PT2399 reverb design problem.

[nguitar12]

So I am playing with PT2399 chip and take a reverb pedal as my next project. This is the post I current looking at:

http://www.diystompboxes.com/smfforum/index.php?topic=103259.0



As I understand the goal of achieving a reverb sound out of a PT2399 chip is to make the repeat of each chip to fill the gap of other. Above diagram seem make sense but one thing I don't understand is how to ensure the repeat of chip2 not to feed back into chip1 (as their output are connecting together). Or this is what we are looking for?

[anotherjim]

Not a problem. Real reverberant spaces contain sound that is echoes of echoes that have been back & forth between opposite and adjacent surfaces.

OK, consider what reverb is. It's echoes of course, and every space has echoes - and echoes of echoes. The size and shape of the space determines the echo return times and the number of reflection paths. A nice sounding echo chamber is big enough so the return times are long enough for our brains to register them as echoes distinct from the original sound. It also has dimensions scaled so the echo paths between any 2 opposite boundaries are not integer multiples of other paths between all the other boundaries -  then all the echoes have different delay times. Then you have a lush reverb.
Using at least 2 different delay times, and allowing the 2 delay signals to mix and be fed back for further repeats, you get a richer mix of echoes than from using only one delay. 3 Delays would be even better, and that's how many PT2399's are in the Accutronics/Belton "Brick" reverb module. 
   

[nguitar12]

Not a problem. Real reverberant spaces contain sound that is echoes of echoes that have been back & forth between opposite and adjacent surfaces.

OK, consider what reverb is. It's echoes of course, and every space has echoes - and echoes of echoes. The size and shape of the space determines the echo return times and the number of reflection paths. A nice sounding echo chamber is big enough so the return times are long enough for our brains to register them as echoes distinct from the original sound. It also has dimensions scaled so the echo paths between any 2 opposite boundaries are not integer multiples of other paths between all the other boundaries -  then all the echoes have different delay times. Then you have a lush reverb.
Using at least 2 different delay times, and allowing the 2 delay signals to mix and be fed back for further repeats, you get a richer mix of echoes than from using only one delay. 3 Delays would be even better, and that's how many PT2399's are in the Accutronics/Belton "Brick" reverb module. 
   

So the assumption that repeat of chip2 will also feedback  into chip1 is correct? And this is the what we are looking for?

[PRR]

You would want a "mixer" to get both a combined signal and still have separate signals.

Mixer can be as simple as two resistors.

The sketch-up above actually does mix through the two R10 1K resistors; however 1K may be low and it does not have individual un-mixed 1 and 2 outputs. Since you are driving an >10K combined load, add a couple 2K or 5K mix resistors after R10, tap your loop-back signals before that.

In general you DO want everything to bleed-back every which way. In the concert hall the floor ceiling and wall reflections bleed-back to ceiling floor and other wall, a dozen times before they fade away.

However if you are faking a room with several delays and they are all the same length, such bleed-around probably does not add anything to the reverb.

Leave the key points exposed for jumpering and mixing listening-tests.

[nguitar12]

You would want a "mixer" to get both a combined signal and still have separate signals.

Mixer can be as simple as two resistors.

The sketch-up above actually does mix through the two R10 1K resistors; however 1K may be low and it does not have individual un-mixed 1 and 2 outputs. Since you are driving an >10K combined load, add a couple 2K or 5K mix resistors after R10, tap your loop-back signals before that.

In general you DO want everything to bleed-back every which way. In the concert hall the floor ceiling and wall reflections bleed-back to ceiling floor and other wall, a dozen times before they fade away.

However if you are faking a room with several delays and they are all the same length, such bleed-around probably does not add anything to the reverb.

Leave the key points exposed for jumpering and mixing listening-tests.

Thanks. I will breadboard the circuit and see how it perform actually.