DIY Reverb Pedal Circuit: Spring-Hall Reverb with PT2399 IC
Introduction to Reverberation
Before getting into the design of DIY reverb pedal circuit, lets see the concept of reverberation first. Real reverberation happens in a performance chamber when the walls and every object inside reflects the sound back and forth. As the sound bounces back and forth in three dimensional space, the reflections creates more and more complex pattern. The natural process shown in the Figure 1 is the illustration of reverberation. In the old days, the only way to reproduce a reverberation effect is by using a real reverberation chamber.
For good effect, someone should build a large room with complex geometry with carefully selected wall material. After that, he should place some loudspeakers at certain location and place a microphone at another location inside the chamber. The first attempt to mimic a room reverberation without a real reverberation chamber is done by a spring reverb tank (see reference ). Figure 2. shows the basic construction of spring reverb tank.
The audio signal drives an input coil that vibrate the spring at one end. The created vibration is then travels to the other end and bounce back and forth between two ends. During it’s travel, the amplitude is decreasing until deceasing. Because of the different traveling speed of high and low frequency waves, the reflection produces more complex waves. In addition, the spring joints adds more complex reflections in the middle of its traveling path. Moreover, different springs arrangement with different characteristics would produce rich and unique resonant characteristics. Finally, a pickup coil picks the reverberated sound at the other end of the spring. The pick-up then returns the signal to the amplifier or the processing circuits.
Digital Modelling of Reverb Effect
Many expert has researched reverberation effect processing in digital domain, and (in my opinion) can be classified on several mainstream methods. Firstly, system response reproduction: this method treat the modeled system as a black-box. However, we don’t care about what happen inside this box. All we need to do is just measuring the output response after we apply an impulse input. After getting the response pattern, a digital processor compute the convolution of the pattern and the input. and use it to modify the signal by convolution processing (see reference ). Whether the modeled system is a real concert hall or a real spring, it will just work. It would be very simple to implement but the convolution processing would require very high processing power to run.
Secondly, physical modelling: this method analyze the physical process of the modeled system simulate every physical process in the computation. This produces very realistic realistic sound but the computational cost could be high. However, it depends on the optimization or mathematical simplification of the model. For example, you can see one example of spring reverb modelling at reference .
Lastly, synthetic modelling: this model construct a computation model that produce a (perceived) similar effect on the audio output. As long as the perceived result is similar, the computation model might not be related with the physical model. Sometimes I see such model is just an oversimplified model of the physical. Similarly, approximation of the system response by trial and error computation experiment might produce such effect as well. For example, a Schroeder reverb [see ref ), can be adjusted to mimic the medium-size hall reverberation by setting up some parameter to certain values.
Electronic Circuit Implementation of DIY Reverb Pedal Circuit: Delay Line Network and Tuned Analog Resonators?
After we analyze that the reverberation phenomenon as the complex echo pattern, we know the principle is simple. Therefore, we can intuitively build such reverb effect circuit using some delay line networks. On the other hand, if we analyze the reverberation phenomenon as a continuous resonance, we might think differently. Therefore, we can think that multiple parallel analog resonators (with different frequency tuning) could generate such effect.
I have been thinking of this analog resonators for years since I learned electronics. It is well known on the oscillator subject, where many of oscillator designs are basically an unstable resonator. Please let me know if there is already analog reverb circuit design which is based on analog resonators. Please write a comment if you find it, so I won’t be reinventing the wheel someday. For the moment, let’s focus on the delay line circuit solution. We will use it for the reverb circuit design, since it has been proven in such applications.
PT2399 Digital Delay Line IC Chip, Low Cost Solution for DIY Reverb Pedal Project
In the old days, the bucket-brigade device is very popular. It uses switched capacitors array to store audio samples in analog voltage is for implementing a delay line function. With the advanced CMOS technology, PT2399 IC from Princeton technology is getting popular. Because of its low cost, it replaces almost all BBD chips. Figure 3. shows the block diagram of the internal circuitry of PT2399 IC chip.
This digital delay line chip is available in 16-pin DIP package, so it will be easy to be handled. Certainly it would be the preferred option for do-it-yourself projects. The minimum delay length is 30 ms and the maximum is 340 ms. This delay setting is easily adjustable by selecting a proper value of an external resistor.
