Reverb - look What’s in this mini pedal...

[Rob Strand]

FWIW,   

http://www.valvewizard.co.uk/equinox.html

has link to Belton patent.

[knutolai]

> PT2399 is just as analog as a BBD

Disagree.

Sorry Merlin, I'm with Paul on this one. If it has an ADC, RAM, and a DAC I have trouble calling it analog.

Another disagree from me, I'm afraid. A BBD is not terribly analog, and a PT2399 is even less so. To describe the pedal as "analog" is just too much of a stretch.

Are 3 x PT2399 cheaper than 1 x FV-1? It seems like an odd design choice.

At those volumes (Malekko) I imagine it will make more sense going with the PT2399 considering the volume discounts.

[11-90-an]

FWIW,   

http://www.valvewizard.co.uk/equinox.html

has link to Belton patent.

No values, though...

> PT2399 is just as analog as a BBD

Disagree.

Sorry Merlin, I'm with Paul on this one. If it has an ADC, RAM, and a DAC I have trouble calling it analog.

Another disagree from me, I'm afraid. A BBD is not terribly analog, and a PT2399 is even less so. To describe the pedal as "analog" is just too much of a stretch.

Are 3 x PT2399 cheaper than 1 x FV-1? It seems like an odd design choice.

At those volumes (Malekko) I imagine it will make more sense going with the PT2399 considering the volume discounts.

I bought 10 pt2399s for about $2... It didn't arrive yet, but i will see if fake or not.. any pointers to see if legitimate?

[niektb]

[...]

Are 3 x PT2399 cheaper than 1 x FV-1? It seems like an odd design choice.

Well, it could be my issue but I have yet to find FV-1 below $10... whereas tayda has pt2399 DIP for $0.79/piece and SMD for $0.29/piece

[anotherjim]

Language -  "...just as analogue as...."
A Tiger is ...just as fishlike as... an Elephant.

[Rob Strand]

No values, though...

True.   You can see the structure if you wanted to play.
You can spend a lot of time tweaking this stuff.

Merlin said he spent ages tweaking the his two PT2399 version and I believe it.    It could sound better than a more complicated design that hasn't been tuned-up.

The filtering can affect things too.  Even on a real spring reverb.   Current drive vs voltage drive, and how much bass and treble you let through.    At least the "reverb" part of the spring reverb is pretty much there.    The simplest case of a flanger + feedback is pretty dismal regarless of the delay setting  .

[ElectricDruid]

The PT2399 does have some characteristics similar to analog that straight digital systems don't have:
- the sampling frequency changes according to the modulation clock
- the effective delay when a sample exits corresponds to the average of the clock over the whole delay
- since the sample frequency is changing the sinc function hf roll-off is modulating with the clock

The characteristics follow from variable clock frequency.   For a fixed delay like a reverb the only thing you will notice is the sinc function hf roll-off.

Digital systems have a fixed sample clocks, although there's a number of schemes on how to do the sub-sample delay times for modulated delays like chorus and flangers.   

In the days before fixed sample rate digital systems was common there were plenty of variable sample rate digital systems: LinnDrum,  most of the early samplers (Emulators and such), PPG Wave, etc etc. They were all variable sample rate systems, just like the PT2399. Doesn't make them less digital, just different. It does make them much more similar to variable sample-rate analog systems like the BBD.

> A BBD is not terribly analog

For amplitude, BBD is analog as hell. Hiss, warp, clip.

In time it is discrete but well below Nyquist that's no big deal.

When Nyquist is 10KHz in *the datasheet for the device in question* that's a big deal. Otherwise, yes, I agree entirely - it's all true - BBDs are analog in as much as the amplitudes they handle are analog. But as we know, that's only half the story at best. It's still a sampling system. Hence "not terribly analog".

There are various grades here:

1) fixed sample rate, discrete time, discrete value - what everyone nowadays understands as "digital"
2) variable sample rate, discrete time, discrete value - early digital, with crazy pitch modulation, like the PT2399
3) variable sample rate, discrete time, analog value - BBDs. Can't think of any other example of this.
4) continuous time, analog value - what everyone understands as "analog"

Those middle two categories are the interesting bit where analog and digital cross over, but the PT2399 is closer to Rob's "straight digital" than a BBD, and neither of them are "straight analog".

Tom

[moid]

Hello everyone

I'm not sure if it will help anyone, but a few months ago I bought a Caline Reflector Reverb (CP44) on ebay because it was ridiculously cheap (£12 secondhand) and I'm a sucker for a cheap stomp box and it's quite fun - it's not a real reverb (didn't sound like it, but sounded weird and delayed and interesting when maxed out). Anyway I opened it up wondering what sort of DSP was inside and hey presto; it's got 3x PT2399 in it! See photo below, some of you might be able to work out the resistor values from it in case that helps (I've left the image hi res, sorry phone users). If anyone needs close ups of specific resistors let me know and I'll try to get closer with a different camera.

[anotherjim]

Thanks Moid.
Pin6's...
1K8
5k1
3k2

[niektb]

what makes one 3x-pt2399-based reverb sound realistic and the other fake? Is it the configuration of delay times?

[EBK]

what makes one 3x-pt2399-based reverb sound realistic and the other fake? Is it the configuration of delay times?

Could be modulation too.  The Belton reverb bricks, for example, have a built-in modulation circuit that many people dislike (a carefully drilled hole can "fix" it).  Ever see a real room with walls that modulate (without mind-altering drugs)?  That's probably why it sounds fake.

