I have is that it has capacitor on the cables. Finding the right wires to put them on was a pain, especially given the limited documentation (at least at the time). Building these custom cables took over a week’s worth of time. I also laser cut some cable combs out of black acrylic to keep everything looking nice on the front side of the case.

Case Mods

As you might expect, I did have to cut a bit out of the case as well. The original DVD drive bays blocked part of the window, and stopped me from mounting the 280mm higher up than it was supposed to go in the front of the case (to allow room for the power supply and 240mm rad in the bottom). With the help of a rotary tool, those came out, and I repainted the edges with some simple black acrylic paint. I can hardly notice that anything was cut until I look pretty closely.

Custom LED Fan Rings

On 3 of the fans in this build, I made my own LED rings. I popped the fan blades out and drilled holes around the edges to match up with the LEDs in my RGB LED strips. They turned out really well.

LED and Pump Control

I made a custom Arduino control board for my pumps and LEDs that is able to control pump speed and will cycle through the RGB color spectrum fairly slowly.


As I mentioned at the start, this is now over 2 years old and needs a refresh. I still have not changed the coolant, and have noticed that it has lost its milky white color – it is now more of a light gray. I have absolutely no idea how much time I poured into this computer, but I have to say, the end result is simply amazing.

Posted on

Half An Hour’s Worth of Burning Arduino Bootloaders

Today I spent about half an hour burning bootloaders on AtMega328P ICs for DIY Arduino Kits.

It was fairly repetitive, but without Nick Gammon’s excellent bootloader burning program, it would have been lot more complicated. I downloaded the code and instead of having it ask for serial input, just gave it all the values required.

I used 2 of my DIY Arduinos connected with 6 jumper wires. For each AtMega328P, I first bent the pins inwards by pushing it against a table so that it fit in the IC socket. Then I put it in the IC socket, press the reset button on the other Arduino, and wait for the LEDs to stop flashing.

In about half an hour (while playing Chinese Checkers), I was able to burn the bootloader on around 80 AtMega328 ICs.

Posted on

DIY CC CV Variable Bench Power Supply 1-32V, 0-5A

I have gone without a variable lab bench power supply for too long now. The PC power supply that I have been using to power most of my projects has been shorted out too many times - I have actually killed 2 by accident - and needs a replacement, at least for low power loads. There are now extremely cheap 5A CC Buck converters available that are perfect for something like this. I also added a Voltage and current display, a switch, and replaced the onboard 10K trim pots with regular potentiometers. I also desoldered one LED that lights up when the output is shorted (indicates constant current mode), and added some wire extensions and a 3mm LED to mount to the case.

18650 batteries are lying around all over my workshop, and I needed something to do with them. I found a design for a 4S10P holder on thingiverse that I printed out and put cells in it and soldered them up with 2A fuses to give me 8S4P. The rest of the space in the holder is used for the CC CV buck converter and other electronics. This allows the highest voltage possible for the buck converter, so we get the biggest voltage range on the output. The maximum voltage will decrease and the 18650 cells are drained, but I don't anticipate needing 33V DC very often.

The display is powered with 12V through a 7812 12V voltage regulator, that can handle up to 35V max input. Finishing this up, I added an XT-60 connector and a balance connector to the main battery so that I can charge it up. I also added some cardboard on the top and bottom to protect the fuses and avoid shorts. To finish it up, I printed out my logo on a used label sticker page and transferred it to the top of the battery.

I have used this fairly often, mostly to simulate 18650 batteries. I would love to find a way to get coarse and fine adjustment on voltage and current levels, so that it is much more usable. Right now, it is fairly difficult to get an accurate voltage without the tiniest of turns on the potentiometer. I might make a similar one using the same parts, but instead of attaching it directly to a battery, use an XT-60 connector and then it can be used with any battery I want. That will need a boost converter as well to get higher voltages, but that it easily fixed.

Posted on

Fake TP4056 Charge Curve Tester with INA219

Why its needed

I’ve been using TP4056 modules for a while now, and have just recently found out that there are tons of fake modules out there now. It’s actually really difficult to find genuine TP4056 chips. This blog has a great outline for identifying some of the chips and the potential problems with them. I wanted a cheap and effective way to test my TP4056 modules to make sure that I am not damaging any 18650 cells.

Dummy 18650

To interrupt the current path in the 18650 charging circuit, we need to either slot 2 pieces of wire and a separation material in the positive end of the 18650 holder, or make a dummy 18650 cell, and put another 18650 holder on top of everything. I designed an 18650 cell in fusion 360 (it was very simple) and added a loop on the top of it to easily get it in and out of any testing station or TP4056 modules. You can find the file for it on here (coming soon).

Other parts and connections

The only parts needed for this project are an INA219 current sensor, a micro SD card holder, and of course, an Arduino nano. On each end of the dummy 18650, insert a nickel strip (used for spot welding) or a piece of solar busbar. Connect the all together, using SPI for the micro SD Card holder and I2C for the INA219 module. One ground wire from the Aduino must be connected to the negative side of the 18650 cell to allow the INA219 to measure the voltage as well. The CS (Chip Select) pin of the micro SD card reader can be connected to any Arduino Pin, but most examples use pin 4, so I will stick with that to avoid modifying code.


