I admit it. I have been a closet Bitcoin geek for more than two years. Yes, the unfinished portion of my basement is full of GPUs running at full bore 24/7. My wife used to be annoyed by the noise, but that was when Bitcoins were worth $5 each. Now, she asks if we should add more. :)

Anyway, my interactions with Butterfly Labs started in January, 2012 when I ordered two BFL Single FPGA miners from them. And then I waited. And waited. But after a frustratingly long time, my FPGA miners actually arrived, and even better, worked (and have kept working for more than a year).

Fast forward to June, 2012. Butterfly Labs announces they are going to develop and sell the BFL SC Single, an ASIC miner for Bitcoin that will hash at almost 100x the speed of a typical GPU for about the same amount of electricity. Has to be a scam, right? Then they announce that they are going to let current FPGA owners trade up to an ASIC miner and get full credit for the price they paid for their FPGA miner. Oh, and they will also be developing and selling a tiny little USB miner called the Jalapeno for about $150 that will double as a coffee cup warmer. For whatever reason, I decided to place an order to upgrade my two FPGA miners, and what the heck, I ordered one Jalapeno miner also for the fun of it.

I placed my order on the afternoon of June 23, 2012. My order numbers were 1662 and 1666.

And then the waiting. And waiting. I should have learned my lesson with the FPGAs, right?

Imagine my surprise when, two weeks ago, I arrive home to find a notice on my door that the postman had tried to deliver a package from Butterfly Labs that required a signature. You’d think I would have received an email ahead of time letting me know that it had shipped. Nah.

I had heard of a few others getting their Jalapenos. I assumed, then, that this package contained my Jalapeno, and the next day I confirmed my suspicions were correct.

The packaging was decent (you can find unboxing videos on YouTube), and the device was easy to setup. I had already compiled a version of cgminer with support for the Jalapeno and so all I really had to do was plug my Jalapeno in to USB and power and start up the miner. Sadly, my package did not include a PCIx power adapter (as some of the early development units apparently did) and so I am powering mine with the standard power brick.

It’s been running non-stop without any problems now for two weeks. Here are the relevant statistics:

• Hashrate: 5.59 GH/s
• Reject Rate (using stratum/btcguild): 0.57%
• Hardware Error Rate (yay, cgminer): 0.00038%
• Power Usage (while mining): 43.6 watts
• Power Usage (while idle): 29.6 watts

Of special note is the fact that my Jalapeno seems to use more power that most other reports I’ve read from others who have received their early Jalapeno. There seems to be some variation in power usage with early units, but most others were around 30 watts while mine is at 43.6 watts. Interestingly, my idle power usage is 29.6 watts which is also higher than the reports I have seen from others.

I’m not at all sure what the difference is, nor am I going to complain (too much). In the two weeks I have had it, the Jalapeno has earned more than enough BTC to cover the $150 I paid for it. Yes, I paid for it in BTC, but I used BTC that I had purchased that weekend for the purpose, so I choose not to dwell on what would have happened if I bought the BTC and just held on to it until today.

Now, I guess it is time to return to waiting for my BFL SC Singles. Maybe this time, I’ll get a shipment notification in advance so that I know to have someone at the house to sign for the packages…

Obligatory pictures…

It’s been a while since I posted about TrendWeight and wanted to give you a quick update of what’s been happening and what’s on my todo list…

It’s been a busy winter with my primary focus, outside of my day job, has been on staying healthy. As such, I haven’t gotten much done with TrendWeight in a few months. That said, I am probably going to spend a little time doing some work on it in the next month or two and here’s what’s on tap:

Arrows for Lean Mass

Right now, the arrow shown when you are gaining lean mass is red and the arrow for losing lean mass is green. This is backwards of what most people expect, so I’ll be flipping the color scheme when in lean mass mode.

Data Export

I’ll be adding a way to download your data from TrendWeight as a CSV file that can be opened in Excel or any other spreadsheet to do your own analysis or to make your own charts.

Actually, this functionality is already there, but not publicized. If you visit https://trendweight.com/export/ you’ll get a CSV version of your TrendWeight data including Scale and Trend values for total weight as well as fat %, fat mass, and lean mass.

That said, I’ll be adding a button to do this data download to the TrendWeight web site so that it is easier to find in the future.

Account Management

Finally, I’ll be adding a couple minor features: a way to change your email address and a way to completely delete your account if don’t want it anymore. Mostly I am adding these features to reduce the time I spend manually doing these things when people email me asking for help.

Then What?

Well, I’m not sure. There certainly have been many suggestions. Some are significant changes (or additions) to the site. Others are probably doable in a single weekend. But before I decide what I want to do next, I thought it would be interesting to get your feedback. To that end, I created a UserVoice site for TrendWeight where you can add your own suggestions and vote on suggestions that others have already made:

http://support.trendweight.com

I pre-populated the site with most of the suggestions I have received by email over the past few months, but feel free to add your own if you have an idea that is not already there.

