PT2399 delay module

this section is currently organically growing


all you need to know about vactrols (this section is in progress, more content will be added)


under preparation

Powered by:

PT2399 KaraOK! module - circuit bending

When I saw an article about transforming a cheap Karaoke board into a eurorack delay module, I was excited and ordered a few boards at once. So I started building, and yeah, it all worked: I could adjust feedback (rather a dry/wet control), delay time, input level … and from a cheap board you can not expect WOW results anyway! 

So I started to use a karaoke delay module in my set-up, but soon noticed that the input level control was not useful between 1.5 and 10, due to clipping distortion and all low end disappearing. No problem, I thought, a simple voltage divider at the input will do the trick. 

The next thing was the feedback control: it went from 0 to 10, but my preferred setting was definitely over 11: no way to achieve an endless delay! 

So I thought: no problem, I’ll do that with a mixer module. So I connected a decaying noise source to a mixer input, and connented the mixer output to the delay module input. I also splitted the delay output and fed it back to the mixer as well, ready for heavy feedback... but no, the output and the input were in opposite phase, so the signals were just cancelling out each other. Of course this made sense: you don’t want endless feedback in a karaoke bar when a singer puts himself in front of a loudspeaker: having input and output in opposite phase is the right way to go!

For my setup, it means I had to change the phase of the delay output by sending it through an attenuverter before feeding it back to the input mixer.

And we’re not at the end yet: the delay time range was just covering the sweet range of karaoke.

As you can guess, my satisfaction level was completely in line with the low price of these karaoke boards!

The problem was: I had ordered a few boards at once!



Looking at the karaoke board you can spot a PT2399 (a lovely characterful lo-fi delay chip, used in several circuits) and there are also 2 NE5532 opamps: good stuff! Because I could not find the schematics of the karaoke board, I decided to sacrifice a board for reverse engineering on a Sunday morning. In the afternoon, the sun got through the clouds, also on my drawings, so I went on with the karaoke board till the sun was down. 

Here’s a short description of the board connections: there are 2 microphone inputs, a stereo input and a stereo output.

Interesting, because the PT2399 delay chip is mono! First of all, the microphone signals are connected together via a simple resistor, so it’s basically a single input. The stereo input connection is where you add the stereo karaoke music to the board: we don’t use this. 

The microphone signal is amplified (1 NE5532 is taking care of that job) and input to the PT2399 delay circuit. The mono output from the delay circuit is mixed with the stereo input (which we don’t use) by the 2nd NE5532 opamp, and fed to the stereo output (in our case left and right are the same, we just use 1 output).

Time for some mods to make a great module from this board!

Before we implement the mods, we need to make the board eurorack compatible.


The first thing to fix is the input level: this board is designed to connect microphones, and by just swapping a microphone signal by a modular level signal, it’s no surprise that the input level is an issue.

So we need to make the microphone input compatible with eurorack levels. There are 2 ways to do this: adding a voltage divider at the input to lower the input level, or bypass the mic preamp circuit. Bypassing the first amplification stage (34x amplification of the low voltage mic input is the best option, also from a noise perspective.



A second thing is: more delay time range! 

The delay time of the PT2399 is determined by pin 6: a small resistor gives very short delay times (up to chorus effects), higher resistor values make delay times longer, way beyond the official specs!! On the board, there’s a 10k resistor + a 50k pot to adjust the delay time: from 10k to 60k. If you want to get the shortest possible delay time: simply put the pot on zero, and change the 10k resistor by a smaller value. In practice, this should not be lower than 2k2 at startup of the module because otherwise the PT2399 may latch. Once the board is on, making this resistor 0 is perfectly OK). So, to expand the delay in both directions, the 10k resistor may be changed by a 2k2 resistor (practically, it’s not needed to desolder the installer 10k resistor, just mount a 2k7 resistor in parallel!), and the potmeter may be changed by a higher value. Interestingly, the time pot (like all pots) is a stereo pot, where only 1 side is used. With a simple trick, we can keep the installed 50k pot, and change it into a 100k pot without desoldering or damaging it!) Now we have created a tamed beast till 12 o’clock, and a wild beast for the second half of the range!



The third thing is: creating endless feedback.

Although it is possible to achieve endless feedback using an external mixer with attenuverter function (remember in and output are intentionally out of phase for karaoke use), doing this with a simple mod is so much nicer!

