I moved over to SMD components

At first, I always thought that using SMD components is something that requires heavy duty industrial processes. The components are incredibly small, and I thought that a computer controlled robot is mandatory in order to place them correctly. I also had no idea how the actual soldering happens. But when I started looking at it, I quickly realized that these assumptions were simply incorrect.

When I made the first iteration of the watch winder, I decided to use two SMD mosfets as a trial. It turned out to be not as hard as I had first thought. As I moved to my next projects, I decided to start preferring SMD over through hole. SMD components have a lot going for them:

• Much wider variety of components
• Easier to manage lots of different components
• I can make smaller PCBs with SMD components.
• Turned out not to be as difficult as I had first thought, there are tons of Youtube videos to explain different techniques.
• Components are generally cheaper
• I always like to learn new skills

At first, I thought that I would simply use a soldering iron with a fine tip to do this, and I did manage to make some boards on my next project that were sort of OK. The solders weren't pretty, but they worked. The problem with that though is, that if you want to remove a component, particularly a SOT-8 or SOT-16 package, it just can't be done. So after some research I decided to buy a hot air station, I bought an Atten ST-862D as it had good reviews and wasn't too expensive. The one advice I would give to anyone considering buying one, is to choose a station with the hot air unit in the main box, not in the handle. It is a bit more expensive, but it is capable of producing much better airflow.

So now soldering was a breeze, but I thought that I could still improve my workflow. I was afraid that I would damage the components if I needed to heat them too long. So then I decided to add a hot plate to my arsenal. That made a big difference ! Instead of trying to heat the entire PCB with hot air, the board was already close to the reflow temperature. The hot air station only needed to provide the final bit of heat, making the whole process much easier to control.

Workflow

After a lot of trial and mostly error, I found my preferred workflow

• I also order a stencil with new PCBs (there are exceptions). It makes applying the solder paste so much easier. I printed this jig to help the process.

The black part is the same for all PCB:s, the green part is specific to each PCB design. The holes have magnets to keep the stencil in place.

• After the paste is applied, I lay the components with tweezers. I found that angled tweezers (of good quality) are best for the job.
• Then I put the PCB on the hot plate and heat it up to 170 °C. I let it stay there for one or two minutes and then apply the hot air from the heat gun.
• I apply 350 °C with 10% airflow from the heat gun until the solder paste reflows.
• After the PCB has cooled down a bit, I inspect it visually with a digital microscope for obvious faults or shorts. I check all ICs with a multimeter for shorts. If there are any, I try fix them with the soldering iron.

Now there are exceptions. If I have a PCB that has QFN packages, I will not use a stencil. The main reason for that is that the hot air station easily blows small QFN packages out of alignment. For similar reasons, I have for now limited myself to 0805 or larger packages when it comes to resistors or capacitors. I did try to put a 0603 in place, but it just kept flying away.

QFN packages require special attention. Those can be a real pain to get done right, but the best method that I came up with is this. I first apply regular leaded tin to all the pads on the PCB where the QFN package is to be installed.  After I am happy how the pads are tinned, I put a drop of sticky flux on the footprint, and then position the package in place. The flux keeps the component from flying away while it is being heated. After that I do the reflow as I did with the stenciled PCB, but now I just solder the QFN in place and once it is soldered, I can inspect it visually.  Now since I didn't put any other components on the PCB yet and I no longer can use a stencil,  I need to apply solder paste with a syringe to each pad of the remaining components. That usually takes some time, and I usually also put too much paste on the pads. The paste likes to stick to the syringe, so the best way to handle that I came up with was to use a wooden cocktail stick to fine position it.

Did I mention I always put too much paste on the pads ? So much so, that more than once while the PCB is on the hot plate the paste simply becomes more fluid and becomes one big puddle of paste on both ends of a resistor. That is normally not a problem though, because once the solder reflows, surface tension usually pulls all the solder into the right place.

If I have a short between two pins of a SOT-16 package, I fix that with the soldering iron (and a hoof tip) and some flux. Normally that is enough to make the solder flow to the right place due to surface tension.

