The other day my Bluetooth heart rate monitor arrived. While I already had a heart rate monitor, it only gives an average and peak reading for the measured time period, and I wanted much more data. This monitor should allow me to use my cellphone to log data.
I've decided I hate most Android applications. They either don't work, or don't do anything useful unless you pay, and almost all of them are plagued with ads. Took me forever to find an application that would simply log the heart rate data to a file. Once I had it though, I could start to get my data.
There are some numbers I really want, and I got two of them pretty quick. First was my resting heart rate. The easiest way to get this is to sleep and log your heart rate.
Here is a chart of my heart rate as I slept. I am a very light sleeper and it is fairly clear how often I wake up and roll over during the night. But based on this data I can see that my resting heart rate is around 57 BPM.
This graph shows the second piece of data I've wanted—heart rate recovery. I biked downtown to get a slice of pizza and recorded by heart rate for the ride. On the ride back I push my heart up to right around 170 BPM (92.9%) and held it there. Here is a graph of the my arrival back home. The time is in hours and minutes. Using the data from this chart I can see I dropped 43 BPM over 2 minutes, which is pretty normal. I've wanted to know this because during roller skating I push my heart rate way up—over 190 BPM (>103.8%) at times. From what I've read this is safe as long as my heart rate recovers quickly and doesn't stay high. I hope to find out this weekend at skating.
There are a lot of data points I want to have, and I will be logging my heart rate 24/7 for a week just to have some data to start with.
Heatsink compound arrived today and it was time to do a little maintenance surgery on the Blue-Dragon. After disconnecting hoards of cables I began work on extracting two heatsinks. The first was the CPU heatsink, which came off fairly easy. The second was the north bridge heatsink which required me to pull up the motherboard in order to have access to the pins that held it in place.
The CPU heatsink compound was white which means it was not silver based. I used a little rubbing alcohol to clean off both the CPU and the heatsink to remove the old heatsink compound. The heatsink itself looked to be in good shape. The fins were clear of dust and I saw no areas of buildup. I washed it off in a strong blast of water and blew it down with compressed air just go remove any fine dust.
The north bridge heatsink was a mess. The heatsink compound used had solidified and cracked. It took some light scrapping on the die in order to chisel it off. On the heatsink I also chiseled off the remaining compound and then used some high grit sandpaper to prepare the surface.
I used a generous amount of silver based heatsink compound on both chips and reinstalled the heatsinks.
Something I noticed when pulling everything apart was how stiff the CPU fan was to spin. When I blow down the system I do not turn it off. 110 PSI of compressed air spins the fans with relative ease so I figured the CPU fan was in working order—especially since I could see it spin on the monitor. But it was clearly resistant meaning the bearings were shot. I oiled up the bearing and used a blast of compressed air to spin the fan for awhile. Afterward the fan spun a lot better. Still, I don't trust the CPU fan and I am going to get a replacement.
After the system was reassembled it was time to give it a test. My threaded prime number counter is a simple program that finds all the prime numbers from 0 to 232. It isn't designed to be the most efficient algorithm—just a good example of running a program over multiple threads. The program uses all cores and provides a 100% CPU load, and therefore does a pretty good job of heating the system up. On the Blue-Dragon this runs for 36.6 minutes. During that time the CPU temperature only got to 117.3°F (47.4°C) and the north bridge 138.2°F (59°C). That is significantly better than the other day when the temperatures were 163°F (73°C) and 185°F (85°C) for the CPU and north bridge. I also noticed my CPU fan was spinning close to 2,000 RPM when before it only was reaching about 1,500 RPM. So I think this should solve my shutdown issues.
The Blue-Dragon has been having issues where it randomly turns off from time to time, and the problem is becoming more frequent. After some investigation, I believe it is because the system is overheating.
Here we see a graph of temperature right as the computer shuts down. Notice how the temperatures spike up before the shutdown. I'm not sure why that is happening, but I'm fairly sure it is the cause of my problems. The BIOS is setup to shutdown on over temperature.
The system was assembled in September of 2012 and has run almost 24/7 since then. Every month I blow the computer down with compressed air to clean out dust and there are always clouds of dust. I also inspect the fans during this blow down to make sure they are all functional. The CPU heatsink is a massive heat-pipe assembly designed for overclocking gammers. The motherboard provides large heatsinks for both the north and south bridge, and the switching power supply for the CPU. My RAM is equipped with with finned heatsinks. And my case contains 7 fans. There should be no issues with heat.
So I did a test and put a 100% load on the CPU. Sure enough the temperatures climbed, and kept climbing. My north bridge reached 185°F (85°C), and the CPU 163°F (73°C) before I stopped my test. These temperatures are above what should ever happen in a 65°F (18°C) basement. Clearly there is something wrong, and runaway temperature is the problem.
I noticed my speed controlled fan was maxed out shortly after the test began, and its speed looked fine. All the other monitored fans held their speeds. So I don't think this is a fan issue.
What I suspect is the problem is old heatsink compound. The heatsinks were installed 4 years ago and perhaps the heatsink compound has degraded. So I ordered some silver based heatsink compound and I will reapply it to the system to see if it helps.