PT2399 Reverb Pedal

[JKowalski]

I'm working on a digital reverb using PT2399 chips. I suppose it's like a discrete version of the Belton brick - I have no idea how Belton does theirs, and whether it even uses PT2399's for sure (though I've heard it does).

I had a bunch of the chips laying around, and I felt experimental today.

Anyways, my design is rather simple - it only uses two PT2399s and the decay time and all sorts of interesting properties can be adjusted as well. I just breadboarded a preliminary stripped down version of the circuit today as a proof of concept, and from what I've managed to get out of the prototype I think this is definitely worth developing into a full fledged pedal and offering it up to you guys!

The rest of today I have got to study for a Calc II test so I can't mess with it anymore but tomorrow and the rest of the week I will be tweaking and redesigning it. (Unless my high school teachers unload a ton of stuff on me like they always do at the end of semesters  ) If you guys have anything you want me to add or consider in the design, please!

Maybe expect some preliminary sound clips this week, and a schematic+PCB the next if all goes smoothly?

[aleister]

hello, actually i would love to see the preliminary schematic/layout that you are working with. these are neat ic's, i have never used 2 in one circuit yet. thanks in advance.

[aziltz]

hello, actually i would love to see the preliminary schematic/layout that you are working with. these are neat ic's, i have never used 2 in one circuit yet. thanks in advance.

sounds f'n awesome.  I want to take a stab at this when I get into PT's.

[Lurco]

check Neunaber: http://www.diystompboxes.com/smfforum/index.php?topic=76484.0

[Taylor]

The basic innerworkings of the Belton reverb can be seen in Neunaber's patent. The images provide a pretty good idea of what's happening :

http://aiw1.uspto.gov:80/.aiw?docid=20090003614&PageNum=7&IDKey=7584A5BC9FB0&HomeUrl=http://appft1.uspto.gov/netacgi/nph-Parser?Sect1=PTO2%2526Sect2=HITOFF%2526p=1%2526u=%25252Fnetahtml%25252FPTO%25252Fsearch-bool.html%2526r=1%2526f=G%2526l=50%2526co1=AND%2526d=PG01%2526s1=neunaber%2526OS=neunaber%2526RS=neunaber

http://aiw1.uspto.gov:80/.aiw?docid=20090003614&PageNum=8&IDKey=7584A5BC9FB0&HomeUrl=http://appft1.uspto.gov/netacgi/nph-Parser?Sect1=PTO2%2526Sect2=HITOFF%2526p=1%2526u=%25252Fnetahtml%25252FPTO%25252Fsearch-bool.html%2526r=1%2526f=G%2526l=50%2526co1=AND%2526d=PG01%2526s1=neunaber%2526OS=neunaber%2526RS=neunaber

It uses 3 PT2399s.

What sort of reverb sound are you aiming for here, out of curiosity?

[jacobyjd]

Check this out: http://electro-music.com/forum/topic-35276.html

[aziltz]

i think we should aim for something that DOESNT sound like the Belton.  It sounds pretty good, but it only does spring.  It would be really nice to have something that will do  other sounds that are more ambience than spring/echo.

[JKowalski]

i think we should aim for something that DOESNT sound like the Belton.  It sounds pretty good, but it only does spring.  It would be really nice to have something that will do  other sounds that are more ambience than spring/echo.

That's kind of what I am going for - Having the reverb circuit discrete let's me adjust a whole bunch of parameters and get some interesting sounds. I can get really long extended ambient reverb, interesting reverb swelling effects, and god knows what else. The echo unit method is however more suited to the mechanical sounds of a spring reverb... I.E. it's hard to get perfectly smooth kinds of verb like you would see with DSP and stuff. I'll have to mess with it and see what I can do!

[Mark Hammer]

There are two fundamental aspects to what makes "reverberation".  One is the absence of anything that sounds like standing waves.  This means that iterations should possess as little mathematical relationship to each other as is possible to achieve, and is the very basis for the MN3011 having taps along its 3328 stages at mathematically unrelated points (i.e., no tap is a multiple of any other).  

The second requirement is that it manifest the sound absorption properties of a real space.  This means that iterations need not be filtered over time like wah, but that the general pattern of iterations include an overall loss of high frequency energy commensurate with the "distance" the sound has travelled.  So, longer delay iterations should lose more sonic energy and especially high-frequency energy with each repeat, and iterations intended to depict waves bouncing off immediate hard surfaces should retain high-frequency energy over more iterations.  In my own experiments with analog delays over the last 25 years, I found that the now-omnipresent hi-cut that so many install on their digital delays does a nice job of making the resulting sound feel more "ambient".  I should emphasize that the hi-cut can be installed in two loci: the overall wet-level control, and the feedback-level control.  It produces a more ambient-like sound when installed on the feedback-level control, for precisely the reasons outlined above: it mimics the natural treble-eating properties of real spaces.

