Author Topic: DSP based delay question....  (Read 6259 times)

G. Hoffman

DSP based delay question....
« on: February 15, 2010, 05:31:40 AM »
So, I'm a BIG fan of delays, but I like them long, and usually clean.  

I'm also in no position what so ever to pursue this as a project, but in the interest of personal growth, what type of architecture are they using for commercial delay pedals such as the the Boss DD-20, the Empress DD-20, etc.  It can't just be a chip like a 2399, is it?  Or is it something more like a PC - a processor, with separate I/O controllers, and separate RAM to store the information?  Or something else that I'm just not familiar with right now?

I'm sure it must be simply hardware wise, right?  Just ASMOP? ::)


Gabriel

cloudscapes

Re: DSP based delay question....
« Reply #1 on: February 15, 2010, 07:20:17 AM »
DSP delays I've looked (dd20, ehx smm/h) at are basically a "computer chip" and a few megs of ram. most of the other parts in those delays are to support the chip (power supply, filtering) and to buffer/convert the signal. the rest is done in programming.
~~~~~~~~~~~~~~~~~~~~~~
{DIY blog}
{www.dronecloud.org}

peterv999

Re: DSP based delay question....
« Reply #2 on: February 16, 2010, 07:21:09 AM »
So, I'm a BIG fan of delays, but I like them long, and usually clean.  

I'm also in no position what so ever to pursue this as a project, but in the interest of personal growth, what type of architecture are they using for commercial delay pedals such as the the Boss DD-20, the Empress DD-20, etc.  It can't just be a chip like a 2399, is it?  Or is it something more like a PC - a processor, with separate I/O controllers, and separate RAM to store the information?  Or something else that I'm just not familiar with right now?

I'm sure it must be simply hardware wise, right?  Just ASMOP? ::)


Gabriel

Gabriel,
It's all depending what your idea is about delays. In the old times Delays could be single delay's and multi delays, based on tape disc systems, of which each delay could be adjusted to a required amplitude. These days the multi-delays are all single delay lines with multi-tap repeats mostly based on BPM rate. I've found a DSP that would cover all requirements. The FV-1 DSP chip and the piggyback board from OCT are perfect candidates to get you fully on-board so to speak. I guess your requested explanation of what you're looking for could be extending this functional range!

-P


scratch

Re: DSP based delay question....
« Reply #3 on: February 19, 2010, 09:30:51 AM »
technically you can do delays 'simply' with analog to digital converter(ADC), some memory (preferably SRAM) and a digital to analog converter(DAC), a DSP is not required for that. However what is simple  in broad terms is a different thing in implementation in terms of of bits of resolution, sampling rate, size of memory, etc. never mind attempting modulation. When you want to start modulating (Chorus, Flanger,etc), doing reverb and combinations thereof then a DSP is the way to go, but then you are in a whole other world of programming ...

Chips like the PT2399 integrate the ADC, RAM and DAC into one chip making it  very simple to use. The FV-1 Digital Reverb 'engine' is in fact a DSP. It has integrated ADC, RAM, DAC and a processor core that can run instructions, you can point it to an external EEPROM (Electrically Erasable Programmable Read Only Memory) where you can store your own programs (algorithms) to implement your own 'effects' ...
Denis,
Nothing witty yet ...

audioartillery

Re: DSP based delay question....
« Reply #4 on: February 19, 2010, 11:32:19 PM »
I have a Line6 dev kit so I have some visibility into the hardware.  DSP core, ADC on one side, DAC on the other.  Pretty much what you'd expect.  And an external memory module (512kB... ~5 seconds-ish).  Implementing delay on this kind of setup is trivial, you just treat the memory as a ring buffer and output samples from whatever time in the past you want.  Very simple code.

New-fangled delay pedals seem to be all about emulating this or that vintage delay sound.  So if you want to do that kind of thing there's a lot more DSP work to do.  But I don't know if I really can tell the difference.

peterv999

