Breaking the Sound-powered Barrier?

Greetings and salutations. Y'all are an incredible group of people! Although I doubt I'll ever be able to read through all the relevant posts, I'm fairly confident that what I propose will be widely viewed as heresy. Well, go get a large balance and a duck[MMP], if you must, but I recently stumbled across all the evidence I need to convince me it's worth trying to make a headset for crystal radios that outperforms vintage balanced-armature headphones.

Now, the obvious route would be to re-engineer the proven balanced-armature technology using modern design tools, components,and materials. As I understand, magnets have been improved some orders of magnitude over the past few decades, and many smaller gains in other applicable technologies could vastly improve the performance of the original design, especially if optimized for crystal radio use. But as worthy as this endeavor would be, I think it is beyond the reach of all but the most resourceful and dedicated of hobbyists.

So I've been doing some fairly extensive experimenting with a much more accessible technology: piezoelectric elements. The results have been extremely encouraging. This inexpensive little metal disc is about as close to an ideal acoustic transducer for a crystal set as one could hope to find. It has inherently high impedance, is very efficient, is as close to massless as one could reasonably expect, and performs best at low power levels.

Why can't we just go out and buy a good piezoelectric headset now? I am speculating here, but I suspect it is because the combination of low power and high impedance has little application in the commercial headset market. (The sound-powered designers didn't pick 600 Ohms at random; this likely matches the characteristic impedance of their intended transmission line.) [EDIT: The transmission line would need to be many kilometers long for reactive line losses to be significant at audio frequencies.] But enough history. How can we hobbyists outdo sound-powered performance with piezoelectric elements? Please take a look at the following specifications and illustrations for the ceramic receivers KBT-33-RB-2CN excerpted from a product catalog[AVX] that also lists the blue Kyocera KBT-44SB-1A buzzers some of you have tested.

Illustration of Ceramic Receiver Specifications of Ceramic Receiver Frequency Response of Ceramic Receiver

First, notice the frequency response graph. It rolls off high frequencies a little earlier than I'd like, but how did a piezo element get such a nice, flat graph? Some of you may really like (and others may despise) the answer: impedance matching. Specifically, in addition to high electrical impedance, piezoelectric elements have high acoustic impedance. If not properly matched to their load (in this case sealed to an artificial ear), they don't work very well. If this sounds familiar, it should. The blue Kyocera units, though they work just dandy as buzzers, suffer from an impedance mismatch between the transducer and free air. Their blue plastic acoustic enclosures effect only a partial match. That even when mismatched like this, these blue buzzers are reported to be as sensitive as good vintage magnetic units and poor sound-powered units is very promising. My experiments confirm that effecting a better impedance match results in vastly improved performance. I used a good-sized exponential horn to match the element to free air, and though there is research showing that people are better able to distinguish sounds reproduced by a pair of speakers in a soundproof room, the use of headphones is vastly more practical for most crystal radio hobbyists.

Now, this nice, flat 107 dB(SPL) response of the ceramic receiver is characteristic of a good impedance match. Considering 1 kHz as representative (as is conventional), the power needed at 1 Vrms to produce 107 dB(SPL) in two headphones is 2*V^2/R = 2*1^2/2800 = 0.714 mW = -1.5 dBm. How does this compare to good sound-powered headphones? As is often the case, Ben Tongue, master of crystal radio research, in one of his excellent foundational articles, provided enough information about the power needed to run a good sound-powered set to make a rough comparison, provided we make one assumption. (Please, Mr. Tongue, accept my apologies in advance. I was trained as a civil engineer to make 'worst case' assumptions and mean to cast no aspersion nor to imply any disrespect.)

Mr. Tongue performed a good approximation of a speech intelligibility threshold test to find the lowest power level needed to operate a good set of sound-powered headphones. The test he performed implies the sound pressure level (dB(SRT)) he was listening to should correspond, within 5 dB(SPL) to his hearing threshold. Mr. Tongue reports “poor” hearing acuity at the time of the test, but does not admit to suffering clinical hearing loss (20 dB(HL)). Therefore, assuming the sound pressure level produced was at most 17 dB(SPL) he reported these good sound-powered units required -84 dBm of power to produce at most the assumed 17 dB(SPL). The ceramic receivers should require only about -91.5 dBm to produce this same 17 dB(SPL).[BT][DB][HAIN]

[EDIT: This wasn't clear. 17 dB(SPL) is 90 dB less than 107 dB(SPL). The amount of power required to produce 17 dB(SPL) will be 90 dB less than the power required to produce 107 dB(SPL). -1.5 - 90 = -91.5 dBm.]

These ceramic receivers, when properly sealed to the ear as indicated in the product literature, should require, at worst, 7.5 dBm less power to produce the same sound level as good sound-powered headphones also used correctly. In addition, at 5600 Ohm in series they are a better impedance match to a crystal radio set than the 1200 Ohm presented by a pair of sound-powered units.

Finally, please take a look at how these ceramic receivers seem to be constructed. I was planning to get into it in more detail, but I have probably already gone on too long. So briefly, after figuring out a better geometry for crystal radio use (even higher impedance and either narrower or wider frequency response), a nice metal acoustic chamber similar to the product shown could be made by anyone with modest metalworking skills from a disk of metal a bit larger than the piezo element, two washers, a small perforated dish, and some screws. An effective earpiece could be carved from a suitable suitable material or may even be made (like a quilt) from a piece of dense closed-cell foam with a smooth coating (like wet suit material or that gel stuff). Even without access to a hobby-sized machine shop, a very good pair of phones could almost certainly be made from appropriate parts in your favorite line of plumbing supplies.

Anyway, I do look forward to comments, corrections, criticisms, and grotesque ad hominem attacks. (Well, maybe not the latter so much.)

Cheers! ~David

P.S. There seem to be a few thousand of these ceramic receivers still in the supply lines, and Newark lists them as a factory direct order with a reasonable minimum. A suitable substitute should have similar performance. I have not yet gotten my CFO's approval for the quantity order, so if anyone beats me to it, count me in for at least dozen!

[MMP] (Mandatory Monty Python reference.)

[AVX] “Piezoelectric Acoustic Generators” by AVX and Kyocera. (If this PDF file goes away, I'll post a copy of it elsewhere.)

[BT] “A New Way to look at Crystal Radio Design. Get Greater Sensitivity to very Weak Signals, and Greater Volume, less Audio Distortion and Improved Selectivity on Strong Signals” by Ben H. Tongue.

[DB] “Decibel Table - Comparison Chart” (It may look like I'm playing fast and loose with decibels, but dB(SPL), dB(HL), dB(SRT), dB(SIL) and dBm all correspond. If I made a mistake, please correct me!)

[HAIN] “Hearing Testing” by Timothy C. Hain, MD Page last modified: January 5, 2008

Personal Tools