Making the input and output in phase is not a simple option without doing a lot of mods around the opamps, so we have to find something else. And luckily there an other solution: a 10k resistor that gives us endless feedback when we reduce it!! The board has a 10k resistor installed that is “safe”: no endless feedback. Depending on the individual PT2399 characteristics (it’s analog anyway) 10k is too safe or just on the edge of endless feedback (I have 1 board going into endless feedback with the 10k resistor, most boards don’t). The challenge is that the resistor and the PT2399 need to match, and the sweet spot is in a narrow area. Again there’s a perfect solution and easy solution: we’re going to use the 4rd pot (meant to control the stereo input which we don’t use!) and put this 50k pot in parallel with the 10k resistor. To have perfect control of the sweet spot for all PT2399 chips, we first extend the 10k feedback resistor by 1.5k: 11.5k. Now if we put a 50k pot in parallel we can still achieve 10k (50k//11.5k => 10k), and the sweet spot between 7k-10k is now spread over half the range of the pot!



Next we will add a vactrol to modulate the delay time. How to make a vactrol is out of the scope of this description. Look at the dedicated "Vactrols" section. 
After installing, it's time to check out your new eurorack delay module, and let yourself carry away away away abay abay obey obey ...

output lvl


I’ve experimented a lot with the PT2399, abusing connections in extreme ways, and NEVER had holy smoke. I use the word “experiment” for a reason: aside from the theory and the specs, it feels nice to play around and see what happens. There’ a way to avoid holy smoke: be aware that the PT2399 is working on +5V. Applying voltages to an IC that are outside the supply voltage is the best way to damage integrated circuits, so just don’t do that. In my experiments, I tried to short circuit any pin with any pin (except voltage supply pins), just to see what is does to the sound. Then, I used resistors, capacitors and diodes instead of short circuits, and made a list of interesting mods. To avoid problems using external control signals to this 5V operated IC, DIY vactrols are super handy: it enables you to adjust the value of a resistor in the delay circuit without any electric connection of the controlling circuit (the connection is via LED light shing on a light depending resistor!
One of the results is PSYCLICK NOISE. I decided to limit the number of mods, and kept 4 toggle switches, 8 momentary switches and 4 vactrols for the mods. Oh yes: 2 rotary switches as well, to switch between different capacitor values which affect filter settings. The filters are interesting with extended delay settings: the PT2399 is pushed out of specs (longer delay time with a fixed memory means a lower sample rate. Those who are familiar with the Nyquist theorem (sample rate should at least be double of the highest audio frequency) will immediately understand that lowering the sample rate opens the door to glitches and artifacts: very interesting, and even nicer if you have some control over the filters. 2 times 7 capacitors may seem out of proportion for lo-fi applications, but bear in mind that a rotary 10 position switch costs the same as a simple toggle switch, and with a little twist you can use it to switch between different capacitors (when used in filters: always do this in parallel with a fixed capacitor, since while switching of the rotary switch there’s no capacitor connected.

The small pcb with the 8 pushbuttons is held in place by the 4 vactrol pcb. I needed to close another 2mm of space, so I wounded a light yellow cable around the vactrol pcb for this purpose. Making a 3D puzzle of individual boards and parts looks very easy once you see the final assembly, although it happened to me a few times that I got stuck, and things just did not fit together. To avoid that, I oblige myself to draw, check and measure the last details before starting the actual job. When you stop at 95%, thinking that the last 5% will turn out by itself when you get there: that’s when things go wrong, and it may be too late to change some parts: make it again, Sam! Putting 1 hour extra in planning and preparation saves several hours in the end.

I will soon make a video about PSYCLICK NOISE, so that you can hear what it does.



When you try to put your module in an external feedback loop, you will notice that the delay does not become endless as you would expect. The reason is simple: the input and output are in opposite phase, they cancel each other out! For karaoke systems this is exactly what you want: when people sing in front of loudspeakers, reversing the phase is the best measure to avoid feedback.  

When you want to use your delay module in Alessandro Cortini style, you want to send the output to a distortion, and feedback the distorted signal back to the input. For this application, having input and output in phase is exactly what you want, unless you have too many attenuverters around...

There's a simple mod for this, and this mod also increases the output level at the same time.

Look to the picture below how to correct the phase and increase the level:

- unsolder 1 leg of the 1k resistor and 1 leg of the 22k resistor as shown.

- solder the unsoldered legs together.

- make sure that the wire going to your output jack is connected to the left output pin, because the right output pin does not work anymore! 

What you have actually done is the following: the 8-pin dual opamp mixes the delay signal with the left and right karaoke music inputs (which we don't use anyway) by using inverting amplifiers. The resistor hack puts these 2 inverting amplifiers in cascade, so that the phase is corrected as well as the output level. As a result 1 output is no longer in use.

All you need to know about


What is a VACTROL?

A VACTROL is the combination of an LED (light emitting diode) and an LDR (light dependent resistor) in a single package.

When the LED shines light on the LDR, the resistance of the LDR goes down. That’s it!

If you expose an LDR to daylight and measure the resistance, you will find a value in the range of 1kOhm (the actual value depends on the type of LDR, and may be higher or lower). If you put the same LDR in a dark environment, the resistance goes up to 1MOhm or much higher. 

