Hand Hygiene

This test was performed by going through the power levels 1,6,11,16,21,26 and 31. Because of the power levels a boxplot would look terrible unless I made one for every power level so instead I have two CDF plots at power level 21. The plots are are of the two different heights because when I was looking at different plots in Minitab they appeared to have some interesting information:

Top Motes

Bottom Motes

The top motes break apart from 1m to 0.75m while the bottom motes have a very clear distinction between 0.25 and 0.50 meters.

This was a test using the spider antenna tilted downward at a 45 degree angle. This was only true when the antenna was facing the motes head on. As we rotated the antenna, the plane through which the 45 degree cut changed and you can really see how this affected RSSI values at the farther distances of the boxplot below:

Here is boxplot of just the top motes to help get rid of the noise produced by the two different heights:

Also, here is a Cumulative Data Function Plot

On the top motes it looks like we could discern between .25 and .5 or we could go between 1 and .75 meters. Again there is little distinction between .75 and .5 meters (the range we want).

The following is a plot of the bottom motes CDF

On the bottom motes however, there appears to be little distinction across the whole range of pertinent distances.

This was a test using a 7MM Antenna that was directly mounted to the board. I believe the mote was oriented with the antenna pointing down.

And here is a Cumulative Data Function Plot

From these two plots it doesn’t appear that this antenna has any clear cut off points, but further testing will most likely be done to see if the antenna can be useful in any fashion.

I added a Wiki page for the steps on mold making.

Power Level 10

Power Level 20

From these two plots it appears that an increase in power does not shift the cutoff point to a different difference. Instead it appears that as power increases the RSSI values will shift for each distance about the same amount, but at the higher power level the cuttoff for .25 meter is a lot sharper. This means that at power level 20 it appears that the spider could have a cutoff of roughly 6 RSSI and catch all the motes within .25 meters and ignore all motes farther out.

As the title suggests, tests were performed with these two antennas at a power level of 10. Both antennas were oriented vertically pointing downwards.

The following test was performed with an angle resolution of 20 degrees from 0 to 180 and at the height differences of .35 and .1 meters. The 180 degree mark corresponds to the side of the purell dispenser opposite of the mote side facing the receivers. The following two plots are of the RSSI for groklab’s Purell setup and for Ted’s purell setup

Some progress has been made towards the next generation pager design. As mentioned in an earlier post, I have been picking up where Gray left off redesigning the pager configuration. We have added a battery charging circuit and changed the USB connector. Gray’s design had problems physically fitting the internals into the enclosure and managing wired connections between the boards and the battery.

Fig 1 Board layout

In Figure 1 you can see that the board layout is similar to the boards that Gray had made. I replaced the Hirose IDC connectors he used with Molex right-angle Sherlock connectors. I have also moved all of the components that were placed under the daughter board out to make debugging easier.

A few hurdles during the redesign of these boards:

–The pinout for all of our boards using the Mini-USB connectors has been backwards. It had gone unnoticed because we didn’t reach the prototyping stage where we would power the boards via USB. (The pinout, from top to bottom as in Fig 1 is GND, Unused, D-, D+, Vcc)

–We had an unworking MAX1555 that was a bit hard to diagnose at first. Be aware of this if you have erratic voltage characteristic from this chip. In our circuit design, the input (DC) voltage differs from the output (BAT) voltage by no more than 5% (with a working chip). Keep in mind that this chip only operates within certain input voltages. This datasheet has been useful diagnosing these problems.

Fig 2 Mote Interface	Fig 3 Mounting		Fig 4 Vertical alignment

Some issues with this design:

• We still aren’t within the ideal vertical thickness for the boards. This is illustrated in Fig 4. Everything fits in the pager, but space is quite tight and requires brute force to close the thing up completely. This will eventually cause issues with reliability of connections, pager body fatigue, etc.. With recent design changes, the daughter board does not provide extra PCB real estate and serves only to raise the USB mini connector to the middle of the “bay” on the face of the pager. It also increases the thickness of the boards by tens of millimeters. The next design revision should concentrate on a way to place the USB connector in a reasonable position without a raised board.
  • Note the elevation of the white Sherlock connector in the picture on the top of this page. It is nearly the same elevation as the daughter board. I didn’t realize until I started writing this, but getting rid of the daughter board alone may not be enough.
  • We could do away with these connectors by soldering directly to each board (or using tiny single pin friction fit headers) and wire-to-wire connectors, as Ted used in his proto design.
  • Another option is to experiment with more pager modification/gutting. It may be possible to recess the charging board a few more millimeters into the enclosure. I am guessing we only need about 2mm.
• I question the reliability of wire-to-board connections (illustrated in Fig 2). During the assembly of this board, I had to resolder the wires that connect to the mote interface pins twice. The problem is similar with the USB I/O leads (not pictured), although much harder (impossible?) to avoid without being able to redesign the mote. Possible solutions include using small single or dual pin friction fit headers to solder into the mote interface, rethinking the soldering process, using a direct board-to-board connection.
  • I started looking at the diagram schematics for the mote to see if there is a way to access the USB I/O pipe from the expansion pins. I’m having trouble because I don’t really understand them. If we can get someone to take a look at them and see what they think, that would be great. (There are two pins on the 10-pin expansion that are labeled UART Tx/Rx).
  • If we can eliminate the necessity of soldering the USB I/O directly to the board, it would be possible to make a board to board connection, putting the USB directly on the charging board (rather than a daughter board). This would solve clearance and connection reliability issues simultaneously.
• It’s pretty hard to see, but Figure 3 illustrates part of the mote that has been destroyed by the grommets we use to space the mote and charging board. There should be a resistor where there is not, across the pads labelled R4. The outer diameter of the grommet interferes with this component on the mote. It does not appear that either of the two other mounting holes are likely to cause interference on either side of the board. This poses a large problem because there is very little clearance between the edge of the mounting hole and the component in question.
  • I will look in to other design solutions and other hardware solutions to avoid this problem.

I think the most efficient solution to these design problems concerns the daughter board on the charging board and the fact that we are limited to soldering the USB I/O connection directly to the mote. Further improvement will immediately concentrate on reducing the profile thickness, getting rid of the daughter board.

Ted comes through again with a fascinating video about a (relatively) simple way to visualize the RF field.  I’d like to see some of these in the lab.  Seems like it would be relatively simple to do if we attach an LED to the output of one of the motes and set it to go off when a threshold RF is reached.

Ted sent me a note pointing to low self-discharge NiMH batteries, which might be a good choice for the pucks and pagers.  He recommends coupling these with a better charging system, as cheap chargers can decrease battery life. 

Just wanted to note the idea here, so it is not forgotten.