Feb 05

Car PC “Completion” and Installation notes

In atypical fashion, I have completed a project to a certain extent! I have my CarPC installed in my VW TDi MK4 Golf, and these are notes about the assembly, trouble shooting, bench testing, etc portion of the project. This may be my last post on the project, unless I feel that enough work goes into perfecting the usability of the system to warrant an update…

I’ll just let the pictures do the talking:

In the middle of my project, I saw this on my way in to work. Whilst my VW was gettin’ pimped, these two unfortunate brothers of mein (see what I did there?) were getting wrecked. As you can tell from the salt on the ground, the floor was slick. My suspicion is car three (which is not pictured here) turned into the spot going way too fast, and drifted into the jetta. How careless. I took a moment of silence, don’t worry.

The hardware on the bench. I used quick crimp connectors as they seem to be preferred in car installations, for whatever reason. Probably because a soldering iron isn’t easy to power from a cabin.

Internals of the Carnetix P1900. I ended up desoldering the input for the power supply, as the space got really tight. I think If I did this again, I’d desolder both sides so the install is as low profile as is possible.

I tweeted or instagrammed this picture because I thought it was funny. Apple products are said to be magical, but I’m pretty sure this takes the cake.

4″ exactly. Impressive hardware and technology density in the Intel NUC (I’m biased).


The HDMI cable provided with my monitor was about 5.5 feet too long (it’s a ~6foot cable). Since the HDMI cable also had a breakout for USB over HDMI (for the touch screen interface), I had to end up hacking together one of my own. This wasn’t a breeze, but I’m pretty handy with a soldering iron. Above is the desoldering of the existing solution. Looks to be a plastic, or injected glue strain relief.

Comparison of what I have to solder with what I have to solder to.

Testing the “final” product. I added hot glue for strain relief, but will probably use a product like “Sugru” next time, which is moldable rubbery clay-ish material. It’s commonly used to repair electronics. I would also like to try a right angle connector, but the base material is already pretty high profile, so I didn’t bother. I could get away with something as short as 3-5″ for this custom cable, but didn’t want to do it twice.

For anyone curious, the USB connection is as follows:

USB side [1,2,3,4] = [VCC,D-,D+,GND]


When I assembled my cable, the twisted pairs seemed to be mismatched with what I “ohmed out” with my multimeter, just a heads up to check your connections when you’re done. You can find pictoral pinouts on pinouts.ru or a google search…

Testing touch screen functionality.

Test benching the hardware. I connected to a power supply and “pulse started” by touching the Carnetix Pulse start (blue wire) momentarily with a +5V supply. In car I use the “IGNITION” line attached to my ignition switch key sense (key in, power supply on). In the MK4 2003 golf, this line is only accessible via the ignition switch bundle. You can see this on the LEFT side of the steering wheel (right above). There’s a bundle of cabling on the top of the steering wheel opposite of the ignition/keyhole. The wire is BROWN with a red stripe (NOT THE OTHER WAY AROUND). Some guides I found online said that the IGN signal was routed to the stereo wiring harness, but I found this to not be the case, and I suspect this is only true with the earliest 99.5 MK4s (the ones without the double DIN monsoon).

Side by side shot of the PSU and NUC. I used screwed together stand-offs as a stopgap solution for spacers and prototyping.

Took a flash shot to show how much “additional” room can be had by cutting out the back side of the radio cage. I have a spare cage for test fitting indoors, but I didn’t end up needing the extra space. Attached is a METRA wiring harness adapter, so I don’t have to cut/splice into the original harness. I ended up removing all non-audio cables from this harness and wiring up RCA cables to the harness. Right Front/Rear are hooked up and tied together, same with the left. This implies there’s no fader control. I’m willing to sacrifice it at the moment so I don’t have to purchase a separate amp.

GPS USB module out of case. Honestly, I don’t know why I removed this from the case… I think I just have an unexplainable fascination with voiding warranties… IIRC this is a SiRFIII chipset. If the reception sucks (behind the fuse panel), I’ll probably just end up using a bluetooth module.

