Intel Shelton Board

It been over a month now since I start to experiment with the Intel Shelton board. I have a Demartech 300 mW radio card and I use it as an uplink node. So far no failures. As an uplink node it is more powerful than my current Compaq computer. I will start looking for a case to house the unit in.

I have been watching a lot of streaming media through the board and no hiccup or jerkiness. So far it has accepted a lot of bandwidth and the kids are playing games through it. I think it is time for some field tests.

Double opt in

I know that a lot of you have signed up for my email notification program when I update my blog. I now have to change it to a double opt in as per Can-spam rules. For those who don't about the rule in order for this site to send you a email I need to ensure that you are opt into a permission base system. The double opt in mean that I have to do it twice.

Try out: http://www.moskaluk.com/mail_notification_form.htm

Thanks

Add more power

Almost all the advice I get from people is to add more power to better RF signals and performance. So over the years I have slowly migrated from a standard 802.11b radio card to a whopping 300 mW radio card. As the power increase I improved not only coverage but I also improved performance.

Low and behold. To improve performance I tried staying away from LMR-400 solution and built an out door unit. I can tell you the out door units have been very successful. The power out put has minimum lost and I get great performance.

The indoor units are a problem. I tried to upgrade radio cards but the radio loss from the large cables reduces signal performance. I have some LMR-400 over 75 feet long. The loss is significant.

So I did the next big thing to increase RF signals. I added an amplifier to my LMR-400 cable. The amplifier now sends a strong signal to the omni directional antenna. The loss is still significant but the improvement is very impressive.

The size of the amplifier is lower than the legal limit set; however, I can theoretically buy a larger amplifier because of the cable loss. I forgot to mention that the amplifier is located between the Mesh node radio card's pig tail and the LMR-400 cable connection.

I'm about to test is out shortly between to weak links. Stay tuned!

Old fashion Skip

Tropospheric ducting is radio propagation that tends to happen during periods of stable weather. In this propagation method, when the signal encounters a rise in temperature known as a temperature inversion, the higher refractive index of the atmosphere will cause a radio signal to be bent. Tropospheric ducting affects all frequencies, and signals. In other words good old fashion radio skip.

there is a good site that tells you when skip may occur http://home.cogeco.ca/~dxinfo/tropo.html

I can stop scratching my head. As I look at the temperature and humidity I believe are indicators of a temperature inversion. The local traffic reporter in his plane always says there is a temperature inversion today. From now on I will listen to the traffic reporter rather than the weather man.

What does this tell me. Well it looks like I have to manually set each stationary mesh node. What I setting is to block any nodes that are not in the area of the radio signal. The reason for doing this is to illuminate the skip. Now my experience with skip was that I was experiencing very good throughput, but the signal did not last very long.

To block all the radio signal seems impossible, so my strategy is to block any signal that is skipping. It still a manual approach but at least I can stop scratching my head!

Just when you figure it out

Well, I back to scratching my head. Again it is weather related. The last few days we been having thunder storms. It has clean up the air quality in Toronto and the RF defraction and reflection issue have gone away, so I thought. I observed that the outer links that disappear after the first thunderstorm have now reappeared. It is not hot, humidity is way down and actually it beautiful day. I still getting weak signal strength between nodes and the atmospheric reflection is back.

There seems to more than a reflection problem going on.

Then it rained

The last 7 days have showed poor link quality of mesh between regular nodes and improved distant nodes links. It has been very trying week. In previous blogs the RF signal quality has been poor. But at the same time I seen nodes link together and signal strength between nodes that do not have a line of site.

What I am observering is reflection and refraction of the RF signal from the atmosphere due to pollution, humidity and temperature over 30C.

As it was getting hotter we had a thunderstorms with heavy rain that washed away most of the pollution and humidity.

What happen was the mesh node links improved again and the distance reflected links were once again lost.

Extreme conditions have illustrated that RF signal is affected by weather conditions.

Atmosphere Condition

After two weeks of drought conditions and 7 days of temperatures in Toronto being over 30 C I started to get reflection off the atmosphere. I noticed this on two different times of the year. The first when it was extremely cold such that the temperature was lower than -15 C and now when the temperature was greater than 30 C.