Remember that this is a free-time project for me and so I make no promises about when any suggestion from that site will get done. But when I do have some time to work on TrendWeight, I’d just assume spend time working on the things that users want the most.

I just noticed that there is another WiFi scale on the market from iHealth. In fact, it may have been on the market for a while. Actually, they have two scales: a low cost bluetooth only, weight-only scale and a $110 wifi + bluetooth scale that does weight and various other measurements like body fat.

Does anyone have one of these scales? Do you like it?

They appear to have an open API similar to FitBit and Withings and so I could add support to TrendWeight for these devices. I can do the programming with just their API documentation, but I don’t have one of their scales to test with.

If you have one of these scales and would like to help me try it with TrendWeight, let me know by emailing me at support@trendweight.com.

Note: There is a discussion of a new TrendWeight feature eventually in this post, but it is preceded by a bit more of my personal story than I normally share. If you have or know someone that has heart failure or CHF, I encourage you to keep reading even though this is long. If you don’t, you probably won’t care about this new feature and so can skip reading this post if you like.

Sometimes life throws you a curve ball. Late last November, I was diagnosed with heart failure. In a nutshell, that means my heart muscle is weak and doesn’t pump as well as it should. Fortunately, I am currently doing fine (thanks to modern medications).

One aspect of living with heart failure is that you have to weigh yourself every day and keep track of day to day changes. That shouldn’t be hard, right? I already weigh myself every day and keep track of changes, and I have this nice web app that makes it easy. Sort of.

Without getting into too many boring details, let me give you a little background. When a person’s heart gets weaker, a common symptom is that they start retaining water. It was explained to me like this: You heart is not pumping as strongly as it used to, and your kidney notices and thinks the problem may be that there is a shortage of blood, and so your kidney starts keeping more water in your body to help make up for the perceived shortage. The result is that a person will steadily gain weight from additional water over days or weeks. Doctors want to watch for this happening, and so heart failure patients weigh themselves every day looking for sudden or steady increases in weight (which might indicate that their heart failure has gotten worse).

Since I have been weighing myself every day for the past year or so, I have some interesting pre-diagnosis data to look at. Here is my summer / fall:

At the time, I didn’t know what was going on, but I did know or at least strongly suspect that I was gaining water weight and not fat. When you look at just October and November and compare my total weight gain with my change in fat mass, you can clearly see that I was not gaining fat (I was also certain that I was not gaining muscle because at the time I was nearly completely unable to do any kind of exercise):

When I finally got in front of a cardiologist at the end of November, it didn’t take them long to figure out that I was suffering from heart failure. The good news is that with just some medications, all this extra retained water drained off quickly, my symptoms disappeared, and I am back to feeling “normal” and have a normal level of physical activity including daily exercise.

Ok, so let me get back to TrendWeight.

I weigh myself every day, and that data already shows up in TrendWeight automatically, so monitoring my weight for unexpected increases should be easy, right?. But what matters, this time, is the actual scale reading and not the trend. That’s kind of the opposite of the normal TrendWeight philosophy which is all about helping you ignore the actual scale weight and instead focusing on the slower changing trend over time.

When you are trying to lose weight (fat), you want to ignore the day to day changes caused by varying amounts of water retention. But when watching for weight changes in the context of heart failure, it’s the water retention that we’re looking for. The smoothing of day to day weight changes is counter productive, in this case.

I could just look at the normal TrendWeight dashboard, look at the “Actual” column of the recent weight readings table and do math in my head. For example, you can see (to the right) that from Jan 7 to Jan 8, I gained 1.6 lbs. But wait. Manually doing math in my head every day? Isn’t avoiding that kind of thing the main reason I developed TrendWeight in the first place? I can do better than that :)

So, I have been working on a sort of “CHF Dashboard” that uses the same data from WiFi scales, but analyzes it and presents it in a different way. In particular, it doesn’t show the “trend” weight at all, because that is not the important data point. It also automatically calculates daily and weekly weight changes and highlights days where there are higher than normal changes. Per my doctors, what I am looking for is a change of more than 2 pounds in a single day or 5 pounds in a week. If that happens, I call my doctors/nurses and they will likely make adjustments to my medication doses to get things back under control.

The Dashboard

This is still a work in progress, but I have a beta version of the CHF dashboard that I am now using daily (in addition to the normal dashboard, because I am also still trying to lose more fat). If you have CHF and want to try this today, you can view your version of this dashboard by visiting https://trendweight.com/chf/.

The table shows the last 28 days of weight readings and the relative weight changes for that day (1 day and 7 day). As a side note, I expect this will be helpful for my doctors and nurses as well. They usually want to see my weight changes every time I have an appointment with them, and I can show them this dashboard on my iPad instead of keeping track redundantly on a sheet of paper that I give them.

