Thursday, August 21, 2014

New Small Form Factor Desktop Computer - First Impressions

In my April post I listed computer parts that I was considering buying if I were to build a new computer. In August, I pulled the trigger and bought the parts to assemble the new computer. Here is a list of the components I ended up getting:

Motherboard: Gigabyte GA-F2A88XN-WIFI Mini-ITX
CPU: AMD Kaveri A10-7850K APU
CPU cooler: Thermalright AXP-100R
Memory: Patriot Viper 3 16GB kit (8GBx2) DDR3 PC3-17000 2133MHz PVL316G213C1KR
Storage: OCZ Vector 150 Series 240GB SATA III solid state drive VTR150-25SAT3-240G
Optical drive: Panasonic UJ265 slim Blu-ray burner for laptop (slot load)
Case: In Win H Frame Mini Red

The components were ordered from amazon.com and newegg.com

First I take the motherboard and try to mount the CPU. The motherboard comes with the appropriate CPU cooler brackets for the AMD stock cooler. The brackets were removed by removing the 4 mounting screws. I mounted the brackets for the Thermalright AXP-100R, inserted the AMD CPU, apply the thermal paste from the Thermalright cooler package, and secured the AXP-100 cooler heatsink and fan. I plug the fan power connector onto the motherboard. So far so good.

Then I try to install the memory modules. It turns out that the Patriot Viper 3, although deemed low profile compare to many memory modules out there today, would not fit under the AXP-100 in the slot closest to the CPU. It can fit in the slot further away, so right now I have just one module of 8GB installed.

Next I put the motherboard into the H Frame Mini case, plug the ATX power, CPU power, USB, front panel USB, power switch, reset switch, power LED, HDD LED, and three SATA connectors to the motherboard. Everything is in the right place. I installed the solid state drive and optical drive. Fits perfectly.

Plug in the power cord in the back, connect the DVI cable to the motherboard, and press the power button. The power button's blue LED lights up, and the fan on the AXP-100 turns, but no video on the monitor.  I tried the HDMI ports, no video.  I tried resetting the BIOS and unplugging everything except the power switch, still nothing. I thought maybe this motherboard has the old version of BIOS that doesn't support the new Kaveri CPU, so I contacted Gigabyte technical support. They informed me that based on the serial number, it should have the newer BIOS that supports the Kaveri CPU. I checked the memory module and found that it was not completely tight in the socket. After that I got video. I browsed the BIOS settings and everything looks ok; BIOS version is F3. Then I plugged everything back. The BIOS detects the two drives and I can proceed to install Windows 7 Professional 64-bit. I changed the memory speed setting to the XMP setting, and it runs at 2133MHz with 11-11-11-30 timing and 1.5V.

below is the info from CPU-Z:




GPU-Z shows info about the integrated graphics on the APU:

Booting is much faster now with the solid state drive. It took about 20 seconds from power button to Windows 7 login screen.

Tuesday, April 29, 2014

A New Desktop Computer?

Thinking about upgrading my desktop computer these days. The computer I am using now has been working pretty well for the last 6 years. I wonder what are the new product choices I have for a replacement desktop? When I built my current computer in 2008, I wanted a small form factor PC that's quiet and powerful enough for playing computer games. I did not succeed completely. The Raptor X hard drive was loud and it had died after 5 years. The two 60mm back exhaust fans became clunky and one had died after 6 years. The graphics card has a fan that sings in sync with web page scrolling and graphic intensive games. The good part is that CPU cooling is fanless and the computer still works. Although sometimes the computer would reboot itself suddenly in the middle of a task, usually because I am processing video files from the camcorder. Maybe the power supply is getting old or the motherboard components have aged and voltage supply is not quite stable.

computer technologies have advanced since 2008; components are smaller, data storage is cheaper, memory chips have higher capacity, CPUs can do more with lower power dissipation, and operating systems went from 32-bit to 64-bit, so system memory ceiling is raised from 4GB to anywhere between 8GB to 4TB, depending on the operating system version.

What I am looking for with the next build is continue to strive for low noise, small foot-print, and good performance for everyday computing (web browsing and authoring), light office work (word processing, spreadsheets), graphics editing, video editing, and computer games. And strike a good balance in performance per cost (value). With new products out there, it seems like these goals are more achievable than ever before.