What happens is the outer node start receiving clear signals from nodes that are over a kilometre away. The signal reflects off a layer of the atmosphere and reflects back to the antenna. The signal is clear and can link two mesh nodes together.

When it was extremely cold my assumption was that pollution with water vapour was blanketing the city. Temperature inversion were also detected by weatherman. One could only speculate that the cold temperature created ice crystals that help reflect the RF signal.

In hot temperature a combination of pollution and humidity also create the same type of reflection.

Nevertheless in both case the linking of reflect nodes was stable when connected and did provide adequate through put.

It's not the heat it is the humidity

Mesh Connectivity is not only hardware and software, the environment plays a large role. Weather and temperature variation also seem to play a role in connectivity. I have notice in -15C weather that the Mesh Nodes are working perfectly; however, in the 28 C weather there is a noticeable difference in quality.

In my particular setup the nodes are about .7KM apart on average. As soon as I increase the amount of available nodes and start to decrease distance between the nodes I get a constant operations; however, temperature still has an affect on wi-fi. Understanding this concept will help you plan your network out.

Humidity also plays a role. I have not scientifically determine, tested and measure this problem; however, humidity, temperature or pollution play a key factor in network operations. From the weather observations we must take this into account when planning a mesh network.

The article I wrote in planning a mesh does not take this into account. I should update the documentation www.moskaluk.com/papers.htm .

I guess heat and humidity is giving me humility.

Network Problem Resolved

Well, this one goes into the record books as don't assume anything. In the last 4 weeks I had a network problem. I kept loosing 4 of my meshap's. It would go up then down then up again then down. I thought it had something to do with either Northern Lights or leaves on trees or weather temperature. Well, no. It was one of my customers thinking that he didn't like the antenna where I put it and that his position was more pratical for him. It create echoes, and links problems etc.

Moral do not assume anything and always verify that your installation is true.

Compaq finally failed

Well I have been running an experiment for the last year and half running outside a Compaq slim line computer in a protected shelled. Yes through winter, spring, summer and fall. It has been a wonderful device surviving extreme cold temperature and extreme hot temperature. Yesterday it finally gave up. It boots up partly and I got an "Unable to handle kernel paging request at virtual address 0000001f1." Now the unit started to overheat and I think the ram may be gone but Geez this was my very first MeshAP! I paid $35 dollars for the box, $75 dollars for the radio card and $49 dollars for the pigtail and antenna. The box is worthless but the radio card antenna and pigtail are still operational!

For under $200 dollars Canadian this baby worked outside! The problem was the board was already 5 years old and it was on it's last life. The test is inconclusive as I did not put any bench marks. The idea was to observe.

I running mini-itx boards today and I have many in outside. So far I haven't had any hardware failures.

Rebooted all nodes simultaneously

Today was beautiful hot day in Toronto. Summer is showing itself. So too are all the air conditioners in town. We had our first power blackout of the season. I'm not sure if it had anything to do with the warm temperatures but what happen all my MeshAP rebooted simultaneously.

So what happenings when everything reboots due to a power failure. Well the PC-MeshAP did not automatically reboot. But the Mini-itx boards both Via and Intel were set to startup automatically. Everything seems to work better after a reboot. All nodes are reporting in and nothing is falling asleep.

I noticed that as every node found each other again.

I was having problem with blocknodes that some of the nodes where blocking each other. This has disappeared, for now. It too early to tell but the operations seems to be normal again. I don't see these problem of nodes not reporting in on time. Again I see how long that lasts.

Project updated

Well it officially has been two weeks testing the Shelton MB. The Shelton motherboard has performed well as a uplink node and a repeater node. I currently have a normal power supply on it rather than morex unit. The morex unit that was sent to me stop working after 30 seconds. Again I waiting for suppliers.

I also wanted to put a via board into a demarco technology metal out door case. There are many problems with this setup because the demarco case is just barely larger than the via board. I tried to use PCI riser to minimize the space but all three PCI riser boards are not working. This has slow the progress of the solar panel experiment.

The dual boot mobile via board project. I have purchased a car LCD that is touch screen. The touch screen works with Windows but doesn't work with Linux. I checking for drivers! I will see if Locustworld can add the drivers for me!