Plate reverberators were introduced in 1957 with the EMT140. This was touted as a compact alternative to a reverberation chamber. This is true, as long as you consider a 400 pound twin mattress-sized box to be “compact.” By 1957 standards, it was compact enough, as several plates could be put into a single room and controlled remotely.
The principle behind a plate is fairly simple, yet totally brilliant:
- A 1 meter x 2 meter x 0.5 mm rectangular steel plate is suspended via tight springs in a steel frame.
- The signal is injected into the steel plate via an “electro-dynamical actuator,” moving transversely to the plate (i.e. perpendicular to the plate).
- One or more piezoelectronic pickups are attached to the plate at different locations. By using two pickups, a wide stereophonic image is obtained.
- The reverb time is controlled by a felt damping plate, that is placed in parallel to the plate without contacting it. The closer the damping plate gets to the steel plate, the shorter the decay time is.
Plate reverberation has some strange characteristics, due to the unique physics of a plate. This has been described in detail by Jonathan Abel, Kevin Arcas, Stefan Bilbao, and other smart academic folk. A high level summary of some unique plate properties:
- Dispersion. This means that the speed of sound is different for different frequencies, with high frequencies having a much faster speed of sound than lower frequencies. This results in an audible “PEW!” sound [read that as a laser gun sound] for the attack of the sound. At least in some plates. Other plates have a less audible “PEW!”
- Instant onset of reverb. No fade in at all. Not even a subtle fade in. We are talking instant. There is no build up time.
- Instant onset of high echo density. This is partly due to the fairly small size of the plate, and largely due to the high speed of sound at high frequencies versus low. The echos are distorted by this speed of sound, to the point where the high frequency echos will have a bunch of reflections by the time the low frequencies have their first reflection. To our ears, this sounds like a reverb that is instantly “fully mixed.”
- Frequency dependent decay time. Surprisingly enough, a real EMT140 is fairly dark. At 10 kHz, the decay time on an EMT140 is <1 second, at any setting of the decay damper. Conversely, the very low frequencies can end up booming out. Stainless steel plates have a longer high frequency decay time, and a brighter sound overall.
- Constant resonance density versus frequency. This is different than a real world 3D space. However, this is how most digital reverbs behave.
- Not necessarily metallic sounding. You’d think that a reverb made out of a 3’x6′ piece of cold rolled steel would sound, well, like steel. However, a well tuned EMT140 is super smooth. There is a slight amount of metallic sheen for high frequencies, but the sound overall is less metallic than most digital reverbs.
So, a plate has dispersion, instant echo density, instant onset of reverb, frequency dependent reverb time. What are the implications of this for music production?
During the research for ValhallaPlate, I experimented with various models, where I could dial in the amount of dispersion over different frequencies. It turns out that the “PEW! PEW!” sound of dispersion isn’t all that important to the overall sound. Some plate impulse responses had a noticeable amount of PEW! artifact, but real word plates tend to have less of this sound. The algorithmic models that had more of a “PEW!” sound were kind of annoying.
I found was that dispersion, when properly tuned, has a huge impact on the stereo imaging of the reverb. My current theory on what is going on:
- The pickups in an EMT140 are staggered off-center, and are at different distances from the input signal transducer.
- For high frequencies, these distance differences translate into very slight time delay differences of the reverb onset between the left and right channels (i.e. less than 1 msec). The slight time delay causes the reverb to “lean” slightly to one side, due to the Haas effect. This is pretty subtle, both in physical plates and in a properly tuned algorithmic model.
- As the frequencies get lower and lower, the time difference for the onset of reverberation between left and right channels gets wider and wider. For low frequencies, this difference can be in excess of 10 msec.
My theory is that this time difference in the left and right channels for low frequencies contributes to the depth of the stereo image. A plate reverb doesn’t just sound like it is wider than the speakers – the sound appears to come from behind the speakers. It is a big, 3D sonic image. The small time difference between left and right channels for high frequencies helps to keep the reverb from sounding “mushy” or overly diffuse. This is important for sharp transients, such as percussion.
The instant echo density and instant onset of reverberation in a plate makes it a reverb that can be used on pretty much any source. Different digital reverb algorithms are often described as being optimized for different source signals, i.e. “Percussion Plate,” “Vocal Chamber,” etc. This is due to the artifacts of different digital algorithms, and the tradeoffs between echo density and metallic sound. A physical plate reverb just sort of works on everything. Drums, vocals, strings, guitars, whatever.
The frequency dependent reverb time of a plate models what happens in a physical space. In a large space such as a hall, the low frequencies will have a considerably longer decay time than the mid range frequencies. High frequencies in a physical space will always be damped by the air, so that the maximum RT60 at 10 kHz will never exceed around 1.25 seconds for any physical space. In this sense, a plate reverb effectively mimics the “real word.”
This frequency dependent decay time also makes it easy to mix with a plate reverb. The decay time parameter of a plate (which is either controlled via a remote, or by turning a huge wheel on the plate itself) is essentially the RT60 at some arbitrary midrange frequency, somewhere between 2 and 3.5 kHz. This midrange decay time is useful for dialing in the decay on those parts of the input that are audible as pitch. The high frequencies always have a fairly short decay, so consonants and sibilance won’t ring out for too long. The result is a reverb that has plenty of zingy brightness, but without the sibilance turning into annoying hiss, which can easily happen with digital reverbs.
Later plate reverbs, such as the Ecoplate developed in the 1970s, used stainless steel, versus the cold-rolled steel used on the EMT140. The goal was to get a longer decay time for higher frequencies. This seems in keeping with the general trend of the 1970s towards brighter sounds, which can be attributed to higher spec recording gear, the arrival of early digital units, and cocaine abuse [note from editor: you might want to clean this up before hitting PUBLISH]. Today, the darker sound of the EMT140 is still highly sought after. The short decay time of high frequencies means that an EMT140 can be inserted into a mix without being obtrusive. It just works.