When deciding what components to pick for building a personal computer, I start with the CPU, because the CPU is the main computing device and it dictates what motherboard you can choose. Each CPU manufacturer has its own package and corresponding socket on the motherboard to fit the CPU. Next I look at the motherboard, because usually the motherboard determines how many memory module slots are available to insert system memory. The speed of the memory is determined by the CPU because modern CPUs have the memory controller built-in. Once the CPU and motherboard are picked, then I pick the memory modules. Most CPUs require some kind of thermal management, usually in the form of a heat sink and a fan for active convection cooling. The CPU cooling solution affects the computer case, motherboard, and memory selection. Some CPU coolers are large and requires a bigger case. Some coolers take up a large area and limits the height of the memory modules. The motherboard component layout may limit the CPU cooling choices or even get in the way of the air flow. Next is the selection of a computer case and power supply. Usually larger power supplies have higher power ratings, and most power supplies have cooling fans built-in. The power supply need to be able to provide enough power for the whole system including peripheral devices. Some computer cases come with power supply built-in, which can be good or bad; it's convenient and it's one fewer thing to buy, but if it has a noisy fan or does not have enough power, then a separate replacement has to be selected.

CPU

The CPU I am using now is the AMD Athlon 64 X2 "Brisbane" 5000+ BE. This was the first dual core AMD CPU, built on 65nm SOI technology, has 512KB L2 cache per core, uses AM2 socket, and consumes a rated power of 65W. This CPU uses the K8 generation microarchitecture. Since then, AMD released K10 microarchitecture, which was an improved version of K8. K10 devices have up to 2, 3, 4, or 6 processor cores, some with bigger L2 cache, and higher clock rates. Some were fabricated using 45nm SOI technology.

Then AMD designed the Bulldozer microarchitecture from scratch. Piledriver and Steamroller cores followed as incremental improvements on the same microarchitecture. These newer products include the CPU, GPU, memory controller, and the functions of the northbridge that was left as a separate chip previously. For the desktop segment, Vishera, Trinity, Richland are recent products based on the Piledriver core. These are fabricated with 32nm SOI technology. Kaveri is the latest 2014 released product line using the Streamroller core, fabricated with 28nm bulk Si technology.

For an upgrade, I am looking for a dual or quad core processor with at least 512KB L2 cache and a rated power of 65W or less. Here is a quick listing of products available to buy on Newegg:

Kaveri

A10-7850K quad core 3.7GHz, 4MB L2 cache with R7 GPU (512:32:8@720MHz), dual channel DDR3-2133, 95W
A8-7700K  quad core 3.4GHz, 4MB L2 cache with R7 GPU (384:24:6@720MHz), dual channel DDR3-2133, 95W

Richland

A10-6800K quad core 4.1GHz, , 4MB L2 cache with 8670D GPU (384:24:8@844MHz), dual channel DDR3-2133, 100W
A10-6790K quad core 4.1GHz, , 4MB L2 cache with 8670D GPU (384:24:8@844MHz), dual channel DDR3-1866, 100W
A10-6700 quad core 3.7GHz, , 4MB L2 cache with 8670D GPU (384:24:8@844MHz), dual channel DDR3-1866, 65W
A8-6600K quad core 3.9GHz, , 4MB L2 cache with 8570D GPU (256:16:8@844MHz), dual channel DDR3-1866, 100W
A8-6500 quad core 3.5GHz, , 4MB L2 cache with 8570D GPU (256:16:8@800MHz), dual channel DDR3-1866, 65W

Trinity

Athlon X4 740  quad core 3.2GHz, , 4MB L2 cache with no GPU, dual channel DDR3-1866, 65W
A10-5800K quad core 3.8GHz, , 4MB L2 cache with 7660D GPU (384:24:8@800MHz), dual channel DDR3-1866, 100W
A10-5700 quad core 3.4GHz, , 4MB L2 cache with 7660D GPU (384:24:8@760MHz), dual channel DDR3-1866, 65W
A8-5600K quad core 3.6GHz, , 4MB L2 cache with 7560D GPU (256:16:8@760MHz), dual channel DDR3-1866, 65W
A8-5500 quad core 3.2GHz, , 4MB L2 cache with 7560D GPU (256:16:8@760MHz), dual channel DDR3-1866, 65W

the numbers in parentheses for the GPU is the number of Unified Shader Processors : Texture Mapping Units (TMUs) : Render Output Unit (ROPs)

I included the unlocked APUs (denoted with K at the end of model number) because they could be run at a lower clock frequency to lower the power dissipation. The K parts are probably just the slightly higher performing parts from the production lot that allows them to run at the maximum clock frequencies possible. In many cases the K parts are the same price or only 1 or 5 dollars more than the slightly slower parts with otherwise the same specs.

