Posted on

50W Solar Panel As Cheap As Possible

A few months ago, I found a 100W solar panel kit on kijiji for $50. I had previously built a smaller solar panel to see if tabbing cells myself was actually viable. Quite a few cells were broken in the process, but I got the hang of it and managed to build a small panel, so I decided to try a bigger one.

The bigger cells were thicker and easier to work with, but it still took a few broken cells to get used to them. A few broken cells means that I could not build a 100W panel, but after building a 50W panel I’m glad I didn’t have the option of a bigger one – it would have been too big.
I made this panel as cheaply as possible, and used some fairly unconventional and not necessarily recommended methods of framing and protecting the cells. Let’s go over the frame first. As you have probably guessed from the pictures, the outer frame is not made from the industry-standard aluminum, but 4 pieces of 1×3 that I got for under $5 at the hardware store. For the backing of the panel, all that’s needed is a solid material that is preferably as light as possible. I went with some rigid foam insulation, as it fit the requirements fairly well. Usually, EVA sheets are used for this, but that certainly did not fit the ‘cheap’ requirement. I used only about $4 of rigid foam insulation.

To cover and protect the cells, a piece of glass, acrylic, or epoxy resin are commonly used but again far from the ‘cheap’ requirement. I found a few posts on solar power forums using clear spray paint to protect the cells. This is certainly less durable than the traditional methods, but works for my purposes. I would not leave this out in the rain or in heavy wind, but on those days, there is usually not much sun anyways. The other thing I had to consider was the amount of light that gets through to the cells, which I tested by spraying 1 of my small cells with a generous layer of spray paint. I measured the short circuit current of that cell and an un-coated cell, and found about a 10% reduction in output current. The open circuit voltage did not change. I believe the commercial methods get about 2% drop in power.

Now to go through tabbing the cells. This is definitely the most tedious process of building this panel. As mentioned earlier, this is a skill completely different than soldering 2 wires together. The general process is to put flux on the cell, then lay the tabbing wire across the pads, and drag your soldering iron along the whole length of the cell. The tabbing wires should be just shorter than the length of 2 cells, so that when you connect them in series, there is no risk of shorting a cell and losing power. Once 36 cells are tabbed along the front, I laid them out in 4 strips of 9 cells that I will wire all in series later to get an 18V open-circuit panel. Each cell produces roughly 0.5V, 36 in series gives 18V. This is the recommended configuration for charging a 12V battery bank, as when a load is placed on the panel, the voltage will drop.

After finishing tabbing all the strings, I laid them down on top of each other (big mistake) and did not have time to come back to them for a few weeks. When I finally got back to them, some of the had stuck together where the solder joints were, and I broke 2 or 3 cells trying to free them. After laying the 4 strips out on the foam backboard, I passed the main positive and negative wires through to the back side, wired the rest of the cells up in series with some bus wire, and added a 20A anti-backfeed diode to prevent the batteries draining in to the panels.
When spraying on the clear spray paint, one of the brands I tried started reacting with the rigid foam insulation and melted the top layer of it. If you were wondering why the back and front of the insulation look so different, that’s why.

Testing it revealed an open circuit voltage of roughly 17V, and a short circuit current of 3.5A. Total cost of the project was around $60. For a 50W panel, I’d say that’s not too bad.
This panel will be used to charge my DIY Powerwall once I get a boost converter figured out, as the Constant Current on the 1500W one I bought does not work. With a solar panel this is not a problem, but charging from mains I do not want to draw too much current from any power supply that I use.
If I were to make my own solar panel again, I would not put a strict price constraint of the cheapest possible panel on myself, and would go for an epoxy resin cover instead of the spray paint. This would be much more durable, and thus I would use it much more.