Conclusion

I found that SMD is actually easier to do than THT, particularly if you need to remove a component for whatever reason. Cleaning a hole of a THT component is a colossal pain, and you risk damaging the pads doing it. But you do need more tools to achieve that. With THT you only really need a soldering iron, with SMD you need also a hot air station, preferably a hot plate, a microscope (or at least a good magnifying glass or both) and you need to play with more chemicals (solder paste, flux and IPA to clean up the mess).

I still do use THT where it makes sense (for example connectors), but SMD is now my default choice for all new designs. Soldering SMDs turned out to be a lot easier and actually more fun than I had originally thought.

Updated version of the cooking timer

Christmas was approaching, and I thought it might be a good idea to make a gift to someone myself rather than buying one.

Since I already had all the components for a cooking timer, I decided that would be it. However I decided to address some of the shortcomings of the first version.  These were:

• The power solution, I decided that this version would use Lithium-Ion batteries instead. This also made the base a lot smaller.
• The buzzer, I decided to replace that with a speaker and make a mechanical solution to  amplify the sound.

Since making the first version, I had already purchased the 3D printer, so I had a lot more options available to me. I could for example print in red filament, and even do multicolor printing. So another change I made was to print the switch texts (ARM/DISARM) in red.

The electronics bit remained largely unchanged, apart from the fact that instead of soldering a buzzer to the PCB I soldered wires with 2-pole connector that would then allow me to connect the speaker.

So all these changes ended up affecting the following components that needed to be redesigned.

• The base, a complete redesign was necessary to accomodate for the new battery and the USB-charging module. Since 3.3 volts was sufficient for this, there was no need to use the buck boost module. This design did need quite a few iterations to get the module seated right. The problem with the module is, that it has no real attachment points, you cannot  screw it into anything.
• The "dynamite sticks" were also redesigned to house the speaker and the actual battery. Since there was plenty of unused space there, it was a logical place to put them. The battery was put in a cradle, that then screwed into place in one of the tubes. The speaker was put in a holder in another tube with a similar screwed in place holder. 
• The display housing was redesigned to support two color printing for the labels. That meant that the housing needed to be printed in another orientation.

After these changes, this is how it ended up looking !

This time I have all the files in store, you can find the 3D prints in MakerWorld following this link, and the PCB is available from PCBWay following this link. The actual build, schematics and bill of materials is described in the original article. The only change to the BOM is the addition of the charging module, Lithium Ion battery (18650)+holder and a miniature speaker.

I even wrote a user manual for the device since it was given as a gift.

Assembly instructions

Before you attempt final assembly, test that the electronics works as intended !

After you have printed all parts, you can follow these instructions for assembly. You will need some heated inserts for M2, M3 and M5 screws, all the PCB:s and modules are attached with screws as well as all the parts are put together using screws.

1. Bottom housing

Insert 4 M3 heated inserts into the raised legs, screw the PCB into place paying attention, that the dip switches are visible through the hole. Attach the charging module to the slot on the corner of the piece, and put in place the small pin to hold it in place. You need to solder wires to the charging module as per its own instructions, one pair of wires needs from the OUT+ and OUT- terminals to the PCB, but the positive wire needs to be intercepted by the power switch. For now, simply insert a two pole connector on the red wire between the charging module and the PCB. Another set of wires need to go from B+ and B- terminals to the battery. Again insert a appropriate connector on these wires and then later another connector on the battery wires to make the connections. Make sure you use connectors that enforce correct polarity !!

2. Top Housing

Insert 4 M2 heated inserts into the corners of the housing. Do not screw the two housing together just yet.

3. Middle block

Install 4 M4 heated inserts to the bottom of the housing (thicker part) and 4 M2 heated inserts to the side. Now you can screw the Middle Block and the Top Housing together and run all the wires through

4. Display Module

Top

Bottom

Install 8 M2 heated inserts into the Top Module. Screw the seven segment display, rotary encoder and the power switch (arm/disarm) into place. You need to prepare the respective wires to connect the modules to the PCB and now is a good time to install the wires from the power switch to the charging module. The housing beneath the center block is going to be tight, so you should make the wires long enough to connect to each other in the center block area. Avoid these connections in the housing part !