The A10 parts are consistently $20 more than the A8 parts, so after eliminating the non K and A8 parts, the short list is:

$175 A10-7850K 3.7GHz, R7 GPU (512:32:8@720MHz), DDR3-2133, 95W
$130 A10-6800K 4.1GHz, 8670D GPU (384:24:8@844MHz), DDR3-2133, 100W
$120 A10-5800K 3.8GHz, 7660D GPU (384:24:8@800MHz), DDR3-1866, 100W

There's only $10 different between the A10 Richland and A10 Trinity, so I will eliminate the A10 Trinity in favor of the updated Richland core. Now the question is, is it worth it to pay $45 more for the latest Kaveri core?

After reading Tom's Hardware's review on the Kaveri A10-7850K and A8-7600, it becomes more clear to me that the Kaveri APUs performs better than Richland in the graphics and gaming department. The Intel Corei3-4330 is a strong contender for the AMD parts in the same price point, with Intel clearly winning in the raw CPU workloads. The A8-7600 is quite interesting in that it can be configured to have a 65W or a 45W thermal ceiling, and its performance is significantly improved from the Richland 45W APU.
http://www.tomshardware.com/reviews/a10-7850k-a8-7600-kaveri,3725.html

Moving to 28nm bulk for the Kaveri allows for more transistors in a similar die area, but the clock frequencies are lower. Overall the CPU performance was tweaked so it makes up for the lower frequency, but the graphics performance is higher due to more graphics processing units.

$175 A10-7850K 3.7GHz, R7 GPU (512:32:8@720MHz), DDR3-2133, 95W
$129 A8-7600      3.3GHz, R7 GPU (384:24:8@720MHz), DDR3-2133, 65W
$130 A10-6800K 4.1GHz, 8670D GPU (384:24:8@844MHz), DDR3-2133, 100W

The A8-7600 is not yet available but should be coming soon in Q2 or Q3 of 2014.

At this moment the A10-7850K looks like the best choice.

Motherboard

To use the Kaveri APU, a motherboard with the FM2+ socket is needed. I searched on NewEgg for a micro-ATX board with 4 memory slots that supports DDR2133 speed, and only a handful of results came up. Looks like the GIGABYTE GA-F2A88XM-D3H is a good choice, using the AMD A88X chip, with PCIe 3.0 slot, maximum memory of 64GB, 8 SATA 6Gb/s connections, HDMI output, optical S/PDIF surround sound 7.1 audio out, gigabit Ethernet, 4 USB 3.0 and 8 USB 2.0 connections. $80

If I keep using my current LANBOX Lite case, a micro-ATX motherboard is fine, but if I want to get a smaller case, then I will need to go to a mini-ITX motherboard. Choices are Gigabyte GA-F2A88XN-WIFI or ASRock FM2A88X-ITX+. Both have built-in WiFi and blue-tooth module, but I like the Gigabyte's antenna slightly better. The ASRock has an eSATA port and the Gigabyte does not. The ASRock uses a newer audio CODEC chip Realtec ALC1150 vs. the Gigabyte ALC892. Not sure if there is a lot of difference, but if I want really good audio, I can get a PCI-e sound card such as the ASUS Xonar DX7.1 which will probably be significantly better audio. The board layout in terms of CPU and memory slot orientation is different between the ASRock and Gigabyte. I like the Gigabyte one for a slim computer case that will be used vertically, because the memory slots in the ASRock board will be above the CPU heatsink, which will impede air flow and the heated air from the CPU will go towards the memory modules. If I need a firewire port, I probably need to get a PCI-e card with that interface, because firewire ports are obsolete from newer motherboards. I still have an old Mini-DV camcorder that I may use to capture videos.

Memory

The graphics processor on the APU uses the system memory for processing the graphics, so having a fast communication channel to the memory is helpful for overall graphics performance. looks like the latest mainstream motherboard and chip set supports DDR3-2133, so a module that supports 2133 or higher is sufficient.

