In 1990, an experiment was performed by this author while Chief Engineer of WWMX-FM in Baltimore, Maryland. Without the knowledge of anyone, an all vacuum tube gain control amplifier was installed in the on-air audio chain. The following is a summary and result of the experiment.

Although the radio station audio was in stereo and had a clean sound, it lacked realism and depth, something that this author remembered from the mono Hi-Fi systems of the 1950s. The studios and audio chain were all analog from music source to transmitter. After considering any differences in equipment configuration, it was decided that the primary difference was the use of vacuum tubes back in the 1950s.

Using a few decades of vacuum tube experience, a project was started at home to build a vacuum tube gain controlled amplifier, more commonly called a compressor. Audio compression is used by most radio stations to maintain a steady average volume. The design was all 12AX7 triodes, including a gain control stage; triodes were selected because of their second and third harmonic characteristics. The reason for building a gain-controlled amplifier rather than just a simple buffer amplifier was for loudness. One of the pitfalls of radio broadcasting is the fact that every station manager and program director wants to be the loudest station on the dial. This usually results in a lot of clipping and processing of the audio, resulting in a harsh sound. Using a triode as the control stage requires controlling grid bias and varying stage gain. Using grid bias to control gain has about a 30 dB useful range, sufficient to maintain an average level. As the triode goes farther into biased gain reduction, it produces increased second and third harmonics. The second harmonic adds warmth to the audio, while the third adds loudness. The triode increases loudness without clipping while adding warmth, resulting in a much cleaner sound.

In order to get honest listening results and see if anyone would notice any difference, the new gain-controlled amplifier had to be installed without anyone's knowledge. One night after midnight, the all-tube gain-controlled amplifier was secretly installed in front of the existing Optimod 8100A processor. The Optimod was set so its internal broadband input compressor did very little processing, instead letting the vacuum tube gain control do all the broadband processing. The high frequency clipping was also reduced.

The next day, listening at home, the sound of realism that had been missing could now be heard. It was a subtle difference, with the addition of second harmonic content generated as the tube processor performed gain control. Listening also revealed that although the amplifier was controlling gain, louder passages still sounded louder even though the actual level was being reduced. This is attributed to the extra harmonics adding loudness.

A few days went by, and then compliments started coming in on how good the radio station sounded, including calls from other radio station engineers. One day, a music consultant once employed by the station walked in and said he was driving through town and wondered what we were doing that sounded so unique. Opening the back of the equipment rack, his mouth dropped open when he saw all the glowing vacuum tubes.

Not only was this a success, but it was a big success as the radio station's ratings climbed and beat out most other stations in listenership. Higher ratings mean more ad revenue for the station. There is no doubt that just adding the use of vacuum tubes improved the sound such that more people listened longer.

For those interested in the all vacuum tube gain controlled amplifier circuits, a PDF download is available farther down on this page.

Consider building a lower-power system. Components such as transformers and output tubes used in lower-power amplifiers are significantly less expensive. Besides costing less, lower-power transformers are physically smaller, requiring less chassis space. Tube amplifiers with lower-power output draw less current and have lower electrical utility costs.

Amplifier power output does not affect sound. An amplifier with 10 watts of output power will sound the same as a 100-watt amplifier. The difference is loudness. Rather than using amplifier output watts as a gauge for loudness, use sound pressure levels (SPL).

Sound pressure levels above 85 dB are high enough to cause some hearing loss, depending on exposure time. The higher the SPL, the greater the chance of hearing loss. At an SPL of 100 dB, you really should be wearing ear protection. At 120 dB without ear protection, the chance of hearing damage is high, even for short periods of exposure. At 140 dB, ear damage will most certainly happen without ear protection. If you would wear ear protection out in the environment, then why would you want such a high SPL in your listening room? Unless, of course, you plan to wear ear protection.

The sensitivity of speakers is usually specified as "so many dB @ 1W/1m" (1 watt at 1 meter) from the speaker. For example, 93 dB @ 1W/1m. Higher sensitivity speakers are more efficient and require fewer amplifier watts for a specific sound pressure level. A pair of speakers with a sensitivity of 96 dB @ 1W/1m requires less amplifier power than speakers with a sensitivity of 93 dB @ 1W/1m. Assembling an audio system using 85 dB as a reference for maximum SPL will provide substantial loudness. If paired with the right speakers, 12 amplifier watts can provide an SPL of 85 dB 20 feet from the speakers.