The bottom part screws into the center block, and the top part screws into the bottom part after that. at this time you should have all other wires except the battery and the speaker connected. Now you can also screw the Top and Bottom housings together. There is no picture, but the knob for the rotary encoder is simply pushed into place.

Run a fake wire from the opposite side of the power switch, that will serve no functional purpose but it will be the detonator wire in the ready device.

5. Dynamite sticks and battery holder

The part on the left has slightly thicker base, install a M5 heated insert in to the middle of this part.

Install a M3 heated insert into the battery holder. Screw the 18650 Li-Ion battery holder into this holder, and install the battery. Make sure you get the polarity right ! After that, screw the holder into one of the tubes on the left side part. Now you can run the battery wires to the connector you made in step 1.

Put the miniature speaker into the tube with threads on the right side. Run its wires through the above plug and screw the plug in place so that the speaker stays put. Connect the speaker wires to the wire from the PCB.

Now you can screw the Dynamite sticks together so, that they tighten agains the middle block.

6. Caps

There are two types of caps in the printed parts. The one with the "nozzle" is going to be used for the fake detonator wire you made in step 4, and the others will cap the remaining tubes. Leave the tube with the speaker uncapped ! The caps are glued into place.

Now the Timer is ready !

So I bought a 3D printer

 So I bought a 3D printer...

During my watch winder project, I got so frustrated about printing at the library, that I decided I need my own printer. So I started studying the interweb, watched endless Youtube videos and read endless reviews.

My goals were to have a reliable, easy to use and future proof printer. After carefull market analysis, I had limited the selection to two brands, Prusa (which was what the library had) or Bambulab. After more analysis, I ended up choosing a Bambulab P1S with AMS 2 Pro, my reasons were:

• Good reviews
• Ease of use
• Good quality prints
• The AMS2 option
• a very good deal from Bambulab website !

I do not mean to say, that the Prusa would not have been able to do the first three bullets as well, but money does not grow in a tree, so P1S it was. I later learned that the probable reason for the good deal was the introduction on P2S, but that's life...

I was aware of the controversy around Bambulabs policy on using 3rd party slicers. But I am happy using Apple ecosystem products, so I figured that this is pretty much the same thing so I did not think that was such a big deal for me.

I was able to set up the printer real quick, and I was printing in no time at all. That really speeded up my work and made it possible for me to finally finish the watch winder project.

Since then I must say that it was a really good purchase. I've been using it a lot for many different things. It's good for making precise jigs. I've used it to make various spare parts and to try different concepts. But most importantly, it allows me to perfect my designs. If each iteration of a print took weeks, at some point you would settle for good enough. Now I don't have to.

Travel watch winder

The next project I embarked on, was a travel watch winder.

I have a few mechanical watches, and occasionally I like to wear one watch for a week and then swap out to another for the next week. This means that the watches would stop unless I wind them during the week, that is a nuisance as setting a mechanical watch is a bit of a hassle.

Sure there are commercial options, but they tend to be a bit pricy and this seems like a nice project to do. So let's proceed !

Design goals

• Portable enough to take on trips
• Able to work on battery power for at least 24 hrs
• USB-C chargeble
• Programmable on operating hours and turns per day
• Nice looking

Tools used in this project

• Computer for programming and PCB design
• Multimeter for troubleshooting
• Soldering iron
• Wire clamping tool
• Breadboard
• 3D printer
• Sowing thread and needle

Bill of Materials

• Raspberry Pi Pico W2
• TP4056 Charger Module
• Buck Boost Converter
• ULN2003 Darlington Array
• 28BYJ-48 stepper motor
• DS1302 Real Time Clock
• Buzzer
• FDN340P P-channel mosfet
• 2 x BSR802 N-channel mosfet
• BC337 Transistor
• Resistors 3 x 200 kΩ, 2 x100 kΩ, 10 kΩ
• 2 x Diode BAT60A
• Ceramic capacitor 10 nF
• Electrolyte capacitor 220 µF
• 18650 Lithium Ion battery
• Power switch
• JST-XH connectors (5 pole and 2 pole)
• 3 x 2-pin generic connectors
• Lots of black PLA filament
• Brass heated inserts (for assembly)
• M2 and M3 screws
• Black fake leather

Links to resources

• Github repository https://github.com/PSuElectro/Watch-Winder-V1
• Schematics
• PCB Design 

I ordered the stepper motor and the real time clock as ready modules from Amazon.

