Languages
Dear DDJ,
In "Reliability: The Hard and the Soft" (DDJ, May 2005), Ed Nisley dismisses Ada for DO-178B applications, saying "the entire cadre of Ada programmers is reaching retirement age with no replacements on tap." I'm surprised that Ed is propagating this myth. For one example, I'm not "nearing retirement age," and I've got two young engineers on my team, happily learning Ada and hard real-time programming. The AdaCore Academic Initiative (http://www.adacore.com/academic_members.php) has over 70 member universities, all training new Ada programmers. The Ada 2006 standard will be released next year, adding Java-style interfaces to the language, along with other significant improvements. Many Ada vendors are making money and growing; see http://www.adaic.com/index.html for current information on the state of the Ada industry.
The SPARK (http://www.praxis-his.com/sparkada/) subset of Ada is directly targeted to high-integrity systems, and is growing in popularity; several tool vendors are incorporating it (for example, ILogix—see http://www.ilogix.com/newsroom/newsroom_detail.cfm?pressrelease=2004_09_29_035924_126720pr.cfm).
Stephen Leake
Ed replies: Thanks Stephen. While it's true that Ada continues to be pretty much the only high-level language that's a good fit for DO-178B applications, it suffers from a certain lack of undergraduate mindshare. Judging from the vendors I've talked to (admittedly, a small and biased subset), the consensus is that while their Ada biz is doing okay, many customers are interested in a Java-oid high-reliability language. It comes down to economics. The same realities that drove the military to COTS hardware and software (with sometimes devastating consequences) is driving software toward readily available languages and personnel. That those languages aren't such a good fit for the job and the personnel might have the wrong background may be just unintended consequences. Looked at from the other side of the paycheck, though, knowing how to dance with Ada can be the ticket to a rewarding future. Undergraduates reading this, take heed!
Nuclear versus Wind Energy
Dear DDJ,
Jonathan Erickson's February 2005 "Editorial" presented some interesting numbers about U.S. wind energy. That generated a rather inflamed answer by our fellow reader Peter Andre, defending nuclear power, calling it "the cleanest, safest, best way to create energy." Not in the least.
These days, nuclear power is generating something like one third of the pollution put out by fossil-fuel operated power plants. Enriched uranium has to be mined and transported to the power plants, and in the end, nuclear waste has to be shelled and buried. All of this spends energy and produces pollution.
Unlike wind, enriched uranium is a finite resource, so scarce that the amount available on the Earth's crust is sufficient but to three years of world energy consumption (at present rates). Detailed information about this can be found at http://beheer.oprit.rug.nl/deenen/.
But there are other environment impact issues. That's why the future of wind energy lays off-shore. The EU has, among others, a program to build a wind farm in a 44 meters sea deep area off Scotland (DOWNVinD project); see http://home.wxs.nl/~windsh/offshore.html.
Actually, this nuclear/wind war is pretty stupid, because they are the best we have to tackle the oil peak. It seems that nuclear will be the first hour answer, but only wind power can be seen as a long- term solution for the post-oil days. The beginners guide to the oil peak is at http://www.wolfatthedoor.org.uk/.
Lu's de Sousa
Surround Sound
Dear DDJ,
In his excellent article "Surround Sound" (DDJ, July 2005), Don Morgan discusses the difference between the definition of a decibel by electrical engineers and audio engineers. The implication is that a given power ratio can have more than one dB value. This is not true.
The reason for the two formulas shown in the first paragraph on page 80 has simply to do with what unit of measurement is being used—volts or watts. If, for example, the two measurements were made on a voltmeter, dB=20*log(E2/E1). If the measurements were made with a wattmeter, the calculation would be dB= 10*log(P2/P1). The reason for this is that when calculating power from voltage and resistance, the voltage is squared and divided by the resistance (P=E**2/R). When simplified, the equation becomes 10*log((E2/E1)**2), which is 2*10* log(E2/E1), because log(n**2)=2*log(n). The bottom line is that 10 dB is always 10 dB, no matter how you measure it.
Dave Bushong
Don replies: Thanks Dave. You are absolutely right, dB is dB. My meaning was that Magnitude Squared is the way the audio engineer looks at the spectrum while most electrical engineering folk prefer simple magnitude measurements. And I made a point of it because it can become confusing. If we compare a Bode plot of a spectrum plotted in magnitude we will find that the -3dB point is at 0.5 of the magnitude, while -3dB on a magnitude squared plot will be at 0.707.
Actually, multiplying by 20 (magnitude_squared_db=20*log(out/in) is equivalent to squaring magnitude_db= 10*log(out/in) because we are dealing with exponents -> we get magnitude squared of the transfer function by multiplying the output of the transfer function by its complex conjugate; the magnitude is simply the output of the absolute value of the output of the transfer function.
Magnitude squared does not reduce to simple magnitude, and dB is always dB, but its context needs to be understood.
DDJ