Posted on

Testing TC4056 Modules – Fake TP4056 Clones

I recently bought 150 TP4056 charge and protection boards to make a bunch more Eco-Charge power banks. Once I received them, I noticed that the ICs they were using were another type of TP4056 clone that I had not seen before – the TC4056. I had previously used and tested 4056E chips without issue, but I wanted to test these as well just to make sure.
My giant 18650 testing station is ideal for testing, because it has slots that perfectly fit the voltage and datalogger that I made earlier. The problem is that all the chips on it are of the 4056E variety, so I got my hot air rework station and went to work. I first practiced on a few of the boards that I had destroyed, so that I could get a hang of desoldering ICs with a connection on the bottom of them as well. Once I got the hang of it (it takes longer to melt the solder than on small passive components), I removed the 4056E from one of the board currently in the testing station, and replaced it with a TC4056 from one of the new boards.
After letting it cool down, I plugged in the datalogger and charged up a cell. The graph can be seen below, and is perfectly acceptable for charging 18650s. Like the 4056E chips, this one got quite warm as well, so I left the small heatsink on it as I had done with all the 4056Es in my testing station.

Project By: Micah Black

Written By: Micah Black

Posted on

Dual Power 3D Printer (AC Adapter and Battery) – with Supercapacitors

With the recent addition of a powerwall to my workshop, I’ve been running everything I can from it. That includes my 3D printer. I wanted to have dual power sources for and be able to switch between mid-print, so I could use an AC adapter for overnight prints, and could print from the battery during the day while I can monitor it (I do not have a BMS installed yet – I bought a 7 cell monitor with 50mA balance that I will eventually install).

Before jumping in to that, I figured I should clean up the wiring.

I used a 3 position switch to switch between AC power, Battery power, and OFF.

When switching from AC to battery I used a supercapacitor bank to power the board for the few milliseconds while the contact switches position. The supercapacitor module I used has 6 2.7V 100F supercapacitors connected to a overvoltage protection board. I could have used some batteries for this, but supercapacitors work very well and charging/discharging batteries properly would have been more complicated, heavier, and less appealing. Charging the supercapacitors could have been achieved just connected directly to 12V, but if they are dead to start it would draw a huge current and charge up very quickly which is usually the good side of supercapacitors. However, the problem with this is overcharging and staying balanced. The charging current would also be too much for the 12V 20A power supply, and the voltage would probably drop too much. Being chinese parts, I did not trust that the supercapacitors that I used would be balanced, so I decided to limit the charging current to 1A with a 10ohm resistor in series with the supercapacitors. This resistor would also limit the discharge current of the supercapacitor bank, which I did not want – so a 20A diode in parallel with the resistor was added to be able to discharge the pack at up to 20A. I also added a switch inline with the capacitor bank to connect or disconnect it from the 12V system. This will hopefully put less wear on the supercapacitors and let them last longer (especially considering I do not think they will last too long being chinese parts – one was already dead out of the box).

The other reason for adding the resistor in limiting the charge current was to completely eliminate any chance of the supercapacitors overcharging. The bleed resistors on the supercapacitor board can keep the cells under 2.7V while they are being charged at 1A.
An XT-60 connector was used for the battery connector. One issue I have found is the converting my 24VDC powerwall to 12VDC cheaply, and transmitting 20A current over almost 12 ft from my powerwall to my printer. I have a few different models of DC-DC buck converters on order, and will test them out once they arrive – the one that I currently have has too much of a voltage drop when drawing 20A, and the printer shuts down.
This has been a fun project, and it was amazing when it worked on the first try (powered from a smaller 12V battery).

Posted on

2.4kWh DIY Powerwall

My 2.4kWh powerwall is finally complete!

Printing the holders took a long time to do, and I thankfully had a friend help out with the printing. I had to print out almost 100 holders, using just over a full roll of filament.

Then came the brunt of the work – making over 1500 solder connections for this build (It took a while). I did most of the soldering outside because there is much better ventilation, and the weather was nice, so why not take advantage of it?

The positive end of each cell was soldered to a 4A fuse. I chose 4A, as this power wall was also designed to be able to run an electric car that I worked on for the Waterloo EV Challenge with the EVPioneers. and needed to be able to supply 150A burst current. I only had enough 2A and 4A fuses, and the 2A would not give me enough power. For use as a power wall, I would recommend using 1 or 2A fuses because it will keep the cells within reasonable operating limits. Yes, most cells when new can do 4A (2C) continuous, but after a long life in laptops, it is safer to keep them under 1C continuous.

The negative end was connected to the bus bars with the extra legs of the fuse wire that were cut off from the positive end.
And that brings me to the bus bars. I was originally planning on using copper – either flattened copper pipes bars, but after checking prices and feasibility, I decided against it. I couldn’t find an easy way to attach the 8 cell modules to the copper pipes without soldering, and comparing prices of copper bars to aluminum bars, I went for 1/8″ * 3/4″ aluminum bars.