When going with a mini-ITX motherboard, the heatsink is likely going to be over the memory module. I found this page which has nice pictures of the AXP-100 heatsink over different height memory modules:
http://www.geocities.jp/hangaya_craft/pcg3/arekore_test_g19.html

Looks like it's fine with low profile and standard modules without heat spreader. But with heat spreader, the CPU heatsink will be touching the heat spreader. the AXP-100 has about 27mm below the fin area. listing the ones with low profile heat spreader:

Patriot Viper 3 Low Profile Blue PVL316G213C1KB DDR3-2133 1.5V 11-11-11-30 $179.99

G.SKILL Ares Series F3-2133C10D-16GAB DDR3-2133 1.6V 10-12-12-31 $154.99
Team Vulcan TLAD316G2400HC11CDC01 DDR3-2400 1.65V 11-13-13-35 $154.99

AMD Radeon Gamer Series DDR3-2133 1.65V 10-11-11-30 $219.99

this one with very tall heat spreader is popular on Newegg:
G.SKILL Trident X Series 16GB (2 x 8GB) F3-2400C10D-16GTX DDR3-2400 1.65V 10-12-12-31 ~$150 with promotion

looks like I should get either the Patriot Viper 3 Low Profile or the G.SKILL Ares Series having low voltage and low latency.

With 16GB of memory, I would need to get Windows 7 Home Premium (16GB max) or Professional (192GB max) or Windows 8 (128GB max) to support the amount of memory I have.

Hard Drive

A solid state hard drive (SSD) is silent; no moving parts. A SSD is probably going to make the overall computing experience faster, and nowadays 128GB or 256GB capacity is affordable and more than enough for the Windows operating system and typical suite of desktop software. Other things like multimedia can be stored on external hard drives which we still have 1 TB of space.

After reading reviews and learning a bit more about recent solid state drives, I got the feeling that Samsung has good brand name recognition and performance, and OCZ (which was acquired by Toshiba) is also popular but with slightly less name recognition, but seems to chime well with enthusiasts.

OCZ Vector 150 Series 240GB SATA III 2.5-Inch 7mm Height Solid State Drive (SSD) VTR150-25SAT3-240G $179 on Amazon.com is a high performance drive with good reviews, high capacity for the money, and up to date interface and drivers.

CPU Heatsink

In my last computer build, I used the Thermalright passive heatsink SI-128 and I really like the idea of not having to install a fan on the CPU heat sink. This time I looked up Thermalright's latest offerings and found the AXP-100 to be attractive because it is only 44mm high, offset to one side so it can clear the RAM slots easier. It can be paired with a low profile 100mm fan or 140mm fan which is 14mm high. regular 140mm fan can be used but are 26mm high. This heatsink would be more than enough to cool a 65W or a 95W CPU with a slim profile, allowing the computer case to be small form factor and still use a powerful processor.

Another thing I like about the Thermalright product is that they show detailed specs of the products including all dimensions, so I can design and make sure things will fit together before getting the components. Their build quality from my last purchase was also excellent. I will have to experiment to see if cooling can be adequate without a fan, or if the fan can be kept at a low RPM to be very quiet.

Thermalright AXP-100 $50 from Amazon.com

Computer Case


Finding a computer case involves many compromises and many considerations that have to work together. I found a neat website that makes the search and selection process a lot easier:
https://pcpartpicker.com/parts/case/?compatible_with=thermalright-cpu-cooler-axp100

It lists compatible cases by the CPU, motherboard, and CPU cooler.  It happens to have the Thermalright AXP-100, so I started from there. It then allows you to filter the list by motherboard form factor, user ratings, etc. then you can sort by dimension of the case or even the volume in cubic feet. This is like a computer builder's dream come true.

Here is a short list of cases that I down-selected:

No optical drive bay
Antec ISK110-VESA Mini-ITX 90W PSU 0.13cu.ft. $77.02

1 optical drive bay
HEC ITX200B Mini-ITX HTPC 200W PSU 0.21cu.ft. $49.99
HEC ITX200A Mini-ITX HTPC 200W PSU 0.21cu.ft. $59.99

Antec ISK 300-150 Mini-ITX Desktop 150W PSU 0.24cu.ft. $79.99
Antec ISK 310-150 Mini-ITX Desktop 150W PSU 0.24cu.ft. $82.92

In-Win H-Frame Mini Mini-ITX 180W PSU 0.29cu.ft. $175
In-Win BP655.200BL Mini-ITX 200W PSU 0.29cu.ft. $63.99
In-Win BP671 Mini-ITX HTPC 200W PSU 0.29cu.ft. $69.99

LIAN LI PC-Q03A Mini-ITX Tower Case no PSU 0.36cu.ft. $89.99

For comparison, my last computer case was the Thermaltake LANBOX Lite which is a microATX case that occupies 1.05cu.ft. Half the size would be 0.5cu.ft., which is a Mini-ITX cube such as the Lian Li Mini-Q (PC-Q03) with 1 full height expansion slot, 1 external drive bay for optical drive, and can accommodate a full size ATX power supply. Half again would be 0.25cu.ft which is a slim Mini-ITX like the Antec ISK300-150 having 1 external drive bay for optical drive and 1 half height expansion slot in the back. Half again would be 0.13cu.ft. which is like the Antec ISK110 which has the power supply externally, no external drive bays, and no expansion slots. These are small enough to mount behind a monitor or hang on a wall.