Single day changes of more than 1 pound are highlighted (whether it is an increase or a decrease–they want to know about significant changes either way). Weekly changes of more than 4 pounds are highlighted as well. You might notice that the 7 day column usually has 2 numbers in it. The reason is that my weight today may represent both a gain in weight from the lowest weight I had in the past week while at the same time being a drop in weight from the highest weight I had in the past week. So the table shows both unless today’s weight is the highest or lowest weight in the past week.

I use this dashboard every day to see if my weight today is different enough to need to call the doctor (I am fortunate that that has never happened). In reality, I can usually remember what my weight was yesterday, so I usually know, even when just standing on the scale, if my weight has changed by more than 2 lbs in the past day. But I definitely need this tool in order to identify if my weight has changed by more than 5 pounds in a week.

As I said, this is still a work in progress, and it lacks a certain amount of polish. But if you have a reason to want to track your day to day weight changes, feel free to use it. As with the rest of TrendWeight, I created this primarily for myself, but there are millions people with heart failure in the world, so if this is a useful tool for some part of that community, I’m happy to share it.

As always, if you have feedback or comments, feel free to email me at support@trendweight.com.

I just released a new version of TrendWeight that includes a bunch of changes:

• Body fat tracking
• Decent mobile support for more devices
• Configurable day start
• Access to raw scale readings
• Lots of behind the scenes improvements

Also, I’ll take this opportunity to apologize in advance. This update involved a nearly complete rewrite of the user dashboard (the page with the chart and the stats). I probably broke a lot of things in the process. I have tested the new version a fair amount and fixed a lot of small issues, but I probably didn’t find them all. I’ll be watching the error logs closely for the next few days, but if you notice anything that seems wrong, please email me the details at support@trendweight.com and I’ll look at it right away.

Body Fat

If you have body fat data in your connected Withings or FitBit accounts, you’ll now see some new buttons on your dashboard that let you look at body-fat-related trends. There are three new options:

• Fat %. This is the percent of your total body weight that the scale thinks is fat.
• Fat Mass. This is how many pounds or kilograms of fat you have in your body. This is calculated by multiplying your total body weight by your body fat %. If you are trying to lose weight, watching for this trend line to go down over time may be interesting.
• Lean Mass. This is how many pounds or kilograms of non-fat you have in your body. This is calculated by subtracting your fat mass from your total body weight. If you are actively trying to build muscle, watching this trend line go up over time may be interesting. On the other hand, if you are trying to lose fat, making sure that this trend line is not dropping over time may also be interesting.

Fair warning: In my experience, neither the Withings nor FitBit wifi scales are particularly precise with body fat estimation. What I mean is that your scale readings may change significantly from day to day even though your body hasn’t really changed much at all. The result is that you should expect the body-fat-related charts to be more erratic than the total weight chart. Looking at your body fat trends may still be interesting, but my suggestion is not to worry as much about short term changes. If you care about body fat trends, watch the trends over a slightly longer term than you do for total weight (e.g. look at changes over 3 months and don’t obsess about changes over 2 weeks).

You can also play with this using the Sample Dashboard. Keep in mind that the sample user has entirely fabricated data and the body fat data may not be realistically random. On a side note, if you have at least 9 months of weight and body fat data and are interested in volunteering your weight data to be used (anonymously) by the the sample user dashboard, let me know.

TrendWeight on your SmartPhone

As part of this update, the mobile-optimized dashboard is now available on more devices (almost any device in theory as long as it has a decent browser). I have only tested it on an iPhone, though, since that is all I have available.

Day Start

You can now specify what time of day you want TrendWeight to use when deciding which day a weight reading occurs on. See this post for a more thorough discussion of the motivation for this feature. Most people can probably just leave this set at midnight and not worry about it.

Raw Scale Readings

Several people have noticed that the scale readings displayed in TrendWeight do not match the scale readings displayed on the FitBit website. After some investigation, this seems to be because the two websites round numbers differently. The FitBit and Withings APIs actually return weight readings with many decimal places (e.g. 141.4972 lbs). On TrendWeight when that weight is used in calculations, it is used with the full precision (all the decimal places). When weights are displayed, they are generally rounded to a single decimal place (e.g 141.5 lbs). On the FitBit website, that weight would be displayed as 141.4 lbs though because it always rounds down to the nearest tenth of a pound.

Anyway, in order to allow people to better determine if there are discrepancies between TrendWeight and FitBit/Withings, you can now go to a page which shows the raw scale readings that came from the FitBit/Withings API. You can get to this page by clicking a new button on your settings page:

Again, if you notice anything not working or fishy, please email me at support@trendweight.com.