WinMac Windows and Internet Consulting - Fujifilm

Site Map | Search | FAQ

Fluke

Overview

The 1750 Power Recorder (about the size of a VCR) is a high end instrument that connects to power generation equipment (power lines) to measure anomalies such as spikes, transients, dips, and other fluctuations. It generally stays connected from one week to one month recording data, at any time during or after that period, it is downloaded to a PC for analysis.

These fluctuations could be caused by events such as a squirrel jumping on the power line, a power generator about to fail, a malfunctioning transformer, or even too much load on the line. The idea is to detect the problem(s) and fix them or prevent their reoccurrence. This is data intensive (over 2GB), so efficient processing is essential.

There were three teams working on the project: one working on the hardware and firmware, one in India working mainly on the firmware and PDA software, and one working on the PC software with some assistance from India. WinMac initially started on the PC communication with the Recorder (TCP/IP) and went on to lead the PC team. Also, when a new developer was hired, WinMac assisted with the interview process and did the training.

Notable Components / Improvements

The PC client software that was developed has three main functions: 1) configuring the instrument, 2) displaying live data, and 3) displaying recorded data. Since the PC is less costly and more powerful than an embedded machine, the PC does as much of the computations as possible and all the post processing, leaving the instrument more capacity to do real time data capture.

The graph displays use a 3rd party control from IOComp. This control is quite flexible allowing multiple traces, labels, scaling, interpolation and so forth. The data display uses a grid control from Dundas Grid. It allows cell / row / column formatting including colors and fonts. It also allows windowing of the data.

	Refactoring: A major portion of the client application was systematically refactored by WinMac. There were many places where code was duplicated with slight variations, lengthy superficially related clumps of code, and exposed member variables. While this is understandable during prototype phases, it is less desirable in released code. The immediate benefits of refactoring were easier to understand code, uncovering of existing bugs and less bugs for added features. The additional long term benefits are easier maintenance, and easier enhancement implementation. This goes hand in hand with agile programming practices. WinMac is a big proponent of agile programming and thank Fluke for supporting this view. More information can be found on Scott Ambler’s site by clicking here.

	Instrument communication: One feature worth mentioning is that the client PC does a periodic IP broadcast to detect available instruments on the local network. The responding units are displayed in a drop down list. This a useful network discovery feature to implement in many situations.

	Recorded Data: The client was originally implemented for a much smaller dataset, performance was unacceptably slow and large datasets would crash the client. This was a “show stopper” issue. WinMac enhanced the client using caching, decimation and other means to improve the performance by orders of magnitude. The client crash was also corrected in the process.

	Engineering Client: A separate small application was created to assist firmware and hardware Engineers. This application provided a direct front end from which they could control the instrument (or the simulator). This was used throughout development and saved a lot of hours versus the alternative low level peek/poke type of tools available.

	Simulator: A 1750 simulator was also created. This allowed 1) testing of new PC client functions when the instrument development wasn’t ready yet, 2) development of the client when there was a shortage of instruments, 3) easier debugging since both ends of the process is under PC control. This required a clear understanding of the interface between the instrument and the PC client, and good judgment as to what level of simulation was required. All throughout development, this simulator was improved and it saved hundreds of hours.