ABS case taking shape. I chose ABS because it’s easy to machine/drill, and it melts together with MEK “glue.” This is extremely friendly for apartment prototyping. The down side is that I have ABS flakes all over my carpet… not sure how easy that’s going to be to get out…

Size comparison to the enclosure I’m building in. Even with a reference picture, it’s hard to tell how small my project enclosure actually is. I had a LOT of time sunk into fitting…

I ran out of ABS, as I purchased a 12″x12″x12″ piece of ABS from Amazon (raw plastic sheets are hard to find, and they aren’t cheap. If anyone knows where to source some, just let me know). Since I was working on my project mostly during nights, I didn’t have an opportunity to go out and buy material. I found a really old graphics card in my spare parts bin, and used hot air to remove all the parts I wasn’t interested in. After sawing it down, and drilling some holes, I had a spare “PCB separating” shelf.

Finished shelf.

Shelf in between the NUC and PSU. This is required because the mounting holes overlap a little bit.

Burned myself with my soldering iron while working. Still trying to figure out how the hell I did this… my track record is very good…

After test fitting, I quickly found out that using regular cables was going to be impossible due to the bulk. As I didn’t want to lose any precious real estate, I made my own right angle connector (good ones are surprisingly hard to find).

Burned myself AGAIN. This time with Hot Glue. I never imagined that hot glue guns got hot enough to instantaneously burn you, but there’s a first for everything, I guess.

Checking my screen again after everything was crammed together and test fitted.

Another shot before removing the PSU input. You can see the PCB is tilted significantly. I can’t fit the PSU cable next to the 60mm fan I shoehorned in. You’ll also notice the heat sinks on the PSU changed. I removed the outer case due to space limitations and added on my own heatsinks. Since I’m drawing (hopefully) ~<1/3 of the maximum current rating, I don’t expect anything to get too hot. I removed the Ground Loop Isolator that is required when coupling sound to the Monsoon amp. My Sound card is wedged under the PSU.

Another twitter joke: I never regret paying good money for a really excellent pair of strippers. The stripmaster is truly a godsend for anyone who works a lot with electronics. I almost impulse bought a 22-30g stripmaster to complement this one, as I generally work with much thinner gauge wires than I do in this project, but I’ll just wait.

Test fitting into the car. Cables in the back were a pain in the ass.

Shot #2, it boots!

This is just a shot to give an idea of how close quarters everything is. The 60mm fan was used because there’s a USB cable directly under it, and there is just barely enough space for the 60mm fan thickness. If I had opted for an 80mm fan, I would have lost (an extremely valuable) USB port, and I would ahve had to move everything away from the wall of the case, and that would have encroached upon the other side.

You can see where I removed the input cabling. This allowed me to avoid the PCB bending of the previous solutionThis shot shows the cabling nightmare. There’s a ton squished down into the case that you can’t really see, either. I spent many hours just starting at everything, trying to picture how it would all fit in.

Assembled shot, ready to drop into the car.

Cabling mess.

I bought a hole saw, but forgot the mandrel, so I ended up just drilling a circle outline with my hand drill. After punching it out, I was OK with the results.

Same thing happened here, except it was much sloppier.

RCA out to go to the stereo adapter wiring harness. All the cable ran to the back is not even needed at this moment in time.

Free space shot.

Another top down shot.


Hope you enjoyed the picture dumb, I’m propbably going to sell off a lot of junk, simplify my life, and then move on to microcontroller projects!