One of the things I've never seen on any of the MN3011-based circuits that attempted to produce analog reverb was any differential filtering or feedback of the taps.  Ideally, shortests delays ought to have more top end and repeat more than longest delays.  I guess the way to implement this would be to have some simple RC lowpass filter between each tap and a common mixing point which would simultaneously take a predetermined amount of treble off (more for later taps), and also set the mix-level of that tap (more attenuation for later taps).  That way, one could use a global feedback control, and extend the decay by introducing a bit of feedback to the input.  Although, now that I think of it, perhaps one might want to have the short-repeats decay a little quicker than the ones bouncing off the other side of the parking garage walls in some instances.

[aziltz]

are MN3011s gone like the MN3001?

[Mark Hammer]

are MN3011s gone like the MN3001?

Because they tended to have a fairly narrowly-targeted market, there may be little pockets of them here and there.  About 4 years agio, the guy who makes Retro-Sonic pedals (we live in the same city) told me he had a line on a batch of them, and was looking for possible products he might develop with them (nothing materialized, but that was my fault).  So, I assume that they are sitting somewhere unused, and may even be more numerous in NOS form than SAD1024s at the moment, but not enough of them to make a sustainable product line viable, just limited runs.  And because they are not marketable the way that 4096-stage devices are, don't expect Coolaudio to start cranking them out for a buck apiece anytime soon.  Even when you can find them, don't expect them to run you much less than $20@

[JKowalski]

There are two fundamental aspects to what makes "reverberation".  One is the absence of anything that sounds like standing waves.  This means that iterations should possess as little mathematical relationship to each other as is possible to achieve, and is the very basis for the MN3011 having taps along its 3328 stages at mathematically unrelated points (i.e., no tap is a multiple of any other).  

The second requirement is that it manifest the sound absorption properties of a real space.  This means that iterations need not be filtered over time like wah, but that the general pattern of iterations include an overall loss of high frequency energy commensurate with the "distance" the sound has travelled.  So, longer delay iterations should lose more sonic energy and especially high-frequency energy with each repeat, and iterations intended to depict waves bouncing off immediate hard surfaces should retain high-frequency energy over more iterations.  In my own experiments with analog delays over the last 25 years, I found that the now-omnipresent hi-cut that so many install on their digital delays does a nice job of making the resulting sound feel more "ambient".  I should emphasize that the hi-cut can be installed in two loci: the overall wet-level control, and the feedback-level control.  It produces a more ambient-like sound when installed on the feedback-level control, for precisely the reasons outlined above: it mimics the natural treble-eating properties of real spaces.

One of the things I've never seen on any of the MN3011-based circuits that attempted to produce analog reverb was any differential filtering or feedback of the taps.  Ideally, shortests delays ought to have more top end and repeat more than longest delays.  I guess the way to implement this would be to have some simple RC lowpass filter between each tap and a common mixing point which would simultaneously take a predetermined amount of treble off (more for later taps), and also set the mix-level of that tap (more attenuation for later taps).  That way, one could use a global feedback control, and extend the decay by introducing a bit of feedback to the input.  Although, now that I think of it, perhaps one might want to have the short-repeats decay a little quicker than the ones bouncing off the other side of the parking garage walls in some instances.

Great information! I have to head off to class but I will look at it in more depth after.


If your curious as to the layout of the PT2399 chips in my design, its pretty simple. One chip is set for a pretty short delay time, the other is set to a slightly shorter one. The second one is in parallel with the feedback loop of the first one, and it's output also goes to the final mix as well as back into the longer delay chip. SO it goes like this: Long repeat, short repeat, long repeat again, long repeated short repeat, long repeated short repeater short repeated, long repeat, ....etc. Basically the repeats keep multiplying on one another and overlapping, kind of exponentially. Also, the difference in the times of the two chips make it so each repeat is in a different position relative to one another, they constantly shift forwards and don't typically coincide. It makes some interesting sounds, for sure. The format of my layout is very open to specific kind of filtering like you suggest, so that's something I'd have to incorporate next.

I read that the Belton uses three chips in a parallel format... I can't see the diagrams included in the patent for some reason (big red x no matter what) but I read the text version. I think I make this sound good with two chips only, but I will have to see how smooth a reverb I can end up with experimentation....

[frequencycentral]

Chris, there's a thread over at electro-music.com that may be of interest to you, a guy started a project using 6 PT2399 for reverb: http://electro-music.com/forum/viewtopic.php?highlight=pt2399+reverb&t=35276

[JKowalski]

Chris, there's a thread over at electro-music.com that may be of interest to you, a guy started a project using 6 PT2399 for reverb: http://electro-music.com/forum/viewtopic.php?highlight=pt2399+reverb&t=35276

Yeah, I've seen that one - I didn't really like how it sounds, and the amount of chips is a little too high. I don't think his project took enough advantage of feedback systems. There is the figures from the belton on that page I didnt notice before and couldnt see on the patent, so thats helpful!

Looks like the belton uses pretty much the same format after all, except with three delay times instead of two as mine does. Hmm.