I began by planning the circuitry and the initial concepts of the program for the device. The very first bits of progamming had me study the web server functionality. Again, I found an excellent library from Gurgleapps website. I built the web interface with basically three features, adjust the number of revolutions per day, the time window during which those revolutions should take place and a calibration function to adjust the winder orientation. The idea with the latter being that I would keep a tally of the steps that the motor takes, and if it starts in a known position, it would end up back in the same position. This is how the interface turned out.

The winder has a battery backed up real time clock module. I decided (for simplicity) to not include a time setting function, but rather every time you press the save settings button on the first screen, the winder is set to the same time as the device (like your phone) controlling it.

While I was working on the code, I also needed to start testing the circuits. For this I assembled the whole circuit on a breadboard. In this project I was still using mainly through hole components, but for the power rail I needed to use two SMD mosfets so I decided to venture into SMD soldering. It wasn't too hard, as long as you didn't need to remove any of those components. 

After the circuit was completed, I ordered the PCB from PCBWay and started to design the enclosure. And boy did that take a lot of time. Like I mentioned in the previous project, the printer at the library was slow, and I managed to book only one session a week. So it really was frustrating when the parts did not turn out as expected ! So at the end of the day I ended up buying my own 3D printer, but I'll do another post about that later.

The PCB installed in the electronics sled

Then it was time to assemble everything. Using the heated inserts and some screws, I put together all the 3D printed parts. To make it look nicer, I glued fake black leather to the outer surface of the winder. For the cap, I purchased 1.5 mm clear plexiglass. That is just about thin enough to cut with scissors. I glued that and it's frame together with Gorilla super glue.

The one part that I am not at all satisfied with, is the cushion on which to actually mount the watch. In the end I settled on a piece that I sew together and filled with foam, but it could look nicer. Also it is held in place by squeezing it into a sort of a fork, and the PLA fork gives way over time. However the design does allow me to come up with a better solution later.

So this is how it finally turned out !

Pretty good all things considered ! I have taken it on trips, and it does serve its purpose pretty well. The battery lasts for about a day (depending on how many revolutions you want in a day), but I normally leave it plugged in.

Conclusions

I'd call the project a success, although certain things I could have done differently. One thing that didn't work out quite as expected, was the calibration feature. The stepper motor misses steps for various reasons (the watch is causing too much torque if not aligned well), so even if I calibrate it, it normally ends up in a different position. Another surprise was the terrible accuracy of the RTC circuit, it varies even minutes a day ! There was also a tiny mistake on the PCB that I was however able to fix by cutting a wire. For some reason, the real time clock would stop when going to battery power. The solution was to cut the wire to Pin 1 on the PCB. Also there were certain finishing details that didn't turn out perfect.

During the project, I spent way too much time tweaking the mechanical design. Once I got my own 3D printer, it really helped a lot as I could turn around a part in a few hours instead of weeks. I also managed to fry one Pico while soldering it. I suspect, that the large GND plane on the PCB was the culprit, as it meant I needed to heat the pin for way too long to get the solder to flow correctly. I changed the strategy after that, I now solder header pins to the PCB and then snap the Pico in place after soldering. 

My son did ask me to make him one too, but with a different form factor. Instead of making another one just like this, I decided to revise the design to fix the issues I came up with this one. But that is for another post.

New power source

In my first project, I used AA batteries as the power source for the device. While being easy to implement, it did have a few drawbacks. It is bulky, it cannot be recharged and it does cost some money. Also with 3 AA batteries, you can provide a 4.5V voltage. Not quite 5V which would be ideal for some more power hungry devices such as a stepper motor.