Jan 08

Car PC Worklog Update

I decided I’d probably just post incremental updates instead of long “post-mortem” work logs after everything’s done. I’m pretty bad about documenting the process as I go, so this might work, it might not…

For any CarPC, there are a few main subsystems:

  1. LCD/Monitor and bezel (molding)
  2. Core Computer Hardware
  3. Peripherals (sound cards, cams, GPS, etc)
  4. Audio (including car speakers, amps, etc)

This post is largely about the Monitor assembly (specific to my car). Back when I started researching Car PCs (>5 years ago), there weren’t many good commercial(ish) products dedicated to mobile computing (“Infotainment”). That’s all changed, though. I determined that the best use of my time would be investing in a Bybyte Double-Din LCD frame specifically designed for the Monitor I was using (Lilliput 669GL w/ HDMI). After receiving the unit, I quickly found out (by an eyeballed test fit) that my head unit was actually larger than a standard DoubleDin. I ended up buying a Metra dash-kit for my Golf, and assembled it (with some support by the wonderful folks of the Bybyte team). Some assembly notes follow:

I bought a DIY/disassembled Double-DIN kit on ebay via mo-co-so. After my first one came in with some imperfections/blemishes, mocoso support shipped me a new one free of charge. EXCELLENT customer service, I would not hesitate to recommend them to anyone looking for CarPC accessories. This photo shows my impromptu binder clip “clamps” while assembling the mounting tabs.

The disassembled Lilliput HB 669GL HDMI and assembled frame.

Assembly process, routing cables. Hiding the panel control buttons. Well designed kit by Bybyte, no risk of dangling cables that will eventually get broken off. Only slight here is that there is no physical access to the buttons. This isn’t really important as there’s a remote, but to enable “auto on” functionality upon powerup, you need access to the buttons at least once.

Assembled kit.

This is a shot demonstrating how the Metra dash trim kit fits over the Bybyte DoubleDIN enclosure.


How the kit looks when it is actually lined up. The step between the Bybyte and Metra is pretty ugly, but I’m not going to invest any time (at the moment) to clean it up and/or custom fab a better solution. Maybe if it bothers me in the future.

Splitting the enclosures up

This photo shows the modification required to get the Bybyte DDIN kit to fit with the Metra trim kit. Just cut off the bottom shelf.

This shows the one major issue I had during assembly. The DDIN kit mounting holes do NOT line up with the Metra dash kit when they’re centered. I recommend removing the bottom shelf, inserting the LCD frame, and THEN mounting and gluing the mounting arms. They will NOT be centered. I ended up removing some material with a dremel to get the mounting screws to go in.

A quick test fit of my NUC behind the mounted display. My initial thought is to remove the NUC from the enclosure and build a fan system for it as well as the PSU.

Showing the minimal clearance.

That’s it for now. Stay tuned for the next phase, which is cabling my car. Fabrication for PC/Audio/PSU mounting will probably come shortly after that. Once that’s through, it’ll be the test run/test fit.

Nov 26

Car PC – Quick Introduction

Before things get too interesting, I wanted to make a quick note about my next “major” project. I’ll be building a Car PC in the next few weeks/months —  Most of the worklog posts on the blog will be Car PC related. For those of you unfamiliar with Car Computing in general, the forums at mp3car.com is an excellent resource. “Back in the day,” Car PCs were hacked together with various PC components, and interfaced to the car via the stereo. User input is generally done through a 7-8″ touch screen and maybe a small handheld wireless keyboard. The shape and size of these PCs were extremely variable, but in more recent history, the form factor that’s most commonly adopted is miniITX. This is a motherboard roughly the size of a common kitchen napkin. Pretty small, right? Well, the CarPC I am looking to design ideally will fit entirely inside the dash/center console, so it’s actually too big. In a very recent development, my project (that I’ve been putting off for a few years now) has become a lot more simple.

Former Challenges

One of the reasons I’ve put off this project for so long is the complexity of the build. I’ve got a spare radio cage for my 2003 VW Golf TDi, and I’ve been doing some very loose test-fitting of the hardware I’ve got on hand set aside for this build. It is EXTREMELY tight, and to do this project right would leave me with only millimeters to spare. Most people locate their PCs in the trunk of their vehicles, but I wanted to have everything in the dash as one of my design goals. The Power Supply Unit (PSU) was the other major design consideration I had to make — while my motherboard is plenty powerful for a car PC, it is low profile and low power consuming, but confined in the dash of a PC, it’s likely I’d run into heat issues after a long drive. Enter the Intel NUC.

