Impulse vs. Sine Sweep Response

On the Shoulder of Giants – By Don Davis

What’s the best way to get room data? Some advocate impulsing the room with balloons or starter’s pistols. Others prefer using swept sine waves. Don sheds some light on this old argument with some reminders about each method’s susceptibility to amplitude non-linearities.

Impulse responses are akin to using AM transmissions in the midst of a severe electrical storm. Using swept sine wave with a tracking filter is akin to using FM transmission in the same severe electrical storms.

The impulse response is pure amplitude and subject to non-linearity. The swept sine wave is full power at all frequencies and has the immense benefit of being sinusoidal, hence traceable through complex systems for time, phase, and amplitude.

Early radio used the “spark gap” transmitter, and in my early youth, prior to WWII, I witnessed a demonstration by the Ham Radio club at Purdue of a “rotary spark gap” transmitter that in addition to its transmission blanked out everything else on the air.

Reginald Fessenden (1866-1932) was a professor of Electrical Engineering at Purdue in 1892, and later made the first audio transmission of a voice signal using a rotary spark gap transmitter in 1900. He then went on to develop in conjunction with Ernst F. W. Alexanderson, who worked for GE, a series of high frequency alternators for use at radio frequencies. “Fessenden felt that, ultimately, a continuous-wave transmitter – one that produced a pure sine wave signal on a single frequency – would be far more efficient, particularly because it could be used for quality audio transmissions.”

Such high frequency alternators remained in use, especially by the United States Navy, until 1948 on long wave frequencies.

Ernst Adolph Guillemin’s (1898 – 1970) explanation on why the sine wave is the preferred wave form is, “When an electrical network of linear elements is excited by a voltage or current source that is sinusoidal time function, the resulting stationary voltages and currents in all parts of the network are likewise, sinusoidal time functions, differing from each other, and from the excitation function at most in their respective amplitudes and time phases…. The network may have a structure of unlimited complexity (i.e., a room) still the voltages and currents everywhere in this entire system are time functions which in essence are identical in form, namely sinusoidal. No other periodic time function can claim such distinction.”

Fourier, Heyser and many others have recognized these facts to the advantage of analysis.

To achieve linearity and adequate amplitude over a usable frequency span via an impulse is indeed a theoretical “tour de force,” but a real-world nightmare. An impulse response generated via a swept sine wave is a known quantity. Impulses transformed into frequency are rightly called estimates.

Yachting canons, potato canons, starter pistols and all the ingenious ways to harm hearing, should be confined to use by those who still use spark gap transmitters and crystal detectors in their Ham Radio – illegally or outside the U.S. dbd

From Don’s notes, probably BSTN Archives