Attaching the 8 cell modules to the bars was another adventure. On each of the 8 cell modules, the fuses were soldered to a wire with a screw terminal on the end in order to be able to swap out the 8 cell modules without soldering. I was originally planning on using 16AWG wire for this, but after checking out the 12AWG wire that I had lying around, the 12AWG was a lot easier to strip, and would heat up less under heavy loads. On the positive end, I made the wire just slightly longer than the modules in so that they would fit in the smallest space possible, and have just enough room to crimp a screw terminal on to. The negative end got a wire that was bent up to the same level as the positive wires. I covered this longer wire with heat shrink as much as possible, 3 separate sizes to prevent it shorting out where the positive end just pokes out the end opposite of its screw terminal.

Now for actually getting these parts – a $70 trip to the hardware store later, I came back with 8ft of aluminum, 100 12AWG screw terminals, 200 6-32 nuts and bolts (they were the cheapest), and some wood for the frame.

I cut the aluminum into 1ft lengths, then drilled lots of holes in it for mounting the aluminum to the frame of the power wall, and for the screw terminals to attach to. I did not want to have to get out a pair of pliers to hold the nuts in place and risk shorting something out when screwing the packs on to the bus bars, and I had recently seen Adam Welch make some captive nuts on his solar shed bus bars. So I designed a similar system that will hold 2 nuts. After printing out 56 of them, I started putting the nuts in, and sliding them on to the aluminum bus bars.

The frame of this power wall is made out of wood. I really should have used something non-flammable to mount everything to, but I could not find a metal cabinet or something similar in the right size. I also didn’t want to spend $150 on an enclosure, so wood it is. With all the testing I have done on these cells, and individually fusing each one, I do not think there will be any problems. I will be constantly monitor this looking for heaters, and checking voltages.

Each parallel group is separated with a piece of 1×3, which I mounted the aluminum bus bars on top of. Once all 8 bus bars were mounted, I started adding the packs in, balancing capacities as best I could while at it. I used an impact driver to tighten all the screws – I had previously replaced the aging NiCad in the impact driver with 18650s, and it’s still working great. I did run in to one 3D printed holder that I stripped, but thankfully it was at the end of one of the bus bars, so it was an easy replacement.
To finish up, I added a 150A circuit breaker to the positive end, and added a 1/4″ clear acrylic sheet over the top of the batteries to prevent any shorts.

The inverter I used for this is a 1000W modified sine wave inverter. It was one of the cheapest ones on Amazon, and that would probably be the one component I would change if I did do this again. On the other hand, pretty much my whole workshop is powered with DC, so it’s not too big of a problem. I do like it though, because it heats up my 60W AC soldering iron better than the house AC. My regular soldering iron – a Hakko T12 clone – is powered with DC, as well as my lights, and I will eventually add my 3D Printer to that list as well.
I have yet to stress this battery out, or do a proper capacity test, but so far, it’s been amazing.

Project By: Micah Black

Written By: Micah Black

Posted on

DIY Arduino-Based Bike Computer

The inspiration for this project came from trying to repair the bike computer that I had originally. I lost the base for it, and figured I could print a new base, and given that there were only 2 contact points for the magnetic sensor, I figured that it had to be a reed switch.
So I held the 2 leads of a reed switch to the 2 contact points, while moving it over a magnet, and sure enough, my assumption was correct. At this point, I knew what I had to do to fix the bike computer I had now, but there were still a few things I didn’t like about it – current speed was not available, and it had too many other metrics that I don’t care about.
The only logical solution was to build my own custom bike computer. Breadboarding the circuit with an Arduino, an OLED, and a reed swith connected to an external interrupt capable pin (D2 or D3 on the Uno, Nano) of the Arduino was simple. Figuring out the proper calculation to get from number of rotations per millisecond to kilometers per hour was a little trickier than expected. I did easily find one online, but wanted to see if I could come up with the same result.

