Fixing a failed fan and CMOS battery on an Intel NUC7

I have an Intel NUC NUC7i5BNK I use for a home lab that recently started acting a little strange. The NUC is located in a closet so when I went to check on it I noticed the case was pretty hot, so I took it to my desk to check it out.

Turns out there were a couple of issues. First off, I noticed that even though it was blazing hot, the BIOS showed the fan speed at 0 RPM. Second, it was complaining of a CMOS checksum error. Since the device is over 4 years old, it’s reasonable to assume the coin cell battery for the CMOS could be faulty so I decided to check that first as… maybe there’s a correlation with the fan? I don’t know but it was easy enough to check. I found Ifixit’s tear down of a similar NUC to be of help.

CMOS Battery Issues

With the multimeter out, I found that the voltage on the battery was showing close to nil (~100 mV). The bad part about these batteries is that while they’re relatively cheap, Intel has made the decision to spot-weld the terminals to the battery so to get a replacement you end up spending $10+ instead of, maybe $0.50.

To fix this, I simple ripped the tabs off the old battery and then, using a new battery I found some heatshrink tubing that was almost too tight, and I place the black lead on the negative side of the battery, ensuring that I did not ground out the positive portion of the battery and did the same for the red terminal (to the positive side) ensuring I didn’t short out this lead to the negative portion of the battery. I slid the parts into the heat shrink and again checked to ensure both terminals were where they were supposed to be.

Using my heat gun, I shank the 3:1 heat shrink until it was as tight as possible.

I attached the new battery to the NUC, booted it, set the time and saved the BIOS settings, let it reboot, then I shut it down.

I removed it from power for 5 minutes and then tested the boot. It seemed to fix the CMOS issues however fan was still at 0 RPM.

The Fan

BIOS Settings

I manually set the FAN parameters duty cycle to 100% so that the fan should always be on. Saved the settings and rebooted, and there was no change.

Fan Removal

I tried to spin the fan, and while it did move, because it’s a brushless motor there was some resistance so it’s hard to tell if it’s okay or not. I decided to remove the fan by loosening the screws below highlighted in blue to check it further.

Fan Disassembly

I pulled the metal cage off by removing the black screw circled in blue and pried the tabs highlighted in red slightly as I worked the cage off.

I then gently pulled on the fan to separate it from the coil.

Fan Cleaning

Using compressed air and a brush I gently cleaned the fan blades. Then, using some isopropyl-alcohol (IPA) and a lint-free wipe, I cleaned the spindle of the fan (purple) several times and then finally using a dry lint-free wipe I removed any excess IPA. I then applied a very small amount of synthetic grease to the spindle.

Using another lint-free wipe and IPA I cleaned the housing of the fan, being careful to stay out of the spindle hole (yellow) as I thought trying to get debris out of here would be difficult. It seemed clean as-is.

I re-assembled everything in reverse order, before I put the metal cage back on the fan I bent the tabs (red) inward slightly to ensure they gripped the corresponding tabs on the fan body securely. I reattached the screws (blue), and reattached to the heatsink on the NUC motherboard.

Testing

While having the motherboard still out of the case, I booted the system and went into the BIOS. I immediately heard and saw the fan moving and BIOS was reporting 5000 RPM, which was great. The fan sounded “normal” and the temperature of the CPU was at 40c. I let this run for several minutes then I power cycled the system several times to ensure the fan was operating as it should.

I then reassembled everything back into the case and reset the cooling settings back to the “Cool” profile so the fan would cycle instead of running at 100% all the time.

So, all-in-all a success and not to terribly difficult. And I saved $50 or so in the cost of a new fan and a replacement CMOS battery.

Helpful Links

• Intel NUC5i7RYH Disassembly

• Cooling and Fan Controls in Intel® NUC Products with Aptio V BIO

• Monitoring fan speed on a NUC

Converting a Carro Home DC Fan to ESPHome

I’ve always liked the Haiku fans, several features of these fans drew me to them:

• Brushless DC Motors (quiet and fine-grained speeds)

• WiFi Connectivity (for home automation)

• Modern design (doesn’t look ridiculous)

• LED Lighting

But, the price has scared me off. And in this case, it seems that procrastination has paid off. Getting the bug again for a ceiling fan I started looking again and found a few different offerings for brushless DC fans with wifi connectivity.

That being said, I’m not a fan of having proprietary cloud based control in our home automation. We do use Alexa for convenience, but I always want to be able to have off-line and local control of devices. Of course, every fan has it’s own app and my experiences with these types of apps have always left me disappointed and wanting.