Other Design Considerations

Before discussing the NUC too much, I want to touch on the goals of my build, and why/how the NUC simplifies my life. In a sentence, the NUC is a modern/powerful PC with hardware competitive with a MacBook Air in a smaller form factor than a Mini ITX motherboard.

Things that I plan to do with my Car PC

  • Needs to be self contained in a Double DIN shelf
  • Needs to be easily removed for servicing and/or adding media
  • Bluetooth Capability – Data sharing, calls through the car stereo, etc
  • GPS navigation
  • Optical drive input (optional)
  • Dash/Backup Cams
  • USB hub/input (ease of media transfer, charging, etc)
  • Mobile Internet
  • Mobile Internet Radio

The Intel NUC – Fixing my Car PC design issues

The biggest challenge I faced with my miniITX build was simply size limitations. A mini ITX motherboard will BARELY fit in a Double DIN radio cage. Once you add in the touch screen (even stripped down), you’ll run out of space in a jiffy. The way I would have had to get around this would have been to expand the cage, but the only space that was spare in the cage was the cavity for the old-style VW cupholders. This wasn’t a viable option, as I was planning to mount a slot-load slimline DVD drive in that slot for optical media. While the optical drive will only be used on rare occasions, it’s one of those things that I’m not really willing to sacrifice on. Maybe I’ll change my mind about that in the future.

When I initially read about the Intel NUC, I thought “that’s a cool device, but I’m not sure what I could possibly do with it…” After coming back and deciding that my Car PC was going to be my next major item to focus my attention on, it quickly became apparent to me that the NUC was going to be my next major purchase. The only downside to this is that the NUC has a fairly limited (but awesome) I/O available to it. The NUC has a few USB ports, and depending on the model, 2 HDMI ports w/ GbEthernet, OR 1 HDMI, 1 Thunderbolt, and a few USB ports. Coupled with a low-voltage Ivy Bridge i3 with reasonably good integrated graphics, the NUC already crushes the hardware I had set aside for my build (A Dothan style Pentium M). The NUC also has mSATA and mPCIe slots on it for an integrated SSD and wifi module, and all this incredible hardware is in a 10cm^2 (4″x4″) package.  Since the IO (specifically video) is all digital, I am forced to buy a new screen. The touch screen monitor I already had set aside for this project was a standard VGA input. SUCKS.

While there’s a little bit of growing pains with shifting directions on this project, I think the actual build and test time for the NUC based Car PC will be extremely simplified and shortened. I think what was once an enormous project became an extremely doable project. The cost easily tripled, but I’d rather spend less time to get a usable product and move onto something else. After the Car PC, I think I’ll be posting a lot more microcontroller based work 🙂

Until next time (with hardware pictures!), JD

Nov 15

Manufacturing PCBs at Home (PWM Stirplate Example)

NOTE: This is a sort-of continuation of the previous blog.

In any home tinkerer’s lab, you’ll probably find a lot of breadboards/protoboards and miscellaneous prototyping circuits. As a hardware hacker, once you’ve proven a design works, it’s generally worth the time to build something a little more reliable and/or presentable. Thus: Homemade PCB manufacturing. I’ve gotten a few under my belt at this point, and I won’t repeat the gobs of information that is already out there. If you need any information about the general procedure of PCB manufacturing, just do a quick google search (This process is generally referred to as the “Toner Transfer Method”).

I’m going to go through the example of manufacturing (and the challenges) the PWM PCB in my homebrewing related blog entry. This entry serves to improve the reliability and consistency (manufacturability) of homemade PCBs, and personal challenges I faced building my latest PCB. This is a relatively low complexity circuit, so take it for what it’s worth.

