Cox - for the Gigablast loss!!

Wow - Cox just doesn't apparently know how to run a network.  Or customer service.

I've just moved from my old house with Gigabit Fios.  I decided, despite the paltry upload speeds, to get gigabit at my new place.  I was disappointed when I found out that the neighborhood I was moving into, despite it being denser, did not have FiOS.

Several calls to customer service later, multiple resets, and it just starts working.  The only information that I can find about it is that there are issues with MAC address binding.  But of course, not information provided by Cox, just a forum on Reddit.  Super frustrating.  But glad that we have rain in what was our Internet desert.

Dinosaurs!

Not my typical post, just a reminder to my future self that I've recently read the powerful New Yorker article about the day the dinosaurs died.  It is definitely worth a read if you haven't done so already.  The conclusions drawn about the globally catastrophic event are incredible but seemingly well-founded.  Amazing stuff.

Ubiquiti Second Thoughts

Second thoughts not in the way that one normally thinks about that term, but just thinking about the process of having installed a small Ubiquiti configuration in my house and how it's worked out.  As I look through the blog posting history, I don't see any indication that I wrote about the Ubiquiti system in the first place.  So, I guess I should.

What drove my deployment of Ubiquiti was upgrading my FiOS connection to GB at the behest of my boys, who were living with me at the time.  The Verizon-provided router that I was currently using was not going to be capable of 1 GB.  This was after Verizon had lowered GB Internet to reasonable prices (< $100/month) and about March of 2017, right about two years ago.

Our first attempt was to use a new Verizon-provided router for the service.  It could not have been worse.  It kept needing to be reset and otherwise was not terribly reliable.  I researched my available options and ultimately decided that despite the expense, a Ubiquiti-based system would provide a lot of flexibility and provide outstanding coverage across my (not terribly large, but somewhat spread out) house and back yard.  The goal of the system was to have a reliable, but easy-to-manage system that would just work.

I purchased the following:

• Unifi Security Gateway (USG) to be the router
• 2 x Unifi AC AP LR
• Unifi AP Outdoor
• Unifi CloudKey
• 2 x 5 Port EdgeRouter PoE
• Unifi Video Camera G3

After I had made a separate decision to build a simple Linux server which could host the Ubiquiti configuration software, I decided that I could return the CloudKey.  I was also annoyed that the CloudKey was not powerable via PoE provided by the EdgeRouter.  The other thought was that I could use the server to run Ubiquiti's software to run security cameras.  I did get the software working and tested my one camera, but I never ended up deploying the six or so security cameras that I had been planning on doing.

The Unifi AC AP LR (as in 802.11 ac, access point, long range) had to be wired and could be mounted to the ceiling.  I did install one near the router cabinet in the living room as below:

All blue!  Providing super-fast wifi close to most of the usage (in the living room and kitchen)

USG not mounted yet, but you can see the cloud key on the floor.  Server to come soon to replace the box on the left.

Previous outdoor setup relied on a large antenna and some Amplifi gear, but it never lived up to its promise

Here is the new Unifi AP Outdoor all hooked up - but no cable management has been completed

I loved a few things about the network:

• Web-accessible dashboard (both internal and external to our network)
• Single SSID; a seemless transition between access points as you moved across the house
• Overall, quite speedy

Our GB FiOS connection never lived up to the hype, but we do get upwards of 870 Mbps, which is something.  All of the backbone network in the house is also now going through GB ethernet switches, so there is just good speed to be had all around.

I have done very little in terms of network maintenance or fiddling over the last couple of years.  Mostly just updating the firmware on the devices and rebooting them.  I do that probably not more than once per six months.  One of the times that I decided to do that, I found that the outdoor access point and one of the indoor ones were not listed as active.  What was up?