To get the current flowing into the 18650 cells, and the voltage of the 18650 cells, we need the load voltage and the current from the INA219 module. Adafruit’s library is very easy to use, and works well. As for logging data to the SD card, we can use the built-in SD library, use a string to hold each line of data, separating each value (load voltage, current, bus voltage) by a comma so that it is easy to import into excel and create graphs. Code can be downloaded here.

Charging graphs

So far, I have not found any of the TP4056 modules that I have to be problematic, but I will keep testing them.

Posted on

Fixing a Lithium-ion Yardworks Lawnmower Battery

I found a 40V 10Ah lithium-ion lawnmower on kijiji a few years ago for $30. Finding lithium cells that cheap was a steal, so I bought it. Originally, I hoped there would be around 50 18650 cells inside, but when I got it home and opened it up, I saw that there were 10 10Ah cells in series. Not as easy to work with as 18650s, but I’ll probably add them to a powerwall or something soon – if I can ever can them out of the case that is. They seem to be wedged inside with double sided foam tape. I might need to break the case to get them out. Anyways, I found the capacity indicator on the battery was showing nothing, so I measured the voltage of each cell individually. As expected, I found one cell sitting around 1.6V, clearly problematic. The rest were sitting around 3.7V and 3.8V, which is a good sign for those cells.


I had a few TP4056 modules out for charging 18650 cells, so I just hooked one of those up to the problematic cell with some alligator clips. I did bypass the BMS by doing this, so I kept a voltmeter connected to the cell as well. It took a few hours to charge up to the voltage of the other cells, and when I disconnected the charger and re-tested using the built-in capacity indicator, it now showed that the battery was half-full. The battery has been restored to a fully working condition, but given that I don’t have a 42V DC wall adapter, or any high voltage boost converters to charge this as-is, I will probably be taking it apart to use the cells individually. The BMS will probably get used for an e-bike or another similar project.

Posted on

Ice Cream Maker Attempt

This project has been sitting around completed for a few months now, and I finally found the time and opportunity to test it.

I used a few peltier elements to cool down a stainless steel tray, in order to try to make some ice cream on it.



4 12V peltier elements (though more than 4 is advised)

12V computer power supply – at least 100W on 12V rail per peltier element is advised.

Small metal tray

2 CPU heatsinks – any heat sinks are usable, but these are cheap and easily accessible in old computers. I had both already on hand when I started this project.

Attaching Peltier Elements and Heat sinks to the tray

I used some cheap thermally conductive silicon glue to attach both the peltier elements and the heat sinks. I just left them sitting with the weight of the aluminum heat sinks to hold them down, but it definitely would have been a good idea to clamp them down while the glue was drying so that there is less space between the peltier elements and the tray.

To permanently hold everything in place, I contemplated designing and 3D printing a holder, but I was not too sure about the effects that a big temperature difference would have on it (brittle at one end, malleable at the other). So, I drilled some holes in the tray, attached m3 screws and washers, then looped some solid core wire around them and over the heat sinks to hold them to the tray. A 3D printed mount would have been better, because it would allow for more pressure to be applied, but this was much quicker.

Making the Stand

I felt that designing a 3D Printed mount for this would take longer than it was worth, so I just used some wooden dowels and hot glued them to the fan shrouds on the heat sinks. It ended up a little uneven – nothing that can’t be solved by adding a bit of extra hot glue to a few corners – but when you’re not accurately measuring anything, that can be expected.


All 4 peltier elements, as well as both fans were connected in parallel with a wago style connector. To provide 12V power, I used an old 650W Antec computer power supply. The 4 peltier elements draw almost 25A at 12V (300W), so they are not pushing this power supply to its limits. Computer power supplies are really the only solution I have found to power this, as it uses so much power.


As you might have been able to tell from the title, this did not work as well as I had expected it to. It only really got cold in one spot, just above one of the peltier elements, the rest of the ice cream mixture hardly changed from its original liquid state at all. It took almost 20 minutes to make one spoonful of ice cream. Using more peltier elements is definitely a must if you plan on building one. That way, it should get much colder and might actually be useful. If, or when, I do this again, I would also use some more substantial heat sinks, and mount them properly with screws to hold them down tightly.

Posted on

Custom Spool Holder

For the last 2 years or so since getting and building my 3D Printer (a Prusa i3 style clone from China), I have always used the spool holder that came with it. That was not ideal, as it took up space on my shelf to the side of the printer. It also took a lot longer than should be necessary to change filament, and it only easily took regular sized spools – some of the ones I have are wider than usual. I looked for designs of spool holders on thingiverse, but none of the ones I could find matched what I wanted, so I designed my own.

I wanted it to take up as little room as possible, and it could not be mounted to the center of the top acrylic piece like most other designs do because it would be too tall, and would interfere with the hockey equipment above it. I opted to attach it to the top left of the printer as shown.

This spool holder is entirely 3D Printed – no metal rods or bearings necessary. I still have to make a few changes to the design to minimize print time and add a bit more strength. The STL files will be uploaded to thingiverse when complete.