The Antec ISK110 is a minimalist case, about 3 inches (76mm) thick. There is no provision for cooling fan, but there is lots of mesh area for natural convection air flow, which is good. There is indication by one Amazon.com review that the Thermalright AXP-100 heatsink can fit if the mesh is removed, but it is not clear to me whether the AXP-100 can be used with a low profile fan or not. With a fan the height of the AXP-100 is 58mm, which leaves 18mm for the case, motherboard, CPU socket, and motherboard bottom clearance. In any case this seems quite tight. Power supply is another potential issue because the stock one is 90W and some of the AMD APUs have a thermal design power of 95 or 100W. Even if I use the 65W APU, the motherboard, memory, hard drive, and USB devices may draw more than 25W. There is the possibility of replacing the stock power supply with a higher power one, such as the pico-PSU by mini-box (http://www.mini-box.com/picoPSU-150-XT) for $45.

The HEC ITX200A and ITX200B are slim cases about 2.76 inches (70mm) thick. There is a built-in 60mm case fan on the side and a small fan in the stock power supply which provides 200W of power.
The fact it has these 2 fans is not as attractive for me, and the CPU heatsink may be too tall for this case.

The Antec ISK 300-150 and ISK 310-150 are thicker at 3.8 inches (96mm). There is a built-in 80mm case fan, and the power supply is built-in with AC power cord directly to the case. The wiring for peripherals seem to have large connectors and take up space.  There is room to fit the low profile CPU heatsink, and the exhaust fan is on top when the case is oriented vertically, which should help pass air through the CPU heatsink. The power supply could be replaced by the pico-PSU and the internal wiring can be changed or customized by some engineering.

In-Win H-Frame Mini is the most expensive case in the short list, and it is also very different from the other cases in that it is quite open to air with several parallel aluminum plates forming the case. It comes with a piece of tempered glass for replacing the opaque side panel so you can see what's inside. The case is fanless, and the power supply is built-in and has a fan. the cable management is already done on the underside of the case. overall thickness is 4.3 inches (109mm).

In-Win BP655.200BL and In-Win BP671 are similar with different external looks. There is built-in power supply with a fan on the inside of the case. There is a 80mm fan on top near the front of the case, so there are two fans pulling air out near the top of the case. the overall thickness is 3.9 inches (99mm).

LIAN LI PC-Q03A is a bit larger because it can accommodate a full length full height graphics card. It is all aluminum and does not include a power supply, but it can accommodate an ATX or a micro ATX power supply.

The Antec ISK110, Antec ISK 300/310-150, and the In-Win H-Frame Mini are my top choices based on how well I think they will work with the Thermalright CPU heatsink and minimizing the number of fans. I may want to get a pico-PSU if the built-in PSU is too noisy.

Overall Proposed Build
Option A: highest performance and expensive case
CPU: AMD A10-7850K 3.7GHz, R7 GPU (512:32:8@720MHz), DDR3-2133, 95W $175
CPU cooler: Thermalright AXP-100 $50
RAM: Patriot Viper 3 Low Profile Blue PVL316G213C1KB DDR3-2133 1.5V 11-11-11-30 $179.99
MB: Gigabyte GA-F2A88XN-WIFI $106.69
SSD: OCZ VTR150-25SAT3-240G $179
Case: In-Win H-Frame Mini $175
Total: $865.68

Option B: good performance and modest case
CPU: AMD A8-7600 3.3GHz, R7 GPU (384:24:8@720MHz), DDR3-2133, 65W $129
CPU cooler: Thermalright AXP-100 $50
RAM: Patriot Viper 3 Low Profile Blue PVL316G213C1KB DDR3-2133 1.5V 11-11-11-30 $179.99
MB: Gigabyte GA-F2A88XN-WIFI $106.69
SSD: OCZ VTR150-25SAT3-240G $179
Case: Antec ISK 300-150 $79.99
Total: $724.67 ($141.01)