With the basic circuit and programming complete, I started to design an enclosure for it in Fusion 360. The enclosure mounts to the bike through a separate piece which is zip-tied to the handlebars. The reed switch is will be permanently mounted to the bike (again with zip ties) to the fork. To connect the permanently mounted reed switch and the removable Arduino parts, I used an XT-60 connector. It’s only electrical purpose is to pass the signal from the reed switch to the Arduino, but it also holds the removable part in place. I could have accomplished the same thing with some dupont-style 2.54mm pin headers and designed a clip on to the model, but I went the route of XT-60 connectors because I felt it was easier.

In the removable part of the bike computer, I had to fit an Arduino nano, an OLED, a sliding power switch, 4 push buttons for menu navigation, the XT-60 connector, as well as a li-ion battery, a TP4056 charge and protection board, and a 5V boost converter. It was a tight fit, but I managed to get everything in.

The battery I used for this build is not the typical 18650 that I have in so many other projects, as that would have required the enclosure to be a lot bigger. For this project, I took a cell from an old cell phone that I got from an electronics recycling, removed the plastic casing and protection circuit, tested the voltage (it was around 2.7V), and charged it up. The cell was rated at 895mAh, and the TP4056 modules are pre-set to charge at 1A, which would be just over 1C for this cell, I decided to change the charging resistor on the TP4056 board. I could have done this with a soldering iron and a fine tip, but given that I recently got a hot air rework station, I figured now would be a good time to use it. Typically, Li-ion cells can be charged at 1C, but given that this cell was used and sitting at under 3V for a few years, I decided to go for a 0.5C (roughly 500ma) charge rate.

The appropriate resistor for that would be around 2.5K ohm, but the closest bigger resistor that I had was 3.3K, which worked out to roughly a 350mA charge rate. I replaced the resistor, and charged it up without issue. The battery got connected directly to the BAT+ and BAT- pads of the TP4056 module. The OUT+ of the TP4056 gets connected to one side of the power switch, while the other side is connected to IN+ of the boost converter. OUT- of the TP4056 is connected directly to IN- of the boost converter. The boost converter output was originally a USB type A port, which I removed, and soldered wires directly to the 5V and GND pads on the PCB (the outer 2 pins of the USB connector). They are not labelled, but the GND pad is directly connected to IN-, so I got my multimeter out to check continuity. Once I had that figured out, I connected the wires to 5V and GND on the Arduino. All the buttons got connected one side to GND and the other side to a digital pin, declared as INPUT_PULLUP in the code.

It was a very tight fit getting the Arduino into the enclosure. While putting it in the first time, I dislodged one of the SMD capacitors on the back of the board. I soldered it back without issue, but then it would not accept uploading a program to it. Most likely, I also damaged some of the traces going to the capacitor, but I’m not sure. I replaced the Arduino, and got everything properly mounted in the enclosure. I soldered the OLED to the Arduino before screwing it in to the enclosure with 4 M3 screws. Then the Nano went it next, secured with friction for now. I put the XT-60 connectors in both parts of the mount, and secured them with two-component adhesive, which I also used to secure the Arduino at this point, and the power switch. After soldering the 4 push buttons for menu navigation to the Arduino, they got glued in with a generous amount of hot glue. The TP4056 module went in directly on top of the Arduino, the boost converter slid in behind the switches, and the battery fit over top of everything. I’m sure I could have made the wiring more elegant, but it would have taken much more time.

Overall, this accomplished exactly what I wanted it to do – get me information on current speed, and distance travelled while riding my bike. I also added in some functionality to track trips, and save up to 10 of them in EEPROM memory.


Project By: Micah Black

Written By: Micah Black

Posted on

100W LED Flashlight in a PVC Pipe

Back for round 2 of my 100W LED flashlights. I enjoyed the first one such much and used it enough that I decided to build another that solved a few of the annoying problems with that one (terrible battery life, constantly monitoring battery voltage, battery outside the main casing).
I’ve been thinking of building this for a few months now, and from the time finally decided to go ahead and make it, it took me around 8 hours of work on it to complete it. That includes making the custom battery, testing all the parts, and choosing resistor values.