DIY Reverb Pedal Circuit for Simulating Spring and Room/Hall Reverb
I have created very simple reverb circuit that can simulate spring and room reverb effect. It use 5 delay line integrated circuit chips, and the block diagram is shown in the Figure 4. It has decay time control, room size control, and dry/wet balance control. When we set the room size to minimum, it would sound similar to spring reverb. On the other hand, when we adjust it to maximum then it would produce hall or even cathedral-like reverb.
Reconfiguring The Delay Blocks Structure
If we take a look at Figure 4, we can see that the first delay block can’t be set to 0 second delay since the minimum delay for the PT2399 chip is about 30 ms. The result is that the total shortest delay is 61 ms, and this is not good for simulating soundwave wave propagation inside a spring. By reconfiguring the delay blocks as shown in the Figure 5, we can achieve the shortest delay about 30 ms while minimizing the component count by using only 4 delay chips.
The Complete Circuit’s Schematic Diagram
A testing in digital domain (using Deepstomp platform) has shown that the basic block diagram of the circuit works fine. For the actual hardware, here is the schematic diagram of the reverb pedal circuit (Figure 6). It has four controls: decay time, room size, dry/wet balance, and the output level.
You can see in some component values are missing, and that’s because the final circuit is still under development. We will update with the final value right in this page when the circuit has been successfully tested. You can see that the missing values are primarily on the mixer and filter parts, but you can figure it out from the gain, amplitude range, and the frequency response assumptions to try some values if you want to experiment yourself for finding the best component values.
Update (07/13/2020): Restructuring The Schematic Diagram and Adding Some Improvement
The schematic diagram shown in the Figure 6 is hard to read, making it hard to validate and analyze. Now I have redrawn the schematic with some fixes and additional features, see it in the Figure 7. The part list in shown in the Table 1.
The schematic diagram is now added with signal flow marking for easier analysis. The circuit is optimized for the lowest component count. Here some points which are probably speculative and need more testing and experiment:
- There is no documentation on the op-amp’s open-loop gain and input impedance spec specs of the free op amp section (pin 13-14). I just use 100k resistor for setting up gain and mixing impedance resistors at IC1_2, while example circuit in the datasheet uses about 10-15k resistors. By using 100k resistor, the input impedance of this PT2399 reverb circuit is around 50k, determined by the R3 in parallel with R26-R28 series. The use of high resistance for this circuitry is to avoid external op-amp addition, so the components cunt would be minimal.
- To avoid many dc decoupling capacitors, most internal connections between IC uses DC coupling (only after MIXER-2 has a decoupling capacitor). This may lead to offset error since each IC has it’s own internal references which might have some differences from chip to chip. It is assumed that the difference is small enough and got no much amplification since most of amplifier stages is set around unity gain.
- To avoid redundant filters for input sections of the PT2399 delay lines, all the delay lines uses the filtered signal tapped from Delay-1 input filters output. It is assumed that the output connection in this point is not affected by the loading effect of the internal delay line connection inside the IC-1 chip.
- To avoid the redundancy of the output filter, the filtering of the Delay-1, Delay-2, and Delay-2 outputs is done after mixing. This after-mix filtering is done by Delay-4 input pre-filter.
That’s all what I can say about the progress, and be warned that his circuit is not yet tested. I will publish the result when I have tested it myself, but after understanding some of the assumption I think you can try it yourself to test the circuit and discuss the result here. Any comment are welcome!
- L. Hammond, “Electical Musical Instrument,” U.S. Patent 2,230,836, Feb. 2, 1941.
- Fons Adriaensen, “Acoustical Impulse Response Measurement with ALIKI”, 4th International Linux Audio Conference: LAC2006, http://lac.zkm.de/2006/papers/lac2006_fons_adriaensen_01.pdf
- Stefan Bilbao and Julian Parker, “A Virtual Model of Spring Reverberation “, IEEE Transactions on Audio, Speech, and Language Processing, Vol. 18, No.4, May 2010 p.799,https://www.era.lib.ed.ac.uk/bitstream/handle/1842/3718/BilbaoS_A%20Virtual%20model.pdf
- M. R. Schroeder (Bell Telephone Laboratories, Incorporated, Murray Hill, New Jersey),
“Natural Sounding Artificial Reverberation”, Journal of The Audio Engineering Society, July 1962