Materials/Tools Required:

  • Copper-Clad fiberglass board
  • Laminator/(Clothes) Iron
  • Tape
  • Photo paper/Magazine pages
  • Eagle software
  • Schematic/proven circuit
  • BOM/components (measurements)
  • Calipers
  • Small drill (press)
  • Carbide (opt) drill bits
  • Water/Vinegar
  • Steel wool/scrubbing sponge
  • Muriatic/Hydrochloric Acid
  • Hydrogen Peroxide
  • Tupperware container
  • Nitrile/latex gloves
  • Plastic utensil
  • Table Salt
  • Common tools (Soldering iron, drill, etc)

Pictoral Description/Misc Notes:

  1. Step 1: Design a schematic, Lay it out in Eagle (easier said than done).
    MAKE SURE THE ORIENTATION IS CORRECT. I etched a few boards that were useless due to mirroring issues. When you work in Eagle, if you work in SMD, just route it (and flip parts) to the opposite side. Print non-mirrored if this is the case.
  2. Tile (optional) and print out the laid out circuit with a laser printer. A major caution: Not all laser printers are created equal. I own a Brother laser printer and a Dell printer. The Brother is significantly cheaper to operate, but the toner sticks to all types of paper I’ve tried much better than the Dell. In PCB manufacturing, this is a nightmare. I’m not sure if the laminator I used wasn’t getting hot enough, or what, but I will be continuing to use my Dell printer here. Note: At the time of writing, I am using OEM toner in both. If I run out of toner in my Brother printer, I’ll try again with 3rd party toner and update this post if it works well.

Note: Paper choice is a tough call. I’ve used magazine paper in the past with some success. Photo paper (and other options) are also popular. I’ve been using Photo paper with great results, but none of the options I’ve tried worked well with my Brother printer.

  1. Cut out the printout, tape down to the copper clad. It is a good idea to shine the copper clad and ensure the surface is clean before attempting to transfer the toner.

  1. Traditionally, hardware hackers use a clothes iron to get the toner hot enough to melt and transfer to the copper. In recent years, many DIYers have switched to using inexpensive laminators to reduce the effort and consistency in toner transfer success. Clothes irons (especially cheap ones) have generally poor control over heat application and pressure. Laminators have a fixed thickness and roll down on the board with decent pressure. I picked up a GBC brand laminator called the “Inspire.” It looks cosmetically Identical to the H220, which is a highly recommended laminator. Pass the stencil and board through the laminator several times, in multiple orientations.

  1. Post lamination/iron, remove tape. Use care not to peel off the paper too violently.
  2. Dunk the board in some warm water, and let soak for 5-10 minutes. Don’t go short on soak time. Any residual paper fibers will loosen up, and you can rub any excess material off with your thumb.THIS IS IMPORTANT. Any paper residue will slow down your etch (if not outright prevent it!)

(Final “built out” board)

  1. Once you’ve gotten the board cleaned up, it’s time to etch. The traditional approach is to use a Ferric Chloride formula, but Ferric Chloride is very harsh chemically, is difficult to dispose, and is hard to find. A popular new etchant solution is available using common household materials. That, and I didn’t want to get FeCl matter all over my hands. (Engineer/Chemist/Wordplay joke woo!)
    Solution: 2 parts Hydrogen Peroxide, 1 part Muriatic Acid (diluted Hydrochloric Acid). You can find Muriatic acid at pretty much any hardware store, it is commonly used to etch concrete. Table salt (supposedly) helps to refresh the process if it slows down. I saw some activity when adding salt, but I’m not sure it makes a big difference.

Stir/keep the etchant moving for vest results. Keep flipping the PCB in order to get both sides exposed/etched.

  1. “Final” product. If you want, you can “tin” the board using solder. This is advantageous in that it will minimize oxidation.