I did not go on a tear to find out.  I was reasonably lazy and took my time.  I still seemed to have wifi where and when I needed it, so there wasn't too much of a need to do something quickly.  Ultimately, I have figured two things out:

• The outdoor AP not working was a result of some animal deciding to chew through the ethernet cable that goes out of the house at the study and back to the device.  Annoying, but also somewhat relieving that it was not anything to do with the device.  I haven't fixed the cable and I'm not sure that I will before I sell the house (which will hopefully happen in the next couple of months).
• The indoor AP was a bit of a surprise.  I thought that it was the AP located in the study that had stopped working and that was likely driven somewhere in the cable routing upstream (of which there was quite a bit).  It turned out not to be true - the AP in the study was the only one that was working.  The AP pictured above in the living room, despite having it's blue ring on, was not working.  I'm still not quite sure why.  I pulled it off the ceiling and connected it into the EdgeRouter with a shorter ethernet cord (into a PoE'd port) and it has been functioning fine since.  I should note that the cable that runs from the EdgeRouter to the ceiling-mounted AP is one that I put the ends on.  So I might not have done a great job.  This is something that I may have to deal with in the future, but also don't want to do much before I sell the house.  Whatever setup I have, will be good enough.

So despite these two quirks, I'm still really happy with the network and am hopeful that I get another three or four years out of the gear.  If I do, I will declare myself satisfied with the purchase and will look into buying another round of Ubiquiti gear.

More Love for the 935

I still love my Garmin ForeRunner 935.  You can read here about my initial love affair, but I'm now 16 months into ownership and I feel the same way.  It continues to be an awesome watch.

Here is what I continue to like:

1. Battery life: always good.  It's never died on me.  Fantastic.
2. Heart Rate Accuracy: seems spot on when wearing my heart rate strap.
3. Ecosystem: It's fairly well developed.  A lot to like about the software that will integrate (like MyFitnessPal, which I just signed up for).  Appears to be able to integrate new items like the Humon Hex (if I get around to buying it) easily.
4. Customizability of Watch Face: I've gotten stuck on the one I like, but it's great to know that I can choose whatever I want and many (most) for free.

Here are new things I like:

1. QuickFit: I just upgraded to the QuickFit band.  Though I am finding myself most typically just wearing the black band in the package of three that I bought on Amazon for $25, I like having the option.  I also have a black metal band for situations where I need to be a bit fancier.  It was not expensive, nor were the non-Garmin plastic bands.  Happy to have the option to pay full price for the Garmin-branded bands or not.
2. OK, maybe it's just one thing.  But I like it a lot.

Here is what I would change about it:

1. HRM Tri: Mine died and never gave me notice that the battery was low.  Took me a while to figure out (arguably I was being stupid).  There should be better messaging here.  Even eventually an email saying, "Hey, we've missed your HRM Tri recently from your workouts.  Is everything OK?"
2. Default Elliptical Activity: It doesn't capture enough information.  Seems obvious that it should have cadence given that you are given credit for the "steps" taken during the workout.  I can buy a better one, but feel like I shouldn't have to.  This is too obvious.
3. Cheaper Power Meters: But, alas, this is not a Garmin issue, just generally an issue.
4. Easier Integration with Treadmills: Yes, many of us don't love the dreadmill, I get it.  That said, why can't Garmin Connect when running on my phone connect via Bluetooth to the treadmill (or to the Life Fitness app) and get my treadmill milage instead of assuming it from my cadence?
5. Better Reporting: There are some genuinely useful reports provided on the Garmin Connect site about yourself.  Having now over a year worth of data,  I can see trends in the data.  But most of that is confirming what I already knew (last summer was a slow, slow time for me from a physical activity standpoint).  My activity counts and calories along with my VO2max all reflect this.  What other reports could be provided that are useful.  Encouragement along the way?  Learn "normal" behavior and encourage user to keep at it?
6. Easier Spin Bike Pairing: I've gotten into a rut when I do spinning at my gym now.  If it is in a class, I'm on bike #23.  If it is on my own, I'm on the bike on the left.  This ensures I don't have to fiddle with pairing issues.  Which are still not dealt with super well, IMHO.  There must be an easier way, especially to avoid unintended pairings when in a class full of bikes with power meters.