Re: DSP based delay question....
« Reply #5 on: February 21, 2010, 10:34:46 AM »
I have a Line6 dev kit so I have some visibility into the hardware.  DSP core, ADC on one side, DAC on the other.  Pretty much what you'd expect.  And an external memory module (512kB... ~5 seconds-ish).  Implementing delay on this kind of setup is trivial, you just treat the memory as a ring buffer and output samples from whatever time in the past you want.  Very simple code.

New-fangled delay pedals seem to be all about emulating this or that vintage delay sound.  So if you want to do that kind of thing there's a lot more DSP work to do.  But I don't know if I really can tell the difference.

It is understand by me that the amount of controls on the development unit is rather minimal making the unit not as universal as assumed. I've looked into going this path also but this  serious limitation, for mine particular requirement, took it out my scenario.

audioartillery

Re: DSP based delay question....
« Reply #6 on: February 21, 2010, 01:55:09 PM »
Peter, I'm not sure what you mean by that.  6 knobs, 2 3-way switches, and a two-stage pedal.  Is that what you mean by controls?  Or are you talking about the internal architecture?  I'm not trying to talk you into a Line6 kit (I was just giving an example of a DSP-based delay implementation), just want to understand what you mean.

peterv999

Re: DSP based delay question....
« Reply #7 on: February 23, 2010, 05:21:28 AM »
Peter, I'm not sure what you mean by that.  6 knobs, 2 3-way switches, and a two-stage pedal.  Is that what you mean by controls?  Or are you talking about the internal architecture?  I'm not trying to talk you into a Line6 kit (I was just giving an example of a DSP-based delay implementation), just want to understand what you mean.
Audioartillery, Very much agree that for modern echo gear this could be sufficiant. However, in the old ages of vintage gear you should think of minimal 14 knobs and 3 switches and a 60-80 position selector for patches to get all the parameters in.  My remark was related to that limited amount of controls assuming people with interest would also take that into account avoiding costly mistakes.   Kind Regards  -Peter

Taylor

  • Poster2
  • **
  • Posts: 4116
  • Total likes: 61
  • The clean energy source of the future.
Re: DSP based delay question....
« Reply #8 on: February 25, 2010, 05:17:01 PM »
I'll vote again for the FV-1. Making a delay with this is easy as pie, really no coding would even be necessary as the code to do this is already posted at the Spin Semi website. You can get up to 2 seconds of delay by using a 16k crystal. This gives you response up to 8khz which IMO will sound fine for guitar.

cloudscapes

Re: DSP based delay question....
« Reply #9 on: February 25, 2010, 07:24:11 PM »
on the FV-1, do you have random access of the audio buffer?
~~~~~~~~~~~~~~~~~~~~~~
{DIY blog}
{www.dronecloud.org}

Taylor

  • Poster2
  • **
  • Posts: 4116
  • Total likes: 61
  • The clean energy source of the future.
Re: DSP based delay question....
« Reply #10 on: February 25, 2010, 07:29:04 PM »
Hmm, do you mean for doing "buffer override" type effects? I don't know the answer to that, but if you're aiming to do those glitchy buffer-read effects, that can be done just using delay lines, I've messed around with it a bit.

I'm not a guru with digital hardware at this point, so that question would be best directed at their forum:

http://www.spinsemi.com/forum/index.php

It's quiet there, but Frank Thomson, co-designer of the chip, checks in daily to answer questions.

peterv999

Re: DSP based delay question....
« Reply #11 on: March 02, 2010, 01:32:29 PM »
on the FV-1, do you have random access of the audio buffer?

cloudscapes,

Have a read on my www.Echotapper.nl site http://www.echotapper.nl/dev/index.php?option=com_content&view=article&id=41&Itemid=45 it possible provides you with the answer. I've chosen this FV-1 product as it allows directly for multi head/multi amplitudes (like in the old days with multihead tape/disc systems). A 6 head simulation is only using < 35% of the programming memory leaving ample space for filters/flangers. A real GEM this DSP unit from Spin Semi.

Peter