  1. Once you’ve finalized your etch, and checked out all connections, made sure there was no under/overetching, it is time to drill. You can try doing this with a hand drill, but at the diameter sizes of common PCB holes, it’s safer to use a drill press. I’ve seen good results with dremel tools, but I’ve also heard that a lot of drill bits will break due to low precision and tolerance of the “chucks” of the dremel. I considered picking up a cheap drill press from Harbor freight, but it would require another chuck, which would double my overall cost. I found a blog that recommended “Proxxon” tools, so I just made the investment, and picked up a rotary tool, power supply, and drill press. I purchased the version with the adjustable chuck, but I have a feeling it is lower precision than the fixed versions. Prior to this, I had to drive home (to my Parent’s place) and use our old/gigantic milling machine. I would recommend the Proxxon tool to anyone looking at similar solutions, but I’m not convinced it is as high precision as some make it out to be. The drill press attachment is passable.

Above is my awesome dad helping me with the Milling machine. If you require high precision drilling, I don’t think you’ll get much better than a solid drill press with a high precision chuck. A milling machine is a bit over kill…

  1. Picture of a “finished” board.

Note: This is one of MANY revisions of my boards. I don’t think this one actually made it to final production…

  1. Populate boards. I laid these boards out with dual configurations, there are a few versions pictured below. The final version allows for mostly all through hole components, as well as surface mount (my preference). I will probably be releasing primarily SMD/SMT layouts in the future unless there’s significant demand to release through hole. Through hole requires drilling, which is a pain in the ass, but it’s more friendly for people who aren’t as good with a soldering iron.

  1. Case work: Depending on your enclosure, your final setup will probably look incredibly different than mine. If you are “lazy,” and/or want a simple final product, get a large enclosure and use a full fan. 120mm is best, 80mm is fine, 60mm rotates pretty quickly by default, so I would avoid anything 60mm or smaller. If you keep the fan casing (and fan blades), you can just drill and mount directly to your fan. I took a small piece of PVC piping from home depot to make up the difference in the height to my case, and glued a few rare earth magnets (taken from a laptop hard drive) to the top. It’s wise to offset the magnets slightly from the fan hub, as there is a magnet directly below the hub of the fan, and it can cause some problems if your magnets are too close. I’d recommend getting a 1” PVC cap end as depicted below to solve this problem. If you have to offset the fan (raise it up), use some standoffs or spacers/washers to adjust the height. Get as close to the top of your (preferably rigid, and preferably flat) case lid as you can. I’ve also read that using metal enclosures can cause problems, and that the stir plate flat out won’t work, but I haven’t confirmed this throughfirst hand experience. Plastic Hammond cases from digikey or your local electronics surplus store work great for this application.

I superglued two magnets to the top of the PVC spacer. Some people will take a larger single magnet and glue it directly to the top. This works as well, just be aware that most high-powered magnets (from hard drives) tend to be curved, so you may have a hard time finding a good “center of gravity” of the magnet. The nut on top of the magnets was to keep them spaced while the glue was drying.

The “legs” of the fan that were attached to the fan housing were chopped off because my enclosure was smaller than 120mm, and the fan was broken to begin with. Perfect repurposing opportunityJ I superglued nuts in these elbows for mounting to the top of the lid.

  1. A few finishing touches: I used a larger drill bit to scoop out some material and “counter sink” the drilled holes that I created for the top mounted spacing bolts.

Test fitting the board:

  1. Final Product: I recommend leaving long fan leads (if you have removed the fan blades). The reason I left my leads so long was that I don’t have an external potentiometer, so having the ability to remove the lid and make micro adjustments to speed while the fan was still on was a bonus. If you want a clean look, just keep the leads short, or don’t use a clear enclosure.

  1. Circuit in action:

Feel free to contact me if you have any questions/concerns/criticisms, etc. If it needs addressing, I’ll update this post 🙂

Nov 14

Building a DIY PWM Driven Stir Plate for Homebrewing (geek edition)

Disclaimer: Since the tone of this blog is still ill-defined, this isn’t a technical discussion of PWM circuits, nor PCB manufacturing, nor is it strictly a homebrew entry… it’s sort of an awkward in-between.

Introduction/Purpose/Background/Problem Statement (skip this if you don’t care)
This is a two-part blog post. The first blog talks about the concept and theory, and releases DIY build plans for the PWM circuit. The second post talks about the (home) manufacturing of the released files.

