Update 1

Target PC is setup (formerly ATAT computer). It's Network information is:

IP: 171.64.164.18

LAN Mask 255.255.0.0

Host Name: threepio

domain: stanford.edu

It appears that the only 3.5" floppy drive that works is on Sanjay's computer (the one on Sith does not appear to work).

Currently, the Target PC is booted up via the floppy, but I haven't been able to get the HostPC? to interface with it yet. Files are located on Yoda under sphan\sensor_arm\Xpc

For a quick XPC tutorial, search for XPc on the bdml website.

-- SamsonPhan - 04 Oct 2010

Update 2

Plugging laptop into port that Hostpc will use:

IP Address 192.168.1.100

Subnet 255.255.255.0

default gateway: 192.168.1.1

DHCP server: 192.168.1.1

DNS Server 171.64.7.99

171.64.7.77

Running /sbin/ifconfig -L en0 on the QNX env't

inet 171.64.164.18

broadcast 171.64.255.255

Update 3

TargetPC? uses 192.168.1.100. Host PC must be setup to be 192.168.1.101. Both are port specific (blue to target, green to Host). Successful acquisition of signal. Note, AI0 maps to channel 1 in Matlab. All others are offset by 1 as well.

-- SamsonPhan - 26 Oct 2010

Update 4

Successful interrogation using the PWM freq measurement block. The following figure shows compression of the sensor while sampling at 1 Khz.

The signal is significantly noisier when not compressed. All "bad" data points exhibit a tick value of <2, which translates to a signal frequency an order of magnitude greater than previously seen. It can be considered noise. Applying a filter, we obtain a much cleaner result.

The cause of the noise may be improper grounding (highly possible given previous issues and the fact that noise is reduced when the sensor is compressed). The following two graphs show the effect of sampling rate on tick measurements.

• Noise__Test_1e-4.png:
  

• Noise__Test_1e-3.png:
  

In all cases, the number of "noise" ticks is relatively low compared to "real" readings, though it increases with sampling rate (373 vs 94). Note, the bins were automatically set by Matlab, and split the value down the middle. Also, the tick values are always integers. This reduces sensitivity but was expected.