[Taylor]

Though of course I support the DIY spirit of designing things ourselves, I do want to give a quick nod to the FV-1. It is an extremely powerful chip and people have already written some amazing reverbs with it. IMO it sounds much better than the Belton, and anything else that is based on PT2399s can. You can make an amazing, rack-quality reverb with this chip with no programming on your part and very low parts count.

PT2399s will never allow for a really long reverb time without hearing the individual repeats, and the final density can't be as big as you really need in a good reverb. I actually worked on a PT2399 verb with 5 delay chips for a while, and ultimately realized that there's a better way.

Now, that's just my conclusion, which could very well be different from anyone else's, so I definitely don't want to discourage this for those who are liking it.

[Mark Hammer]

I think it is an entirely rational conclusion to reach.

As elegant as something like the FV-1, and any existing or slated-for-release chips are, I think the hardware approach to generating reverb is something that aligns a little more closely with most people's thinking about effects generation.  That doesn't mean it is "best".  Rather it simply provides a usable and approachable platform for them to explore their own ideas.  Put another way, I'm confident that every language of the world is capable of exquisite poetry of the sort that could bring tears to your eyes.  But if I know how to express my ideas and sentiments in English and not nearly as well in other languages, English is what I will use.  So, the digital option afforded by devices like the FV-1 may well permit "digital poets" to do marvelous things, but the rest of us plan out our acts of tortured beauty with through-hole components.

That being said, few FX-on-a-chip devices are open-ended enough to permit some of the truly bizarre and targeted-use FX that some folks feel the need for.  That is, they do what they are intended to do extremely well, but there are some things they just won't do.

[JKowalski]

I'm starting to adjust my opinion of this project quite a bit. (In the negative sense) I started this hoping to get a relatively low part count, decent sounding, adjustable reverb, and so far it's proven to be more of a interesting wierd effect board rather than a nice reverb. I've been making all kind of adjustments with the sound and I've been intrigued many times (Weird note swell effects, resonating plate kind of sounds, odd bouncing reflection messes)  but never truly enthusiastic about any of the "normal" reverb sounds I have managed to dial in. I think Belton had the right idea when they stuck three chips on there, I think three is the absolute minimum where you can a decent sounding reverb-reverb.

Perhaps this will just be another one of those "it's not worth it" projects, as Taylor concluded. It's sure starting to look like it to me.

As for the FV-1, I am sure it has great capabilities but that's not my kind of area. I'm not actually looking to build a reverb for myself, unless it would turn out to have special interesting qualities that set it apart... Rather, I am building this as an exercise in design, an opportunity to mess with interesting results, and a chance to give people another option. The FV-1 is a predesigned, program only chip that really offers no design challenges in incorporation. The PT2399, however, being a building block for more complex designs, interests me much more. Mark Hammer said it well: "(The hardware approach) simply provides a usable and approachable platform for them to explore their own ideas. "
I'm not much of a builder, (I rarely assemble something unless I absolutely want to use it) but more of an experimenter and learner, many of my projects are for personal education only and never get past the breadboard stage.

[Mark Hammer]

Let's put it this way.

Two points will always get you a straight line.  You need at least 3 points to have something that isn't a straght line.  When it comes to reverb, a straight line = standing waves and irritating resonances.  It is precisely why setting an analog delay for slapback, and turning up the feedback doesn't get you a reverb sound.

So yeah, I think both you and Belton are spot on with respect to the need for at least three sources of delay to produce a reasonable solid-state reverberation.

A couple years back, I worked out a conceptual design for an ambience generator using 3 cascaded BBDs set for different delay times.  The junction between each BBD would involve a mixer stage to incorporate both feedback from other chips and feed-forward from preceding ones.  With each chip set for a different delay, the assorted feedback and feedforward signals would amount to a broad range of total delay times.  Never built it, but in theory it seemed plausible.

I suppose the alternate way to do so would be to use an MN3011, and assign some taps to the output and route others to the feedback path.  If the path you travekl back is a different number of stages than you travel forward, then you end up with many different reflection times.  Whether it nets you something that would have the appropriate "sound absorption" properties is another matter.

[aziltz]

jkowalski,  did you put together any kind of schematic or block diagram of what you were working on?  One thing that would definitely be tricky here is how much control is enough, and when might it become too much that its unusable?

[JKowalski]

jkowalski,  did you put together any kind of schematic or block diagram of what you were working on?  One thing that would definitely be tricky here is how much control is enough, and when might it become too much that its unusable?

No, sorry. It's really pretty simple. The Belton reverb turned out to be basically the same thing, cept with three PT2399's. Multiple echos, with each echo output going into all the other echo inputs so you get a kind of wave upon wave multiplication of repeats.

The amount of control wasn't the problem, it was simply the inability to actually get a clean, decent reverb with only two delay chips. There simply wasn't enough unmathematical interactions to make it sound natural enough to be worth building. Perhaps I didn't give it enough time, and didn't try all the possibilities to fix the issue, but I had a feeling that it ultimately wasn't going to go anywhere.