Option C: use current case (LANBOX Lite)
CPU: AMD A8-7600 3.3GHz, R7 GPU (384:24:8@720MHz), DDR3-2133, 65W $129
CPU cooler: Thermalright SI-128 (from existing computer)
RAM: Patriot Viper 3 Low Profile Blue PVL316G213C1KB DDR3-2133 1.5V 11-11-11-30 $179.99
MB: Gigabyte GA-F2A88XM-D3H $80 (no wi-fi)
SSD: OCZ VTR150-25SAT3-240G $179
Total: $567.99 ($297.69)

someone has shown a build very similar to Option A:
http://www.overclock.net/lists/display/view/id/5752211

Saturday, April 12, 2014

Time Warner Internet Speed Test April 2014

I noticed that Time Warner Internet monthly bill went up again from $34.99 to $37.99 in April 2014. This is the Lite service with 1Mbps for upload and download.  I am signing up for the Standard service with 15Mbps download and 1Mbps upload for $34.99 for first 12 months. Then after that price will go up to retail price, which is not specified.

This is the screenshot of the speed test with the Lite service:









The next day, 4/12/2014, I ran the tests again and the highest I got was with the CT server:









Most other servers give 2 to 6Mbps for download and 1Mbps for upload.  Going farther to west coast servers, Hawaii server, and Taiwan server, the download speeds are all between 0.5Mbps and 2Mbps with upload still at 1Mbps. So it seems like the benefit of having the Standard service with up to 15Mbps download speed is limited and highly depends on the origin of the data we are trying to download.

Monday, January 28, 2013

New Cable Modem and Internet Speed Test

Around October 2012, Time Warner Cable started charging cable Internet customers in our area a $3.95 monthly Internet modem leasing fee. The actual cost increase is slightly higher, since there is sales tax on this fee. This amounts to $51.24 more per year for using their Internet service. Users can opt-out of the leasing fee by purchasing their own modem and returning the leased modem. That's what I did around New Year Day 2013. Time Warner has a list of approved devices on their website. I had the Motorola SB5101 leased from Time Warner, and I thought if I were to buy one for myself, I might as well upgrade to a newer device.  I was deciding between the Motorola SB6141 and SB6121. They look the same from the outside, and both support the latest DOCSIS 3.0 standard for very high speed Internet. The older SB5101 supported up to DOCSIS 2.0 for speeds up to about 30 Mbps, while the SB6141 can support up to about 320 Mbps download by combining 8 channels, and the SB6121 can do up to about 160 Mbps combining 4 channels. Both SB6141 and SB6121 have upload capability up to about 160 Mbps. According to Time Warner's approved devices web page, SB5101 is not suitable for Road Runner Turbo service or above. From the modem specifications, SB5101 should be able to handle the 20 Mbps download and 2 Mbps upload speeds, but maybe the higher level services uses a different communication standard that requires the newer models. The highest speed offered right now is the Ultimate 50 Mbps download and 5 Mbps upload, and I don't think the SB6141 would offer any advantage over the SB6121 for my current or future situation, while the SB6121 is about $20 cheaper, so I ordered the Motorola SB6121 from Amazon.com.

I got the modem before New Year's Eve, and I tried to activate it by calling Time Warner and by using their web site's online chat, but on 12/31/2012, no one seemed to be on the other side of both the phone and the web site, so I waited until 1/2/2013 to try again. This time, the online chat was able to help me activate the modem. I just tell them the MAC address written on the back of the cable modem, and within minutes, it was switched over. When that happens, my online chat still says it is connected, until a few more minutes later, which went dark. I then quickly go to the old modem and swap the cable and router LAN wire to my new modem. Within a few minutes, I have Internet again, and the Time Warner online chat was back too. The chat support person confirmed that it is all working fine, and that they have put a partial credit for the modem lease fee on my account, and we ended the conversation.

That same week, on Saturday, I took the old modem to the nearest Time Warner office to return it. They are open on Saturday from 10 AM to 2PM. There was a line of people waiting to get customer service, plus a line of cars outside using the drive-through, but I didn't have to wait a long time. They seemed very efficient. When I got to my turn, I just said I am here to return the leased Internet modem, and the Time Warner employee was very professional and friendly, and ran the return process smoothly. It was done within 2 minutes, and I was given a receipt that detailed the transaction, including the MAC addresses of the old and new modems, saying that I've switched to my own modem and is returning the leased one. I was impressed by the thoroughness of the modem return process.