Of the fans I’ve found, one manufacturer kept catching my eye “Carro Home”. Doing some research it seems as though they had a Kickstarter a few years back and I noticed that they were using the “Tuya” smart home solution.

A lot of the Tuya devices can be flashed with Tasmota or ESPHome (or any custom code you want) but it’s not guaranteed. In fact, the newer devices use a new chipset which is incompatible with the previous methods.

Searching online, I found no-one confirming that custom firmware could be installed on these fans so I decided to take the risk and order a fan.

After receiving the fan, I tore into it and then discovered my first setback, the wrong Tuya module.

The WBR3 Tuya Module - Public Enemy #1

That’s the bad news, the good news is this module is pin-compatible with the ESP12F module, so one can swap one module out for another. This isn’t for the faint of heart though as it requires de-soldering the old module (surface mount) and re-soldering the new module.

Then, there’s one more point of bad news… How the heck does this thing work? What does the Tuya module do? How does it interact with the rest of the device? For that i turned to a few resources. The Tasmota wiki, the Tuya wiki/developer pages, and finally the most helpful was DigiBlur’s YouTube channel and Discord. Good video of the steps here: https://www.youtube.com/watch?v=d_HpkIiWC3Y

Before I removed anything, I wanted to see if I could probe the device as it was running and see how it worked. It seems that a lot of these Tuya modules just act as a dumb pipe (as DigiBlur put it), basically just relaying commands from the app to a secondary microcontroller (MCU). In doing that, it has a simple serial connection at usually 9600 or 115000 bps.

While I was fairly certain that this was how the device was working, I wanted to know for sure. So I setup a pair of FTDI USB to serial adapters and put the RX lines from the FTDIs to both the TX and RX lines of the Tuya module. I also tied the grounds together between both FTDIs and the Tuya module. As I booted up the Tyua module I was able to see traffic two and from the Tuya and the secondary MCU. I then used the Carro Home app to exercise the different functions and recorded the messages for each function for later use. This also showed me the Tuya ID as well.

Do you expect me to talk? YES! 👉, I actually do…

After that, I then removed the Tuya module from the fan, cleaned up the pads.

WBR3 removed and pads prepped and cleaned.

I prepped an ESP8266-Mod (ESP12F) with a base install of Tasmota for testing.

Flashing tasmota to ESP12

I then verified that Tasmota was running properly on the ESP module and then soldered it to the fan module.

Brain transplant complete

I held my breath and booted up the fan module from it’s DC power supply, and after a few tense seconds I saw the ESP show up on the network and I was able to connect to it.

I then worked on sending the commands i learned via the TuyaMCU commands and verified they did indeed operate the various functions of the fan… Once I was satisfied with that, I decided I wanted to flash ESPHome on the unit (as I was more familiar with that). I had to flash the TasmotaMinimal firmware as an interim step, and then I configured what I thought the ESPHome configuration would be, and flashed that to the ESP module.

Holding my breath again, I waited for the device to appear on the network….and success. All of the functions worked for the most part, I need to still figure out how i want to handle the reverse and breeze functions of the fan, but the ESPHome interface to the TuyaMCU seems a bit easier for me in ESPHome vs Tasmota, could just be my familiarity with it.

Heres a link to my current ESPHome config for this fan: https://gist.github.com/billchurch/2feb716ef30f03df775da81822f59b12

And to Home Depot where you can find the fan I worked with:
https://tinyurl.com/yp44ux4w

Maslow CNC Upgrades To Increase Accuracy

Z-Axis

Depending on whether or not you purchased the Z-Axis option (and you should have) it's still not quite perfect. In my case I'm using the Rigid R2202 router and the Z-Axis option for this model makes use of the existing adjustment knob which works good in theory but there's a lot of slop in it especially when changing direction (going between raise and lower functions).

Here are some things to improve the accuracy and reliability of the Z-Axis significantly.

Smooth and deburr router body and base. Clean, and wax.