Let’s start out with the choice of parts. I mounted everything inside a 4″ PVC pipe because I had seen it done before (link), and it is much sturdier than the MDF I used for the original.
As for a heatsink, I had to find one that fit inside the 4″ pipe. A stock Intel CPU cooler is perfect for this. For the control circuitry, I used pretty much the same parts as the last one – a 150W boost converter, an XL6009 Buck Boost converter, 2 potentiometer, and I also added an extra switch and USB buck converter to have a USB charging port.
The batteries I used are 12 Grey Panasonic NCR18650 from old laptops, all around 2800mAh. The BMS is a 4S 30A BMS from aliexpress, and works perfectly, as far as I can tell. I added a voltage monitor to the back of the flashlight as well.
And of course, we cannot forget the 100W LED, and accompanying lens.
I used M3 nut and bolts for all the attachments, as I have plenty of them lying around, and they are very common.

Starting with the control circuitry, I used a rotary tool to cut out a circle of MDF slightly smaller than the inner diameter of the PVC pipe to mount all the electronics to.
The boost converter is being used to boost the voltage of the battery pack up to a maximum of 32V for the LED. Anything higher than that, and the LED will start to draw too much current, heat up, and possibly explode due in incorrectly matched diodes. If you want to find out more about why this happens, check out BigClives’s video on it. Always be sure to know what you are doing when playing around with high power chinese LEDs.
The original potentiometer on the boost converter is a 10K trimpot, but that obviously had to come off if we were going to be able to adjust the brightness form the outside of the case. I started out with a 10K potentiometer, and figured out what resistance caused a maximum voltage of 32V, which turned out to be around 9K. I used a 5K potentiometer in series with a 4K of resistors to max out the voltage at 32V, but still have an adjustable voltage.
I also wanted to be able to control the fan speed, so I did the same procedure for the XL6009 buck boost converter, max voltage of 14V to overvolt the 12V cooling fan to give maximum cooling performance. I feared that the small intel heatsink would not be enough to properly cool the 100W LED at full brightness for very long. It turns out that the stock Intel fan has a built-in speed controller, so this turned out to be useless, but I did fry one fan while figring this out.
While testing the buck boost converter for the fan a potentiometer failed, and created infinite resistance between the wiper and the edges. This triggered the buck boost converter to boost to its maximum voltage which turned out to be over 60V. This let the magic smoke of the stock Intel fan, so I had to grab another from my bin, but I didn’t put it back in circuit until I had replace the potentiometer and tested the voltage numerous times on the output. I was surprised that the buck boost converter went up to such a high voltage, since its max adjustable output voltage is around 35V, the same as the capacitors are rated for. I’m glad (and surprised) I didn’t blow any of the capacitors, pushing 25V over their limit through them. Just another example of chinese engineering. If I had not caught this before I mounted it, the capacitors would have been taking that 60V for a much longer time before I realized what had happened, and most likely would have blown.

The USB Buck converter was also added with its own switch, and required no special wiring. Interestingly, there are no markings on the board to mark input polarity, so I got out my multimeter and testing for continuity between an input pad and the grounded USB shroud.

One quick note – controlling these LEDs with a voltage limit is not the proper way to do it. A current limiting circuit is much better, and will prevent the LEDs from burning no matter what the voltage is. They are much more expensive though, so I am sticking to voltage control, but limiting it below the max voltage. These LEDs can take up to a maximum of 36 volts (I believe) if properly controlled with a current-limiting device. I would highly recommend not driving chinese LEDs at their maximum specs, as that increases the chances of danger (again, see Big Clive’s video that explains much better why this is dangerous). I tested my LEDs, to make sure that they were not too far out of balance with each other. As you can see from the picture below, mine were matched fairly well – much better than the ones shown in Big Clive’s video. I am driving my LEDs at a max of roughly 33V.

To attach the LED and lens to the heatsink, I drilled 8 holes around the center, one set of 4 to fit the LED and the other set of 4 to fit the lens mounting points. I used M3 screws, and they tapped themselves into the aluminum very well. Before screwing the LED down, I put a blob of thermal compound in the middle of the heatsink. Same procedure as CPU mounting CPU coolers to a CPU.

Once I got all the control electronics figured out, I went on to cutting the PVC pipe and mounting everything to it. I drilled holes for the potentiometers, switches, and screws, then went outside to use a rotary tool to cut out the ventilation holes, cut the tube to length, and enlarge some of the drilled holes. It is very important to do this is a well ventilated area, and ideally use a face mask to avoid breathing in the PVC dust.