As I said before, there’s going to be a lot of nerdy interests that cross-pollinate in my blog, and today, an example of this manifests itself in a DIY (overkill, perhaps) stir plate for homebrewing.

One of the many challenges of homebrewing is having the ability to pitch a healthy yeast culture into your pre-fermented solution (called must [wines] or wort [beer]). An extremely simple summary of the fermentation/brewing process is: Create a solution with sugar in it, and add yeast. The yeast consumes the sugars and releases alcohol as a byproduct. The reality of this simple formula is that if you don’t have a good population of yeast that is added (“pitched”) into your solution, the yeast can become over-worked/stressed, and you may find some “off-flavors” in your final product. The other obvious benefit is that when you have more yeast, your fermentation takes off faster! Thus, many homebrewers [especially those that are brewing high gravity/high-alcohol brews] will make a “yeast-starter.” All a yeast starter does is promote a healthy yeast colony prior to pitching into your solution. This is accomplished by providing an easily fermentable solution in a good environment for yeast. This implies an oxygen/sugar rich environment.

Solution (no pun intended)
Thus enters the “stir plate.” If any of you have taken basic chemistry, you’ve probably worked with one before. You set a beaker or container of liquid on a magnetic platform, and drop a magnetic (generally teflon coated) “stir bar” into the beaker, and the “stir plate” will spin the bar around using magnetic coupling. In a homebrew setting, the stir plate allows for higher oxygen permeation in the solution, which will help “jump start” your yeast starter. Mead, or honey-wine is notorious for slow starts in brewing, largely due to the lack of natural yeast nutrient in the must. I had a mead brewing last year for roughly 9 months, and I still had to stabilize before bottling. I told myself I wouldn’t start another without going through the process with a hearty yeast starter. I’ve got two gallons of honey on my counter, so I figured it’s time to get going…

Design considerations
In a DIY/homebrew setting, it makes little sense to buy a commercial stir plate. A quick Amazon search shows a price range of ~$80-250. The first goal of this project is to beat the commercial price. Many DIY solutions use a computer case fan with a strong magnet glued to it instead of electromagnets, which cuts costs significantly. With this major consideration out of the way, the only real problem left to solve is speed control. There are three (in my opinion) solutions to the problem, all with their respective Pro/Con lists.

  • Solution 1: Simple Potentiometer circuit
    The Potentiometer (sometimes referred to as a “Rheostat”) is simply an adjustable resistor. We abuse this property and essentially create a variable “voltage-divider” circuit <LINK>. At the output of the circuit, this just provides an adjustable voltage output.
  • Solution 2: LM317 (or other variable voltage regulator) circuit.
    The LM317 is designed to be used as a voltage regulator. You can tune the output with a few passive components. Since you can change the output on the fly with a potentiometer, we use the pot to create a variable/controllable voltage regulator to use instead of the potentiometer.
  • Solution 3: Pulse-Width-Modulation (PWM) circuits.
    A PWM circuit basically operates by pulsing full power to the output in varying widths per time period. This is most easily explained with a graphic <LINK>. As you vary the output, the full power is applied to the circuit in varying pulses. At 100% on, the circuit remains powered all the time, at 50%, the circuit remains powered about half the time, and so forth. This is used commonly in controlling LED brightness, and of course, to control the speed of fans.

For a hardware “hack,” any of these solutions is sufficient, depending on your needs. They will all provide some sort of speed control to a fan, and thus, your stir plate. One major consideration in a DIY stir plate design is that lower speed control is generally considered to be pretty desirable. This is to prevent violent torque on the stir bar, which can cause the stir bar to be “thrown” out of its rotation origin. One problem in the inherent design of a case fan is that there is a hard “cutoff” voltage where the fan ceases spinning. The potential applied to the fan motor cannot overcome the resistance/friction to torquing the fan, thus, extremely low speed control is not possible ONLY by lowering the voltage. The first two circuits (potentiometer/LM317) are capable of regulating/outputting low voltages, but if a case fan can’t use them… well, you’re SOL.

