Echoplexus Tape Delay
A PT2399-Based Workalike of the Classic Tape Echo
Overview
I know, I know. Another PT2399 delay. Big whoop. But wait, I promise you'll like this one! The Echoplexus is a PT2399-based take on the legendery Echoplex EP-3, complete with discrete JFET preamp. But that's not all. The preamp can be switched out for an opamp-based preamp if you prefer. There is also a “wet kill” function that will allow you to use the preamp on it's on and run the pedal in buffered bypass mode for that mojo and give you trails functionality. The filtering has been painstakingly tailored to get that “tape” sound where both high and low end are rolled off on the repeats. Additionally, a modulation section with two different frequencies is implemented to provide modulation at frequencies that correspond to the physical tape motion of the EP-3. In short, it's not the run-of-the-mill “analog” voiced PT2399 delay. So stick around, you just might like what you see.
How It Works
In concept, the Echoplexus is pretty straightforward: input section, PT2399 delay section with modulation, summing amplifier to output. However, I tried to get some good old fashioned tone out of this thing.
The input section allows for selection of one of two different input sections. The first is a standard opamp buffer stage. This is done because I had an extra opamp and sometimes you just want two different options for tonal color. This buffer is pretty standard.
Echoplexus Input Buffer
One of the more characteristic parts of the EP-3 is the discrete preamp. This is based around a TIS-58 JFET, which is no longer in production. However, the 2N5457 is a pretty good substitute. Another part of the preamp is the fact that it is run at a relatively high voltage of about 25 volts. I decided that this was necessary to make this a more convincing implementation, so a TC1044 DC-DC converter is used to boost the voltage up to 25V (more on this in the power section discussion).
Echoplexus Preamp
Next up is the delay stage. For the most part, this is pretty standard. It has multifeedback low pass filters for both input and output and time control is the standard implementation. However, once the repeats are generated, the go through additional filtering that helps roll off some of the low frequencies in addition to some high frequencies. You will notice the presence of cascaded RC low pass and high pass filters prior to going to mix and feedback controls. The feedback values prior to summing at the PT2399 input are also not the usual. When using caps to roll off some of the low frequency, the overall signal level gets weaker and therefore a smaller resistor is necessary to be able to get feedback that will just get into self oscillation when maxed out.
Echoplexus Delay Stage
Now that we have delay, we need to create our modulation. In a tape delay, there are a couple of different frequencies that can come into play on a machine that is not well adjusted. One is the “wow” associated with slightly assymetric rotation of the tape, resulting in low frequency content. Then there is “flutter” associated with the tape physically fluttering due to an incorrect amount of tension in the tape. Finally, there is “scraping flutter”, a very high frequency phenomenon of the tape scraping over the heads and vibrating, much like a violin bow over the strings. Scraping flutter tends to introduce small amounts of signal content in the 3 kHz region and is not actually a pitch modulation of the delayed signal. For this reason, I decided to use two LFO's, one for wow and one for regular flutter. These LFO's are the simple single transistor topology for simplicity and space savings. They are at fixed frequencies. The flutter frequency is approximately 4 Hz, which corresponds to the tape speed of the Echoplex. The wow frequency is only about 1.5 Hz, which is a limitation of the LFO topology and values. As LFO rate is reduced, the output voltage level begins a steep drop after about 1.5 Hz. I would have liked to have it a little slower, but space precluded a more complicated LFO. I might be able to rework it with larger caps, but we shall see.
Echoplexus LFO
The remainder of the signal path is the summing of dry and delayed signal and output buffering. This is pretty standard stuff, so not much to discuss here.
Echoplexus Output Buffer
Finally, we have the power section. Because there is a charge pump section, there is a little more going on here. We have the standard polarity protection, filter cap, reference voltage, and 5 V regulator that is common to PT2399 delays. However, we also have the TC1044 set up in charge pump mode. For regular voltage doubling, two diodes and two electrolytic caps are used. However, we can continue to replicate the diode/cap configuration to continue to increase the voltage. Note that the voltage will not completely double or triple, due to the forward voltage drop of the diodes. Normally, Schottky diodes would be used here to minimize that drop, but because I want about 25V and not 27, I decided to go ahead and use regular silicon diodes. They each have a forward voltage drop of about 0.7V, so with four of them, we get V = (3*9V) – (4*0.7V) which gives us just over 24V. I have used charge pumps with multiple stages to get them over 50V from a 9V input supply, but do note that the current capacity drops in proportion to the voltage multiplication, so you can't run anything that requires high amounts of current.
Echoplexus Power Section
So there we have it! If you like what you see and want to try one of your own, you will find it at Github, here!