So I started to investigate options. I pretty fast found ready made components from Amazon, that would do the trick with lithium ion batteries.

TP4056 charging module

This thing would take care of the charging and protection of the Lithium-Ion battery, but it would only produce whatever voltage the battery has, so it needs a boost converter to go with it. Well I found this piece again from Amazon.

Adjustable boost converter

So with these two components working together, I can manage the battery charging and output voltage. The boost converter is capable of outputting up to 26V at 2A max load (although they recommend max 1A on continuous use). I don't imagine I would be needing anything above 5V for a while though.

The boost converter was a bit of a tricky piece of hardware. It has a microadjustable potentiometer to adjust the output voltage, and it was a bit confusing to dial in the desired output voltage, but at the end I did manage to make it work.

One feature that this combo was lacking, was the ability to bypass the charging circuit for the load when charging the battery, however I did find resources on the internet to build that yourself with a mosfet and a diode. The idea is, that if you route the power to your device through the charger, the battery ends up trickle charging continuosly, which is not good for the battery. This is not an issue if you don't charge and use the device simultaneously.

Let's see how the next project turns out ! The idea is to combine everything into a single PCB to reduce the need for additional wiring.

Amazon links to above components

https://www.amazon.de/GTIWUNG-Battery-Charging-Protection-Functions-Pack-15-Type-C-USB-3-7-V-holder/dp/B0CSJR4CYJ?ref_=ast_bl_cpl_dp&th=1

https://www.amazon.de/-/en/Converter-QIQIAZI-Voltage-Compatible-Arduino/dp/B0FVM69RGB/

Funny looking cooking timer

This was my first electronics project since childhood, and compared to those days it was a project with steroids !

Design goals

• Has to look funny !
• Should be able to count down from 99 minutes or 99 hours
• Battery operated
• Raspberry Pi Pico based

Tools used in this project

• Computer for programming and PCB design
• Multimeter for troubleshooting
• Soldering iron
• Wire clamping tool
• Breadboard
• 3D printer

Bill of Materials

• Raspberry Pi Pico 2
• Dip switch (2 dips)
• Buzzer (eg. https://www.amazon.de/dp/B0DGD4YJ1P?ref=ppx_yo2ov_dt_b_fed_asin_title)
• Rotary encoder (eg. https://www.amazon.de/dp/B079H3C98M?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1)
• Transistor BC337
• Resistor 12 kΩ
• Schotky Diode 1N5817
• JST-XH Connector 2 poles
• 2 x JST-XH connector 4 poles
• Toggle switch (eg. https://www.amazon.de/-/en/QWORK-Toggle-Switch-Position-Arduino/dp/B0B56S4F1X/ref=sr_1_4)
• AA Battery holder ( for 3 AA batteries, eg https://www.amazon.de/dp/B07WGY91YC?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1)
• Misc wires and connectors

I started the project by designing the software to run on the Raspberry Pi Pico. To do that I also needed to build semi simultaneusly the test circuit on a breadboard, and in order to do that I needed to design the circuit with KiCAD. So I ended up doing all these tasks simultaneusly. From the beginning, I decided that I would tolerate simple mistakes, the primary goal would be to make it work.

So the circuit ended up being rather simple, which was good as it made this easy to test and would not set that bar too high.

Download PDF

Not shown in the diagram is the actual power source which was 3 AA batteries and the power switch, as these components would not end up on the PCB. So in addition to the different connectors to connect peripherials, the circuit only had a Schotky diode, a transistor, a resistor and a buzzer. A dip switch was added to the circuit to control simple settings of the timer. The other peripherials are a 7 segment display, a rotary encoder and the before mentioned power switch. This being my first design, I opted to use through hole components for everything, or rather I did not even consider using SMD. At the time I did this I was under the impression that SMD soldering needs expensive hardware and is something that is way out of my league (I later learned that that is not the case).

