Q: What is a Planar Magnetic Transducer ?
A long strip of thin material is impregnated/coated with several conductors which act like the voice coil on a regular speaker. This assembly, typically a kapton or polyester film with aluminum strips, is suspended between continuous rows of bar magnets. When energy passes through the ribbon/voicecoil sandwich, it is repelled and attracted by the surrounding magnets. A planar magnetic driver is capable of seamlessly reproducing almost all of the sound spectrum, from highest highs to mid bass, almost 8 octaves. This eliminates crossover distortion, phase distortion and avoids the different "voicing" of the various cone speakers necessary to cover the same frequency range. The voicecoil has typically no reactive component which makes it an easy, almost pure resistive load for any amplifier. This contrary to cone speakers and ESL's. Impedances are in the 2.5-12 ohm range
Q: What is a "Real" Ribbon Transducer ?
A "real" ribbon driver has a "voicecoil" made of a aluminum strip or strip assembly suspended between continuous rows of magnets. The principle is the same as a planar magnetic transducer other than the ribbon being stretched between the extreme points. A planar magnetic has its voicecoil assembly stretched along the short axis, typically horizontally. This makes the planar transducer mechanically sturdier than a real ribbon. Real "ribbon" drivers boast an extended performance in the high end of the audio spectrum. Planar magnetics show a drop off in the high end caused by the film material weight, mechanical properties and also subject to the horizontal tensioning forces. Because of the "fluttering" suspension, real ribbons need practically a higher low end crossover point as compared to planar magnetic drivers. Impedance of a ribbon transducer is mainly resistive, similar to a planar magnetic driver. True ribbon transducers however have an extremely low impedance, typically between 0.1 and 0.4 ohm, requiring a transformer to be useful with a practical amplifier.
Q: What is an ESL ?
An ESL is an ElectroStatic Loudspeaker. It consists of a thin sheet of material between two charged grids. When fed a sufficient high voltage, the sheet will push and pull between the grids because of the electrostatic forces. Think of it as a large capacitor. The ESL has a need for a high voltage DC powersupply to charge the grids. The high voltage to drive the sheet is typically generated by a step-up transformer. This results in a highly reactive load to the amplifier. Special amplifiers have been designed with 1 to 1.5 kVolt outputs to drive ESL's directly.
Q: What is a push-pull driver ?
All of the above drivers can be built in a "single-sided" or in "push-pull" version. In a push-pull realization the moving diagfragm is attracted to one side and simultaneously repelled on the other side. This mechanism allows for greater linearity since errors are averaged out. Even harmonic distortion is greatly reduced by this mechanism.
Q: What is the "cavity resonance" ?
The cavity resonance in a planar magnetic driver is caused by the construction of the driver. The film is suspended between double rows of magnets which are typically glued to a metal plate. This assembly is sandwiched together. The cavity formed by this assembly provides gain at some frequency. The gain can be fairly large, 6-8 dB seems to be the norm. Q varies from 1.5 to 3-4 in extreme cases. I recently found out that this resonance can be influenced by damping. It is mandatory to "dampen" this cavity resonance in order to protect the ribbon from excess excursion and to restore the tonal balance. The effect is definitely audible !
Q: Why is there a notch filter in series with a planar magnetic driver ?
A passive notchfilter, made out of a parallel combination of a resistor, capacitor and inductor, is the standard way to tame the cavity resonance. The graphs below show the action of the notchfilter as supplied with the B&G RD75 planar magnetic driver. Although a passive notch filter is not a perfect way of eliminating the cavity resonance, it does an adequate job. Since the cavity resonance peak is asymmetric, an active notchfilter, as used in the Clearview active crossover with equalising, does a better job in attenuating the peak under different circumstances while restoring the symmetry in the response. For an overview of passive notch filters for the Radia series of drivers, please take a look in the passive crossover section of the FAQ.
Q: What is a dipole ribbon or baffle ?
A dipole ribbon or dipole baffle is a planar magnetic, ribbon or ESL driver mounted in a baffle, typically of rectangular or trapezoidal construction. As such the driver is allowed to radiate energy in the front and rear direction. This results in soundwaves emanating in a figure 8 pattern. There is hardly any energy emitted to the sides due to quarter wave dipole cancellation. As such there are also hardly any reflections of the side walls in a typical listening environment. Care must be given however to room placement as reflection of the backwall will interfere with the front wave. A minimum distance of 6 feet of the backwall is recommended. As such the attenuated and dispersed backwall reflections reach the listener 10 milliseconds later than the frontwave, creating the famous "dipole" ambiance. Magic with large scale sound stages.
Q: What is a monopole ribbon ?
A monopole ribbon or planar magnetic is a by construction dipole ribbon transducer installed in a sealed cabinet. The backwave is contained and absorbed by suitable stuffing in the wave trap. This construction eliminates the backwave reflecting off the wall and destructively interfering with the frontwave in those situations where the minimum distance between ribbon and backwall cannot be guaranteed. This is a compromise between "dipole magic" and ease of placement in smallish rooms. It is also a good choice where pin-point accurate small scale sound imaging is desired.
The drawing below shows a cross section of a monopole enclosure with a that could be used for a RD75 or RD50. Important is the space behind the driver. A cross section of about 50 inch2 is required. I prefer to line the walls with 1/2 inch open cell foam and fill the remainder of the cavity with hollowcore fiber for trapping of the rear wave. The idea is to contain the backwave without letting it bounce back through the film. The dimensions of the enclosure are not critical as long as sufficient space is maintained for backwave trapping.
Drawing courtesy Douglas Stabler.
The schematics below show a 3th order Butterworth and a 4th order Linkwitz Riley passive crossover implementation for the different BG Radia planar magnetic ribbons. The passive notchfilter is in both cases the same. The passive notch filter components are indicated by Cn, Ln and Rn.
Although this passive implemetation is adequate for protection and cavity resonance suppression, it does not take care of flattening the ribbon response in critical areas.
All capacitor values are in microfarad. The inductors are indicated in milliHenry and the resistors are in Ohm.
FAQ graphs and measurements:
While I was preparing this FAQ, I took some measurements to illustrate the cavity resonance and passive and active notch filter responses. As such I discovered that the cavity resonance center frequency and Q is different for a dipole and monopole setup. Probable cause for this is the extra damping the membrane gets in the monopole set-up.
Above graph illustrates the cavity resonance peak in a RD75 mounted without a baffle. The centerfrequency of the peak is at 5250 Hz. The supplied passive notchfilter is tuned for this frequency. It does an adequate job at damping this resonance. Notice the overall insertion loss this passive filter causes. The effects can be seen over the full audio spectrum. As a side note, the peak at 1031 Hz correlates well with the simple dipole model as described by Linkwitz. The acoustical path of a BG ribbon without a baffle is approximately 6.5".
The graph below is the same RD75 driver mounted in a monopole housing. Walls are foam lined, the remaining cavity filled with hollow core Dacron II and 2 inch thick foam behind the driver. Earlier waterfall displays have shown that this combination yields good ribbon damping as proven by the lower distortion in this set up. As a total surprise we see that the cavity peak shifted up with a significant increase in Q. The peak remains asymmetric. The passive notchfilter is not adequate anymore to eliminate this peak as it is tuned to the "wrong" frequency ! The mismatch is obvious.
The above measurements were taken in a 12.3 milliseconds timewindow and analysed with a 1024 FFT. Display smoothing was set as usual at 1/3 octave. Distance microphone/driver is 12" @ 1 meter high.
To be continued
Copyright (c) 1997-2005, by Rudi A. Blondia, ALL RIGHTS RESERVED. Last update: April 12, 1998.