By putting an LDR and an LED in a package that does not let in daylight, we can control the resistance of the LDR by the light of the LED.

When the LED is off, the resistance of the LDR is very high, when the LED lights up, the resistance of the LDR goes low.


The great thing about the LED shining light on the LDR is that there is no electric connection between both components: the only connection between both components is light!

This opens a lot of possibilities, like modding existing circuits without changing the original circuit.

This way, you can add a potmeter or voltage control input to existing circuits. 

Vactrols are great to switch audio on and off without annoying clicks. This is because it takes some time for a vactrol to react to a transient from on to off or from off to on: switching off the LED does not mean an instant high resistance of the LDR, while switching on the LED does not result in low resistance of the LDR instantly. 

Aside from switching, controlling a resistor by a control voltage is very useful in VCO's, VCA's, LFO's, tremolo, vibrato, etc

Vactrols are not linear...

Aside the fact that a vactrol does not react instantly on a change of light intensity, there is no linear relationship between the LED light output and the LDR resistance. 

This means that when you double the current through the LED, the LDR resistance does not decrease by the same factor. In addition, an LDR has a memory effect: the resistance also depends on the light/dark situation just before the current situation. And that’s not the end of the story: every vactrol will behave slightly different. 

...but they sound very musical

However, the few milliseconds to switch them on or off tend to be very musical. Just as the human hearing system is not linear (to perceive double loudness of a sound, we need to make it 10 times ous loud from an electric point of view!) some of the vactrol imperfections match well with human hearing.


The Art  is of course in using the disadvantages as features! A good example is the so-called LPG or Low Pass Gate, very popular in west coast synthesis.

An LPG is basically a vactrol switch, and its sound is determined by the transient between on and off, the low frequencies are passed earlier colpared to the high frequencies (fastest transients) get through later in time will be less compared to the lower frequencies: no unpleasant switch clicks! After this initial transition, high and low frequencies get through in the same way, unless a lowpass RC filter is added. 


The first commercial use of vactrols was by Teletronix, who used a vactrol circuit in their audiocompressor LA-2. When I was mixing vocals in music productions, I was searching for the best compressor during several months, and finally found a software version of the LA-2 (€30k for a vintage hardware version was out of reach): it is incredibly soft and musically, even when you push it hard. 

Noise and distortion of an LDR 

The noise of an LDR is greatly dependent on thermic noise (like normal resistors), practically above 10kHz in case of 50 volts or more: negligible in most audio applications. For signals below 300 mV and a resistance below 100 kΩ, distortion levels are extremely low. The dark resistance of LDRs degrades gradually over the years.

Are vactrols expensive?

If you want a Xvive VTLC1 from Thonk, you will pay at least €5. If you buy a VTLC1 directly from China, you will pay about 1€.

I’ve tested both: Xvive vactrols are sealed from outside light because the components are poured into a mold, whereas a €1 vactrol is put in a plastic housing that has light leaks via the holes for the leads: one day I thought that a lot of 10 vactrols was out of spec because the off resistance was below 1MOhm, but when I covered a vactrol with my hand, the resistance suddenly increased by a factor 10!

Of course, in the darkness of your modular case this becomes less of a problem.

In addition, 1 out of 10 cheap vactrols is different in terms of specs.

So far the conclusion is:

if you only need a few vactrols and you want to be sure that the circuit begaves like intended: don’t hesitate to spend €5 per piece.

If you plan to use several vactrols: buy a bulk quantity and spend a few minutes to check the values in both off conditions (remember the impact of ambient light!) and on conditions (5V with 1k in series with the LED is perfect).

Because the forward voltage of a red LED is about 1.8V, the remaining voltage over the 1k resistor is 5V - 1.8V = 3.2V, and the current through the LED is the same as the current through the resistor because they are in series: 3.2V / 1kOhm = 3.2 mA): mark the vactrols that show a different behaviour to be used for special projects.

So this is it, €1 for a vactrol?

No! If you make your own vactrols, they will cost you just around €0.1!!!

€5 - €1 - €0,1 ... and all good!

When you go for DIY, you need to experiment with several types of LDR and LED.

The wrong type of LDR, LED or combination can make your vactrol useless.

I’ve made several dozens of vactrols, and for most of my projects, an 5539 type LDR combined with a “normal” 3mm red LED (8.7mcd/10mA) gives the best result.

If I would use a 20mcd/10mA LED, I would need 20/8.7 or 2.3 times less current to have the same brightness. But maximum brightness is not our goal here, it's about finding your own unique vactrol behaviour, to make your own unique sound!

Once you find a good receipe, you can make 25 vactrols for the price of a good coffee!