On crownaudio.com, you can use the 'Amplifier Power Required' calculator to determine amplifier power output. The primary factors are the desired SPL level at listening distance from the speakers, and speaker sensitivity. Enter the listener's distance from the source (speaker) in meters. If 20 feet, then enter 6.1 meters. Enter the desired SPL level in dB at the listener distance, for example, 85. Enter the sensitivity of the speakers, for example, 93 (93 dB @ 1W/1m). For amplifier headroom, enter 3 dB to accommodate audio peaks. Entering those parameters, you should find that it only requires 12 watts of amplifier power.

For music, a normal SPL listening level is around 70 dB; 80 dB is loud. Sound pressure levels above 85 dB can cause hearing loss, especially with long-term exposure. A home audio system designed to produce an SPL of 80 dB to 85 dB at ten feet (or twenty feet in larger rooms) from the speakers should be sufficiently loud. Over time, you will save money if you pay a little more for more efficient speakers and drive them with a lower-wattage amplifier. Besides the savings in amplifier costs, lower-power amplifiers draw less current, which means lower electric power costs. Vacuum tubes used in lower-power amplifiers are less expensive, a savings each time you replace tubes.

More information:
Noise-Induced Hearing Loss
Hearing loss and music
Loud Noise Can Cause Hearing Loss

All capacitors work on the same principal, two plates separated by an insulator (dielectric) that blocks DC while allowing AC on one plate to appear on the other plate; audio is a form of AC. In amplifier audio circuits, capacitors are used in various places along the audio path. From a technical point of view, you would think that something as simple as two plates separated by an insulator would all perform the same and sound the same.

Not necessarily. It is not unreasonable to say that some capacitors may perform better than others depending on the materials or manufacturing process used. Rather than depend on statements of performance stated by individuals or component suppliers (after all, they are trying to sell you something), technical testing with comparison of different types of capacitors would be more definitive.

A Texas Instruments Analog Design Journal article, Selecting Capacitors to Minimize Distortion in Audio Applications, written by Zak Kaye, Applications Engineer, provides insight. If interested, the article is available here. Test results presented in the article are applicable to use in audio amplifiers.

The article primarily deals with solid state circuit voltages. Keep in mind that it is the AC performance of a capacitor that is important. The voltage rating of a capacitor depends on the circuit the capacitor is used in. Also, the article considers tiny surface mounted capacitors used on printed circuit boards. Capacitor size is inconsequential, it is the type of capacitor that matters. For vacuum tubes, larger sized capacitors are used, usually with wire leads for point to point wiring.

On the conclusion page, there are two types of capacitors recommended for lowest distortion. The C0G/NP0 (multilayer ceramic) and film (polyethylene or polypropylene) type capacitors. Multilayer ceramic capacitors with C0G/NP0 dielectric plus long wire leads and high voltage ratings are difficult to find.

Capacitors Across Pot Terminals

However, C0G/NP0 multilayer ceramic capacitors with short leads are useful for close terminal connections such as across pot terminals. Polyethylene or polypropylene film capacitors with high voltage ratings and long wire leads are available for normal point to point wiring.
Sample of suppliers:
Antique Electronic Supply
Amplified Parts
Just Radios Capacitor Order Form

In Zak Kaye's article, he mentions that to reduce its distortion, increase the value of a capacitor until its impedance is low enough in the band of interest (for instance, 20Hz to 20,000Hz). It's a matter of calculating the –3 dB cutoff frequency of the capacitor-load impedance to the lowest possible value. For instance, a .47uF capacitor will have a lower –3 dB cutoff frequency than a .1uF capacitor. It would also be wise to voltage rate capacitors with a generous safety factor. For example, a 600 volt rated capacitor in a 400 volt circuit. If on a tight budget, low-cost polyethylene or polypropylene film capacitors are available. With careful circuit design, they should perform as well as any other polyethylene or polypropylene capacitor.