Speedtest.net Internet speed results
At home, I ran an Internet speed test on http://www.speedtest.net/, and see that my Internet speed is 1.00 Mbps download and 0.97 Mbps upload. I have the Road Runner Lite, and I was getting 0.56 Mbps download and 0.12 Mbps upload prior to their service upgrade some time in 2012. One day in 2012, I got a letter from Time Warner saying they were going to upgrade my Internet speed with no increase to the monthly fee, and that's very nice, because there is a huge difference between an upload speed of 0.12 Mbps and 1 Mbps, especially for the response times of loading web pages. With the current speed, we are able to watch most YouTube videos without lag, and Netflix works fine. We can also watch TV shows on PBS.org, for example. The only problem is that if someone is watching an online video, then other people's web browsing becomes extremely slow. I think the time to upgrade the speed is when we will often have multiple uses at the same time, or when higher bandwidth is needed to support streaming higher definition videos.

Thursday, January 19, 2012

Calculating R-Values

uninsulated garage wall
















R-value is a measure of thermal insulation in a house.  Each of your walls, ceilings, and floors has a certain amount of insulation and a corresponding R-value. The higher the R-value, the better the insulation, and the less heat is lost through that surface.  In this post, I will explain 2 ways to calculate R-value for a given building assembly.

Calculate theoretical R-value from wall structure

If you know how the wall or ceiling is constructed, you can calculate its R-value from its component material R-values. You can look up the R-values of common building materials on the Internet, and the R-values add when they are layered in the direction of heat flow. For example, rigid foam insulation is R-10, and fiberglass batt is R-13. Putting the foam and then fiberglass gives a total of R-23. Keep in mind that wall studs have lower R value than fiberglass batt, so those areas have a lower R-value. You can get an average R-value of a mixed surface by multiplying their respective U-value by the percentage of their area, then sum them. U-value is the inverse of R-value. A material with R-value of 10 has a U-value of 1/10 = 0.1. For example, a 2x4 studs area has R-4, and occupies 10% of the area; the fiberglass batt has R-14, and occupies 90% of the area. You can calculate the overall U-value as 10%*1/4+90%*1/14=0.0893. Convert back to R-value gives 1/0.0893=11.2. Note that the typical fiberglass batt filled walls don’t achieve the R value rated on the fiberglass batt because of the higher thermal conductivity of wood studs; this is also referred to as thermal bridging.

Calculate R-value from temperature measurement 

Instead of calculating the theoretical, optimal R-value of a given wall, you can calculate the actual, empirical R-value by measuring some temperatures. This can be done with a hand-held infrared thermometer. The best way to use an infrared thermometer is to hold it as close to the surface as possible, and avoid shiny surfaces.

Here's a table that will give you an estimate of wall R-values based on outside temperature and the temperature of inside surface of an exterior wall:




Estimated R-Value


1
2
5
10
15
20
25
30
35
40
Outside Temp (F)
40
 20.4
 10.2
 4.1
 2.0
 1.4
 1.0
 0.8
 0.7
 0.6
 0.5
30
 27.2
 13.6
 5.4
 2.7
 1.8
 1.4
 1.1
 0.9
 0.8
 0.7
20
 34.0
 17.0
 6.8
 3.4
 2.3
 1.7
 1.4
 1.1
 1.0
 0.9
10
 40.8
 20.4
 8.2
 4.1
 2.7
 2.0
 1.6
 1.4
 1.2
 1.0
0
 47.6
 23.8
 9.5
 4.8
 3.2
 2.4
 1.9
 1.6
 1.4
 1.2
-10
 54.4
 27.2
 10.9
 5.4
 3.6
 2.7
 2.2
 1.8
 1.6
 1.4
-20
 61.2
 30.6
 12.2
 6.1
 4.1
 3.1
 2.4
 2.0
 1.7
 1.5
-30
 68.0
 34.0
 13.6
 6.8
 4.5
 3.4
 2.7
 2.3
 1.9
 1.7
-40
 74.8
 37.4
 15.0
 7.5
 5.0
 3.7
 3.0
 2.5
 2.1
 1.9


Interior wall T - exterior wall T


This table assumes your interior wall temperature is 70°F, but you can still use it even if your indoor temperature is a little different; the R-value will be less accurate if indoor temperature deviates from 70°F. To use the table, calculate the temperature difference between interior wall temperature and the inside temperature of the exterior wall that you want to know the R-value of. Look at the first column and choose the row that corresponds to the outside air temperature, look for the temperature difference closest to your measurement, then look up on top row for the estimated wall R-value.