Next step was to do the coding. It did not end up being very elegant, this being my first ever Python project. The way the timer operates is simple, when you turn it on, it will default to previous set time, and when you press the rotary encoder it will start counting down to zero. When it reaches zero it sounds a morse code signal. Part of coding these days is copying the work of others (when they publish their work to public domain), so I ended up using a library from a developer called GurgleApps to handle the rotary encoder. You can find the entire micropython implementation in my Github repository.

Now it was time to test the circuit before placing an order for it. After a ton of debugging and problem solving, I did manage to make it work !

Then it was time to proceed to the mechanical implementation. In Finland where I live, most libraries have 3D printers that are either free to use or very inexpensive to cover the material costs. Mine was free. So I started booking times to print the enclosure. I designed the enclosure using Fusion 360 CAD software, which is rather capable and Autodesk also offers free licence to personal use. 

But that turned out to be the most time consuming part of the project. I must have printed each part at least five times before I got it right. I have since learned that this is pretty normal, seldom you get it right on the first try. There is always some small detail or tolerance that you did not anticipate. The reason this took ages was that the printer at the library was rather slow, some of the parts would take 7-8 hours to print, and the library allows you to have only one booking at a time (so that others get to use it too) and since the printers are rather popular, it might take a week before you get a new booking. That is very frustrating if you just spent several hours printing a component that turned out wrong.

The library only had black filament, so I needed to figure out a way to make certain parts of the enclosure red. After I discarded the idea of painting it, I decided to glue red paper on those parts to make the desired look.

So at the end, this is how it turned out !

I use the timer almost every day when I cook, and it does bring me joy. It turned out to be even more convenient than the timer on my stove, as it simpler to operate and it remembers from the last time how long the eggs need to boil. 

Conclusions

Although this was a rather simple project, it did teach me a lot. I learned a bit of Python, I learned to design things for 3D printing and I also learned a lot about 3D printing in general. There were some design flaws too, the way I designed the enclosure means that I need tools to change the batteries (though they last for months in normal use). The biggest flaw however was, that the buzzer does not make a lot of noise. If I am not in the kitchen and there is a lot of ambient noise such as a TV, I may not hear it going off. A big takeaway was also the need to really concentrate on the 3D design, that will take a ton of time if you need to keep printing the same parts over and over again. It is also pretty important to get the PCB design right, as it takes a couple of weeks to get them from China (unless you are willing to pay serious money for quick delivery). This project had a very simple PCB, so I got it right on the first try.

If you'd like to try the same project, you have the schematics and the code available on this page. You can order the same PCB from PCBWay using this link. (Disclaimer: If you do decide to order this, I earn a small commission). I have since I made this altered the CAD designs, so I do not have the STL files for 3D printing available anymore. Also the designs are for the specific display, switch etc that I used, so they would probably not work with other components.

So I decided to start doing electronics

When I was a kid, I enjoyed doing different electronics projects. Now that I am nearing retirement, it's time to pick up new hobbies or restart old ones. This seemed like a natural choice since this was something that I enjoyed a lot back in the day.

Back then, the typical way to do a project (at least for me) was to find a a ready made kit, or at least schematics that someone else had desinged and simply implement it. Even that was a usually a lot of work, since it typically involved making my own PCBs with different types of not so healthy chemicals.

But I quickly discovered, that thats's not how its done today. Today you design the PCB with some sort of CAD software on your computer and then order it professionally made from China. That sounds like a lot more pleasent way of doing this !

So now I needed to figure out how to start all this. I had already thought a lot about what kind of projects I would undertake. I did know that I was interested in doing things with a microcontroller, so I started studying different options. For different reasons I finally decided that I would start with a Raspberry Pi Pico and Micropython, as that seemed like a well supported way of getting started. 

So first I needed to figure out how to write Micropython. Having never done anything really with Python, that seemed like a good way to start. I chose Thonny as my development environment.

Next step was to choose how to design the circuits around the Pico. After some evaluation I decided that KiCAD would be the right choice for me (for no other particular reason than it being free to use) so I started to learn that.

Now the only thing left to do after all this, was to decide what my first project would be, more on that on the next article.