Using some 6-32 screws, washers, and some galvanized strapping, I created a mount for the MDF control board, and then mounted it in the pipe. After soldering the LED to the output and verifying that it worked, I put that inside the pipe as well, and drilled 2 holes through the plastic fan mount in order to attach it to the PVC pipe with some M3 screws.

Next I worked on building and mounting the custom battery. As I mentioned earlier, the battery is a 4S3P configuration, made up of Panasonic NCR18650 cells from old laptops, all around 2800mAh. Each cell is individually fused on the positive end with a 3A fuse, and the negative ends were soldered together with nickel strips.

The BMS output is connected to the input of the boost converter for the LED, and the buck converter for the USB port. I also added an extra XT-60 connector to the main terminals of the battery, as well as a balancing harness in order to be able to charge the battery with a hobby charger. I put a piece of foam in the back end of the flashlight to cover all the screw heads on the MDF board, wrapped the battery in 2 layers of foam, then put the battery in and another piece of foam on top. Packing the battery with foam is definitely not the best for heat, but I do not anticipate it to be a problem. These battries can supply a max of roughyl 15A, and I will only be drawing about 4A. To keep it from falling out the back, I added another piece of foam, and put an 80mm fan grill on top. I cut out part of the fan grill in order to put a 4S voltage monitor and a switch to have a rough idea of the battery level without any hassle. The screw holes in the fan grill were bent downwards and pushed around the outside of the foam so that 4 computer fan screws could be screwed in to PVC where I had previously drilled holes, and hold the fan grill in place.

All the was left to do was add a handle, so I cut a rough shape out of a piece of 1×4 with a jigsaw, then sanded it down with a rotary tool, and drilled a hole in either end of the flashlight and the handle to mount it securely. I added a layer of clear gloss acrylic spray paint to the handle to give it a bit of protection against moisture.

With that, my second 100W LED flashlight was complete! If you want to see the first one, you can check it out here. I like this one much better, as it is all in one self-contained unit, therefore is much easier to use and handle than the previous one.

Posted on

First Step Towards a DIY Powerwall

For the past few weeks, I have been designing, printing, redesigning, and thinking about cells holders for a DIY powerwall. I decided to go with 8P packs, as those will allow for easy replacement of faulty cells, and if one cells fails, I only have to replace a pack of 8. Each cell will be individually fused at 4A. As for spot welding vs soldering, I go with soldering – especially for the fuse wires, it is much easier, and does not damage the cells. I use a 60W soldering iron with a big tip, and have never had any problems.

I finally started getting the packs put together, and realized that the 350ish cells I was planning on using is much more than I originally thought. I am only using cells over 2000mAh, and each cell has ben tested in my 18650 testing station, according to my process here. I may need to find and strip down a bunch more laptop batteries.

7S will be the configuration for this wall, and I will start out with 6 or 7 packs of 8 cells in each parallel group (7S 48 or 56P). This 7S is ideal for a 24V system. Unlike 12V systems, 24V systems allow the use of common 24V inverters, and DC-DC Buck converters can be used to achieve different voltages. If I had gone with a 12V system, then it would be the classic debate of 3S (9V-12.6V) vs 4S (12V-16.8V). Most 12V inverters have a low voltage cutoff of around 10.5V and a high voltage cutoff around 14V. With either option, over half your battery  remains unused. Going with 7S, we can use a 24V inverter. The voltage range of a 24V lead acid system and 7S Li-ion pack are almost perfectly matched, so all the power in the batteries can be used.

I will eventually be building some solar panels to charge this, but for now, I will stick to charging it with a 930W server power supply that I bought on eBay for $25, paired with a 600W converter from Aliexpress. I will also use some high power TP4056 modules or an iMax B6 to top off individual cells when required. The best solution for cell monitoring would be a Batrium BMS system, but those are expensive. Instead, there will be a small voltage monitor attached to each cell with a switch to monitor voltages while charging and discharging, but does not draw power when sitting idle if the switch is turned off.

The whole thing will take up roughly a 1x3ft area, weigh about 30kg, will be able to store around 2.5kWh of energy.

Another idea that might come in the future to complement the solar charging I will be adding to this – attaching a generator/car alternator to a stationary bike in order to charge the power wall whenever you want.