Now, if you want to be more precise, calculate the R-value directly with the formula R=(Th-Tc)/(Ta-Th)*0.68+0.68, where Th is the interior temperature of an exterior wall, Tc is the outside air temperature, and Ta is the indoor temperature. The indoor temperature can be measured on an interior wall or door or an object that should be in thermal equilibrium with indoor air. Outside air temperature can be measured on an outside object that’s in thermal equilibrium with outside air, such as a trash can or a deck. When using an infrared thermometer, avoid using it in daylight or measuring shiny objects. Also avoid measuring objects on or near the ground because the ground is often at a different temperature than air.

Note that all this discussion assumes no air leakage, no convection, no radiation, and no condensation. Only thermal conduction is considered here. Typically, air leakage or significant air movement would dominate heat loss so much that conduction becomes meaningless and usually the best thing to do is to eliminate air movement first.

Appendix

I will try to explain how the R-value formula is derived. First you need to understand the definitions of R-value and U-value. R-value is the thermal resistance of a material. U-value is the thermal conductivity, which is the inverse of R-value. 1/R-value = U-value. For example, an R-5 wall has a U-value of 1/5=0.2. U-value has the unit of BTU per hour per degree F per square foot (BTU/hr/F/sq. ft.). To calculate heat transfer through an R-5 wall with 70°F on one side and 60°F on the other side, just multiply its U-value by the temperature difference: 0.2*10 = 2 BTU/hr/sq.ft. To calculated the heat loss through a wall 8 feet high and 10 feet long, multiply by the area: 2*8*10 = 160 BTU/hr.

The model for calculating the R-value of a wall is that heat is moving from indoor (Ta) to inside surface of the exterior wall (Th) through a layer of air film with R-value of 0.68, and the same amount of heat is moving from inside surface of exterior wall (Th) to outside (Tc) through the wall with R-value of wall being the unknown variable. The equation is Uair*(Ta-Th) = Uwall*(Th-Tc). Solve for Rwall by rearranging the equation. In the above formula, you add another 0.68 at the end because the overall R-value of the wall includes an interior air film with R-value of 0.68. You can leave it out if you want just the R-value of the wall material itself.

Thursday, September 1, 2011

Grass Update

My new strategy is to plant grass seeds every 2 weeks, keep the soil surface moist with sprinkler, and add fertilizer every 4 weeks.  I started in mid-July, and so far had seeded 3 times, fertilized twice.  The grass is growing deep green, but coverage is still not 100%.  Coverage increases after each planting, but it's not easy to get 100% coverage.  I spent effort pulling weeds out, especially the crab grass.  I also made a home-made rain gauge using a shallow bucket with inch markings to keep track of rain fall.  It is quite useful, because sometimes the rain would seem heavy, but accumulation is minimal, and sometimes it doesn't look wet, but had accumulated more than 1 inch.

repeated planting helps to increase coverage and keeping the ground surface moist improves the grass's survivability.  I think this strategy is working out better than previous strategy which is to plant a lot of seeds at once.


Saturday, June 18, 2011

Dead Grass

I planted lawn grass last fall and this spring, and recently there was a 2 week period with unusually hot and dry weather, so about 40% of the last year's grass died, and about 90% of this spring's grass died.  I did a soil test yesterday, and found out that PH is around 7, which is good.  Nitrogen and Potassium are both very low, and Phosphorus is medium.  This tells me that the grass grows very slowly because there is not much nitrogen.  I suspect the grass died because it was not very mature, and the roots were not very deep yet, so it could not withstand the extended dry weather.  I went out and bought fertilizer and top soil.  The fertilizer is the local Country Estate brand by Hewitt's, and the one I am using is the winterizer product with 14-18-14 of NPK.  I first rake the ground to remove the dead grass, then put down 6 bags of top soil on this roughly 100 sq. ft. area, rake it even, apply the fertilizer, apply the grass seeds, rake and mix the soil and seeds, then water until it is very wet.

The recommendation I found online is to make sure the grass get 1 inch of water a week.  Putting down straw will help, but I didn't buy any this time.  The seeding instruction from Hewitt's recommends fertilize the new grass with the high phosphorus fertilizer after 1 month and after 2 months, until the grass is mature and strong.  I am looking forward to seeing the effect of the fertilizer, because this is the first time I am using fertilizer on lawn grass, so I should be able to observe the difference between this planting and last few times which was without fertilizer but with compost.