The other disadvantage to both the LM317 and potentiometer circuit(s) is that there is considerable power “wasted” due to the design of these two circuits. Much of the power used is translated to heat via resistance and/or powering the regulator. Finally, the biggest disadvantage to the simple potentiometer design is that it has a very weak load driving capability (with a standard Pot). The LM317 will be able to power a much larger load (read as: daisy chained fans/stirplates/etc).

The power and drive capabilities in this context aren’t really that important, but are worth mentioning. To me, the major differentiator of the PWM circuit is that it is capable of driving the fan at a very low rotational speed. This is generally desired in the stir plate setting. While it’s pretty “cool” to get a very tall vortex going in a solution, this isn’t required (nor do you benefit from it very much) when making a yeast starter. A slow swirl is all you really need (and possibly want).

Problems with the PWM circuit
The obvious downsides of the PWM circuit is: complexity/cost(/noise). There are more parts required, and this often means the hack-oriented DIYer shies away from building one. I’ll be providing free schematics, board files, and a straightforward Bill-of-Materials (BOM) at the end of this blog entry. This drives up the cost of the PWM by a few cents (dollars?). The other problem with PWM that people experience is that you might hear a low humming, or “knocking” noise while the circuit is in operation. This is due to the (relatively) sharp waveform edges. You are actually hearing the fan “knock” as it is turned on and off rapidly. The simple solution for this is to put a capacitor in parallel with the output transistor. This smooths out the characteristic of the square-ish waveform.

Summary: If you are looking for cheap and dirty, skip the PWM, the potentiometer will probably work in an “acceptable” fashion for you. I run my stir plate at a fixed speed for the most part anyways, if you can get it down to running in the ~5-7V range, good on ya. You can always gut a larger diameter fan for lower rotational speeds, too.

Since this IS an “overkill” PWM circuit, I will be making a separate blog post about the build process and PCB manufacturing. (Next)

Design consideration note: While I’ll be releasing the PCB, I should note that I personally don’t actually care about on the fly speed control. The Potentiometer on my board (which controls duty cycle) is mounted internal to the case. This implies that I am making a fixed-speed board for myself. You can wire a case mounted Pot if you desire.

DISCLAIMER NOTE: As this isn’t strictly an engineering/homebrew/geeky manufacturing blog, there’s a lot of details missing in the overall theory of operation and/or brewing. Please visit the links in the references section if you are curious about filling in the gaps.

The continuation of this blog post will be talking about the manufacturing of the circuit.

Release notes:

Attached .BRD/.SCH/Extremely sloppy BOM. Note that these files are more for self-reference than anything else. If you are actually looking to build this, feel free to drop me a line, I can clean these files up if they’re of any interest to anyone.


Rough Bill Of Materials:

  1. C1,2,3: .22uF – ~$0.08 ea – QTY 3
  2. B3F Switch: $0.35 – QTY 1 (OPTIONAL –> DEBUG)
  3. DC Power Jack: $1.11 (Choose your favorite, foot print is of common board mount)
  4. 555 timer (SO-8 OR DIP-8) – $0.40-0.55
  5. Potentiometer/Trimmer: 100K OHM –  $1.63
  6. Rectifier Diode(s): 1N5819 – $0.53 ea – QTY 2
  7. Resistor: 10K Ohm – $0.10
  8. Power Mosfet: FD 3055 (or compatible. I just had this lying around) – $0.78
  9. Smoothing Capacitor: 100-470uF – Tantalum SMD or any small pitch cap. – $0.50
  10. Fan header: PC Fan Compatible (salvage or buy) – $0.25
  11. Copper Clad PCB OR Protoboard: ~$3.00
  12. PC Fan: Free – $5.00
  13. Enclosure:Free – $10TOTAL

    : ~25 + shipping (worst case). Mine cost me roughly … $0.00, but I’ve got a lot of spare parts lying around…

Yeast Starter Links:



Pulse Width Modulation – PWM