Stay tuned for more updates on my powerwall coming soon!

Project by: Micah Black

Written by: Micah Black


Posted on

Designing my own BMS Module

When I saw Julian Ilett’s video about voltage protection ICs being used improperly on supercapacitor protection boards, I decided I would use them correctly to build my own basic BMS. In no way do I believe that this is better than the options currently available, but it was a fun project.

The main inspiration for this build was the XC61C low voltage detection ICs. These can turn off their output when the voltage falls below a certain threshold – depends on the model you get. I used 3.0V to ensure that my cells do not get over-discharged. It would probably have been a better idea to use the 2.7V versions, as that would allow for more usable capacity and still not discharge to an unsafe level. The cells would probably recover back to around 3.0V anyways once the load is disconnected.

There are 2 main parts for this board – the cell detection side, and the output control side. Both sides are separated with optocouplers in order to isolate them from each other, as the ground reference is not the same for the low voltage detection ICs and the rest of the board.

On the cell detection side, I used the XC61C low voltage detectors across each cell. They are attached to a the input side of PC817 optocouplers, in series with a 470 ohm resistors. Optocouplers are not ideal for this, as they do use a fair bit of power just to drive the IR LEDs (around 10ma). After leaving it connected to some lights in the shed for a week or so, the output would no longer turn on – I’m glad that the BMS did its job. However, voltage on the top cell was below 2V, so too much current is being drawn by the Optocouplers. It was a fun project, and did what it was supposed to do, but with some potentially damaging effects on the battery – overdischarging the cells through the IR LEDs in the optocouplers.


Project by: Micah Black

Written by: Micah Black

Posted on

Another Flashing LED Sign

I recently built another flashing LED sign and wanted to post some pictures of it, so here they are.

Building this one followed a very similar process to the rest (linked below), but this time it worked on the first test!

Check out the original builds and instructions here.

Posted on

Adding LED Lighting to Outdoor Shed

When we need to get something out or put something into our shed when it’s dark, it’s always more of a hassle than it should be. It starts off with trying to find a flashlight, then only having one hand to carry stuff out and put it in the right spot is not easy. Solution: put some lights in the shed!

Let’s start off with the easiest part of this project, the LED lighting itself. I used some very simple 12V LED strips with 5630 or 5730 LEDs for the lighting. 2 strips of them worked well because the shed is fairly deep, and allows much more even lighting. In total, there are 2m of LED strips. Current draw is roughly around 1.5A. Now to find a suitable way to power them.

As this is outside, running AC power to it and using a small power supply was impractical. I have lots of 18650 lithium-ion batteries lying around, waiting for me to do something with them (a power wall is in the works right now), so I made a 4S 4P pack inside a 4×5 cell holder to allow for extra room to hold a BMS circuit and DC-DC buck converter to get 12V for the LEDs from the battery pack’s fully charged 16.8V. I chose 4S because I wanted to use the BMS circuit that I designed (more details coming soon) to prevent over-discharging the batteries. The BMS will cut all power¬† going to and from the batteries if the voltage of any cell goes under 3.0V (accomplished with XC61C low voltage detection ICs). The buck converter I used is a 5A buck converter that has lots of headroom to drive the LEDs. I set the voltage to around 10.8V so that the LEDs are more efficient – less voltage drop across the series resistors, and thus less power lost to heat in those resistors. Some of these LED strips have been known to get fairly warm, so undervolting them is a good trick to keep them cool and prolong their lifetime.

Mounting everything to the shed took some screws, some scraps of wood, and some 3D printed clips. The adhesive on the LED strips did not stick to the wooden shed, so I designed some LED strip clips that I also used to tidy up the cables. The files are available on thingiverse here. I might eventually add a PIR motion sensor and RTC to this setup to make it turn on as you approach, but not during the daytime. One problem that will come up with this is the cold weather in the winter. Discharging the batteries over in -30 degrees Celsius temperature is not good for them – it will reduce capacity, and increase the internal resistance, resulting in a lower operating time and lower possible discharge rates. These batteries are not getting pushed to their limits at all, but I’m not too sure if I want them being used out in the cold. At the very least, they will be charged more often and kept inside as much as possible. I might make 2 more to switch out so that each battery spends as little time outside as possible.