This has largely been a data dump of my current thinking.  I am interested to see how it evolves over time!

What Tech I Want, I Think (Early 2019)

It has been a while since I've written about a list of current tech items that have caught my eye.  I thought that it might make sense to given that CES 2019 just ended.

What I Want:

1. Humon Hex: it tracks oxygen availability in your muscles (specifically your thigh).  It provides more direct feedback on how hard you are working (which I'd always wanted to know).  Maybe I'm being lazy or just feel tired when my body is capable of doing more - this gives you a much better sense of how true that is.  Cost: ~$300.  Availability: Immediate.
2. Garmin Vector 3 Power Meter: I have gotten used to having legit power meter information from the Stages spin bikes at the Equinox at which I am a member.  Going for outdoor rides now are not as information rich.  I much prefer the outdoor ride (in the right weather), and already have speed and cadence sensors, so this is the next step.  It's a lot of money, though.  Cost: ~ $900.  Availability: Immediate.
3. K'Watch Glucose: A continuous glucose monitoring watch.  It requires a consumable which needs to be replaced every 7 days, but it looks super cool.  I'd love to wear for a couple of months to see how my activities, eating, drinking, working out and all affect my blood sugar levels.  I feel like I'd learn a lot over the first month or two of wearing it.  Maybe not too much after that.  Cost: ~ $150 for the watch, $100/month for the consumable. Availability: Unknown, but hopefully in 2019.
4. LIDAR Lite v3HP: A small 40m range LIDAR from Garmin.  I would love to work this into a Raspberry Pi 3 project to build a small device that would measure speeds on the road in front of me and then post the fastest on a website along with vehicle pictures (with ALPR, ideally) and a graph of the distribution of the speed of cars on the road.  Maybe someday I'll get to this project.  Cost: $150.  Availability: Immediate.
5. Naked Labs Body Scanner: I know I drink too much beer to keep my gut in check, but I think it would be super cool to get a long term view of where fat is being added and removed as you move through time.  So you should have a much better view of the aggregate body fat, and also the view of where it's coming and going from.  Also, you get your weight.  Cost: $1495.  Availability: Q219, I believe.
6. Air Quality Monitor: I'm not sure that Awair is the one, but I would like something that tracked indoor air quality that was inexpensive enough that I could have several of them in my house.
7. A compact point and shoot camera: a successor to my Canon S100.  Not sure what's best out there now, nor what the right amount of money is.  Sometimes, the camera on my phone just doesn't capture the magic.

What I Don't Want

1. MagicLeap AR Eyeglasses: Not until somebody shows me software that matters.  Cost: $2295.
2. 8K TV: Not until most content is 8k.  I bought my current 65" Sharp LCD 1080p TV in 2008.  It's still pretty great.  When it fails, I will buy a 4k TV of similar size (we'll see about OLED).  In fact, I'm rather hopeful that the Samsung modular micro-LED approach gathers some steam and we get some cool options there.
3. Bike head's-up glasses: Way too much money.  Way too early in the product maturation cycle.  Same for a motorcycle helmet head's up display.
4. USB turn-table: Maybe a cool mix of old and new, but I'm not a vinyl guy.  Spotify for me, right now.
5. Glowing alarm clock: My Garmin Forerunner 935 vibrating works just fine.  I wake up at a different time than my wife, so I think that lighting the whole room up would be disruptive.
6. Any voice-activated appliance: I have an Echo Show and I like it.  I just don't feel the need for a microwave that has this built in.
7. Any robot: call me when they are actually useful for more than vacuuming.
8. HTC Vive Pro Eye: but only because I don't have the space to devote to it right now.  Also, having bought the original Vive, I feel like there isn't enough software for it.

If I think of anything more, I'll add to the list.