Probably many other possibilities.

[moid]

Thanks Moid.
Pin6's...
1K8
5k1
3k2

No problem, it's not often I can help a thread out; I'm usually the cause of the problem I should point out that I have no idea if my Caline Reverb is the same circuit as the Mosky... but they're both cheap pedals...so one probably copied the other (or both stole from a third party!). I did get the pedal out today for a play and it does sound pretty good - you can hear the slapback delay sound a little, the first PT2399 is definitely a short delay, then it gets progressively longer and more reverby - I think 2.5 - 3 seconds reverb trail? Which is pretty similar to a Belton Brick I think. If you play a note and then tap the string quickly afterwards to deaden it, you get a sort of 'spring' sound... I don't think anyone who has played a spring reverb would mistake it for one though. It's a bit of a one trick pony - it only really sounds good with both controls maxed out, which I suppose is also an issue withe the Belton Brick... I was thinking of modding mine, but I think I'll leave it as it is; SMT parts look frighteningly small for me to screw up!

[11-90-an]

(I've left the image hi res, sorry phone users).

You aren't kidding... even using a 21.5 inch screen doesn't help

[Rob Strand]

I traced most of it.  I didn't really intend to trace the whole thing or produce a schematic.
It pretty much follows the Belton schematic.

https://patentimages.storage.googleapis.com/US8204240B2/US08204240-20120619-D00007.png

Going from top to bottom on the Belton schematic,

Belton      Caline Reflector Reverb (CP44)

Top           U4              Rpin6=5k1               Delay 88ms
Middle      U3              Rpin6=3k2               Delay 66ms
Bottom      U5              Rpin6=1k8               Delay 50ms


The Belton schematic has some duplicated designators so I'll break it down.
My designator orderings for each section are from top to bottom on the Belton schem.

Input resistors R1 on Belton                                        10k  R17, R16
Filter resistors R2 on Belton to pin 16                      6k8   R25, R29, R32
Filter Feedback resistors R1 on Belton  to pin 15   10k   R24, R28, R31
Filter cap to ground, C1 on Belton:                             C13, C8, C15
Filter feedback cap  C2 on Belton pin 15 and pin 16:  C12, C14, C16
Feedback from pin 12  R1/A Belton:                             12k   R23, R18, R33
GF (Global Feedback?):                                                  12k   R26, R27, R30

Top of right pin 13 and pin 14 of Belton  has a mixer:
R3 on Belton:                                                                  10k R22, R19, R21
Feedback resistor R3/N                                                 6k8   R20

Bottom-right pin 13 and pin 14 Belton has a filter
input  R34  5k1
feedback  R26 5k1
cap to ground C18
resistor between caps R35 10k
feedback cap C19

Middle-right pin 13 and pin 14
feedback R30 3M3
feedback C21
modulation to pin 6 R40 220k

The LF347 is the modulation oscillator

The TL072 does the input buffer and output.

The 4558 feeds delay, handles the delay feedback and delay output mixing.

That's the basic idea I probably could fill in a few more details around the opamps.
How the pots connect isn't 100% clear.  There's a track which goes under U5 near the dwell pot.

FWIW, I traced it independently of the Belton schematic, then later I matched it up because I didn't want to draw a schem.

[Rob Strand]

Can't say these are correct for the Caline.   

In order to match the filter cut-offs mentioned in the Belton patent these should be a reasonable starting point.
In one place it mentions Bessel and another Butterworth.

You can see the difference boils down to playing with the C1 values.

Butterworth (target: -3db @ f ~ 13250Hz)      
IC =                  U3 & U4    U5       U5 out
C1 (gnd) [nF]     3.9          3.3          3.9
C2 (fb) [pF]       470          560        680

Bessel (target: -3db @ f ~ 13600Hz)      
IC =                 U3 & U4    U5        U5 out
C1 (gnd) [nF]     2.7          2.2          2.7
C2 (fb) [pF]       470          560        680

[merlinb]

Both are discrete time, only one is discrete signal level.

The PT2399 converts into a bitstream where the analog value is ecoded into a mark-space ratio rather than as a discrete digital number. Since mark-space is infinitely variable, it can sample any analog value, which makes it just as analog as a BBD, IMHO.

[moid]

Hi Rob did you want me to check any of the circuit routing on the PCB for you?

[Rob Strand]

Hi Rob did you want me to check any of the circuit routing on the PCB for you?

It's up to you really.  If it helps you think about doing some mods we can keep going with it.

The main conclusion is it follows the Belton design very closely, including the delay times.  I'm quite confident about that part of it.  Obviously we don't know the cap values.

The Belton schem pretty much covers the 3xPT2399's.    So we know what the thing does.   The only parts left to draw-up are the opamps and the switching.    Drawing these parts are only going to help someone modding the Caline.    I can see what these parts are doing but I can't see enough to draw-up the circuit accurately.  There's a few places where I loose the trail.   It might be tedious to get that 100% correct via posts.  I don't know if you want fill-in the missing parts of the schematic?

The computer I'm using at the moment only has LTSpice on it.  The editor is painfully slow to use.   I can't see myself drawing up the whole circuit with it but I could draw-up the opamp part.

Thinking about the filtering, the 13kHz or so filters seem pretty high for a spring reverb.  They are mentioned in the Belton patent.   Maybe the output filter of the last PT2399 stage has a lower cut-off;  the filter with 5k1 resistors.