1.  Disassemble the router body from the base (1). Clean the mating surfaces looking for any parts of metal-on-metal contact that may exist, file over if necessary. It may also benefit from working the mating surfaces with a maroon/red Scotch-Brite pad.
2. Apply paste wax to the router body and the housing (1) where there would be metal-on-metal contact. This will help to ensure smooth movement. Instead of paste wax. Instead of paste wax, you may also use UHMW tape 3 lines 120deg from each other down the long axis of the router body. https://www.mcmaster.com/#76445A761
3. With the Depth Interlock Lever (6) disengaged from the router body, and the collar handle closed (17) attempt to move the router up and down. If you can not do this while the handle is closed, slightly back out the adjustment screw (20) until it starts to move freely but is not sloppy. Then re-home the router and engage the worm gear latch (6)

Fix sloppy Z lead-screw assembly

Few different options here, this is the route I went based on the research and feedback from forum user arnoldcp

1. Obtain some suitable sleeve bearings these have a 0.5" ID and a 0.625" OD
2. Remove the locking tab (8) and using a step drill, enlarge the hole of each side of the Depth Interlock Lever (6) to 0.635" (5/8") and about 1/8" deep on either side. Be sure to leave some material to act as a shoulder for the sleeve bearing.
3. Insert the sleeve bearings on either side of the interlock lever, test for fit. Ultimately we will epoxy the sleeve bearings to the interlock lever (6).
4. Once satisfied with the fit, and ensuring the locking tab will still work properly, degrease all parts and epoxy the sleeve bearings to the interlock lever. Be sure to remove the locking tab during this operation.
5. Install some shim washers (1 worked perfect for me) between the router body and bottom of worm gear shaft (9). This should eliminate any up/down slop. You'll probably need a couple but it just depends on manufacturing quality at the time. The ID of the washer is 8mm OD probably isn't all that important but 13mm seems fine. Thickness will vary, which is why I used shim washers. You could also sand or file down a standard washer if you have one the right ID/OD.

View fullsize

Secure interlock lever, if you have a drill press it'll probably be better. The material is magnesium though and works really easily. I started my step drill by hand a few revolutions just to get things lined up.

Drill slowly and check your progress often. The step drill will keep the bit centered (which is why I like to use it) but you need to be careful to not bore out the entire hole. We want a place for the sleeve bearing to stop when when we epoxy it in.

Beautiful chips. We could probably go further here but no need.

Test fit the bearing to make sure it fits properly. Then flip it over and repeat the other side.

Good to go, now ready for degreasing (acetone maybe) and epoxy!

Quick test fit before epoxy just to make sure everything still lined up. I would recommend doing this so you can make any corrections. It should be pretty smooth but with a lot less slop than before. Also note that the bearings are not intruding into the channel where the orange mechanical fuse will live.

This part is a bit tricky, showing the finished product. When you epoxy the sleeve bearings the epoxy will want to run everywhere. Be somewhat sparing but make sure you cover all the mating surfaces and wipe up any excess before it has a chance to set. In my case, I had some leak into the channel for the orange mechanical fuse that I had to scrape and file out. Not a huge deal and easy to fix but if you can prevent it even better. I used a toothpick to apply the epoxy in the hole and then around the bearing once it was inserted into the hole than I set it aside for a day.

Testing the mechanical fuse travel. You may need to do some scraping and filing of epoxy if any is blocking it's path.

View fullsize

And the finished product. Note the blood on the sleeve bearing, this is how you know it's good.

Add some pre-load to the Z axis

I did this initially but I'm not running it this way currently. However this is something very simple like rubber bands or a bungee cord. Just need some tension on top of the router. I connected mine around the router base. Not pretty but it worked... I don't think you want a lot of tension here, just a little.

References: 

http://maslowcommunitygarden.org/Inexpensive-add-ons-for-z-axis-slack-imprecision.html
https://forums.maslowcnc.com/t/z-axis-only-moving-the-router-some-of-the-time/133/16
https://cdn2.ridgid.com/resources/media?key=09f9c438-97f3-432b-b1f8-06304f469e11&languageCode=en&countryCode=US&type=document

Terminating CAT6 Cable with The Platinum Tools EZ-RJ PRO-HD

A friend was complaining recently about how working with CAT6 cable is a little more cumbersome than CAT5E has been. Part of that reason is due to the addtion of a cross web in the middle to help keep the pairs where they should be.

Cross section of a CAT6 cable

And if it's quality cable, the conductors themselves will be of a thicker gauge and might be more difficult to insert into the connector. I put together a video showing two different connector styles that can be used, the EZ-RJ45 and the traditional RJ-45.

If you're interested in purchasing the tools used in this video, please consider using my Amazon Associates links below. They help support the site and the channel for coming videos and posts.

EZ-RJ PRO HD Crimp Tool - http://amzn.to/2iu0oWo

EZ-RJ45 CAT5E Connector, 100-Pack - http://amzn.to/2iXNJhJ

Cable Matters CAT6 Bulk Cable - http://amzn.to/2iG8XNI