Another Dead LED: Time for a Disection!

One of my first generation (for me) Feit "PAR" 30 LEDs just stopped working in my kitchen.  Which isn't a very good way of describing the problem.  It turned on, at super low brightness.  Not like L80 failure.  It all of the sudden went from a very bright bulb, to flickering occasionally, to flickering almost every time it was turned on, to not glowing brightly.  The bulb in question is about seven years old, but still disappointing that it is going.  Many other of these bulbs in my house are still going strong.

I'm not knowledgable enough to know what likely failed, but I thought that I would take advantage of the failure to see what was on the inside.  I did a destructive disassembly of everything but the circuit board.  I can say definitely that I did not expect the circuit board to be so complicated.  I'm half inclined to open one of my newer, much lighter LED bulbs to see how they've been simplified over time.

In any event, here are the pictures of the disassembly, step by step, with my amateur descriptions of what is there.  Further conclusions (if any) will be at the end.

This bulb is dominated by the aluminum heat sink that surrounds it.

Looking at the business end of the bulb - 7 actual LEDs

After removing the three screws you saw in the first picture, I was able to wiggle the top of the bulb, but couldn't remove it further.  Time to try the other end of the bulb.

The plastic lens popped off without too much effort.  It revealed four screws - three of which were required to remove the plastic insert plate.

Plastic plate removed, exposing the metal plate that has the actual LEDs sitting on it.  Some identification markings from Feit: ALPCB0589 Rev1 which doesn't mean anything to me, or the internet.  The positive and negative wires bringing the (presumably) DC power required to make the LED's go bright are exposed.  I had to cut them to keep going.

Now I was able to remove the aluminum heat sink from the bulb.  Here was my first surprise: the aluminum heat sink is only about half the weight of the bulb.  I had assumed it was the heavy part.  The core turns out to be pretty heavy too. 

To keep going, I had to start breaking the center core.  It was a hard, but reasonably brittle plastic that allowed me to dig down a ways.  I'd eventually get stuck just using my Nest screwdriver and had to bring in more destructive tools.

Just some perspective of the working part of the bulb - it is tiny, a super thin plate.

Maybe the plate is not metal: it looks like there is a super simple (in series) circuit on the surface?  Why would they choose to do it that way instead of in parallel?  Seems odd.

I've broken all of the inner core that I could with the screwdriver.  Inside there is silicon or some such substance around everything.  The circuit board is already way more complicated than I had expected it to be.

The other side.

Using some vice grips, I pulled off the E26 connector.

The vice grips were also useful to crack off the remaining plastic.  After that it was just clearing out the silicon.

Given the simplicity of the back of the circuit board, it came off in one easy piece.

Here is the top of the circuit board as much as I cleaned it off -- took way too much time even though you can see a ton of silicon still on it.  The right is the 120VAC input, the right is the DC (IDK what voltage) output.  I would guess that the exposed copper coils are a transformer bringing the AC voltage down.  I recognize some resistors and some capacitors elsewhere, but not sure what may of the other pieces are (like the three big brown things on the top and top right of the board).  I'm also not sure what the yellow covered device is (though presumably, it is an inductor).  In the end, this is almost certainly a constant current power supply shoved into the bulb.  It's just so complicated.

Why a constant current power supply needs multiple chips on the back is a mystery to me.

Here are my conclusions:

1. First-gen LED's were heavy due to both excessively large heat sinks and the silicon goop on the interior.
2. It's easy to remove the aluminum heat sink from Feit bulbs - do it and recycle.
3. New bulbs are not readily accessed like this one.  Likely they avoid both the big heat sink and the silicon goop.  When they start breaking, I'll dig in and see how they are different.
4. LEDs are pretty awesome.  Based on the marking here, I likely bought this bulb in the early part of 2009.  It's the early part of 2019.  I'll take a 10 year replacement cycle.  That being said, I'd gladly take longer.