LFO DSP

[Paul Perry (Frostwave)]

Don't forget the one/F noise generator!! and I bet you could sell a lot of long cycle shift register noise chips, the old Motorola ones are out of production and sounded bad to boot...

[R.G.]

In terms of packages…. TQPF gets pretty small and I have yet to try soldering a MLF package. MLF gives you 32 pins in about the same form factor as a DIP-8.

Aye, there's the rub. The DIP package is just about as fine a pitch as the typical DIYer can handle. That's one reason I never put any of my surface mount stuff out in public. There would be three (about) successful builders and a hundred thousand people who never could get it soldered right.

Don't forget the one/F noise generator!! and I bet you could sell a lot of long cycle shift register noise chips, the old Motorola ones are out of production and sounded bad to boot...

This is one where the eight pin 12C5xx series shines. You can make a sixty four bit shift register noise generator in about fifty lines of code.  It's substantially easier than a sine or other table look up LFO if you count the work to generate the table. And a noise generator only three pins: power, ground, and digital out.

In working with programmable uControllers, you quickly learn that it's not the function that's the problem, it's feeding the function out through the limited number of pins.

One thing I've been messing with is a bass drone pedal, similar to the Taurus.  There's a "top octave generator" set of source code for the PIC on the net. It actually generates a middle octave because of speed limitations, but it's pretty simple to get it to make lower octaves. How about a three-chip pedal that has twelve momentary footswitches and puts out a bass note when you press the "key"? Actually, for a CD4046 and a CD4024 per note, this chip does generate all octaves, as you can phase lock the notes up from the octave it does generate.

sigh... so many designs, so little time...

[amz-fx]

Are you going to make that a project and release source etc?

My plans were to make it a project but not release the source...  for some of the reasons you mentioned but also because a user would have to build a programmer in addition to the project and then learn how to use the software to program the AVR...  it's simpler just to sell programmed chips.

re: the noise generator
A 32-bit max length LSFR will have a  2^32 - 1 cycle length, or 4,294,967,295  bits  or 536,870,912 8-bit words. At the fastest rate of 78Hz, it can run for 6,882,960 seconds without repeating (114,716 minutes, 1912 hours, 79.66 days)!

The top-octave was another project I had on my list but as you said, so little time!

regards, Jack

[doug deeper]

hmmm....still dont really follow....
i wish there was a site like this one concerning stompbox dsp....one of you
well versed fellas should make a diy project using dsp....like mabey an electra dist emulator :] hehehe....
naw....if there is any other info around id love to get into this stuff...
like what does the programing intale?
(software ect.)
and why would the $79 programmer be better than the diy one thats would cost about a buck fifty???
thanks alot,
doug

[Peter Snowberg]

Aye, there's the rub. The DIP package is just about as fine a pitch as the typical DIYer can handle.

This is a very good point, but as I'm really just offering my own semi-custom parts for others to tinker with, my focus is on my apps. After doing a bunch of projections, this idea just isn't a money maker in any way unless the chips are in pedals sold in decent volume (make that rather large volume). For DIY with ease, there are the DIP-28 and DIP-40 parts. Other adapter solutions like "surfboards" come into play too.

Don't forget the one/F noise generator!!

White noise is pretty easy and with an external RC filter, you can get pink. The chips just aren't fast enough to do the lowpass filtering in digital at any reasonable bandwidth.

A few months back EDN published this http://www.reed-electronics.com/ednmag/contents/images/di3227.txt which at least gets a white source on a PIC.

My plans were to make it a project but not release the source...  for some of the reasons you mentioned but also because a user would have to build a programmer in addition to the project and then learn how to use the software to program the AVR...  it's simpler just to sell programmed chips.

Understood. Since neither of us is looking at releasing the code, perhaps there is some good room for a collaboration on some articles about the subject? There are so many MCUs out there and the concepts are all quite similar.

hmmm....still dont really follow....
i wish there was a site like this one concerning stompbox dsp....one of you
well versed fellas should make a diy project using dsp....like mabey an electra dist emulator :] hehehe....
naw....if there is any other info around id love to get into this stuff...
like what does the programing intale?

Say you were building a car... it would be easier if you started with an engine and transmission that you know works. Building embedded systems is similar. If you get the real eval board from the device maker, the assembler/compiler is known to work with it, and you can generally load up example code and get it working right away.

I gave up on DIY development systems for things like this a long time ago. If I'm going to program chips, I get a real-deal chip programmer. It's neat that you can program many chips with less, but if your time is worth absolutely anything, it's really a question of the $79 unit from the chip maker, or the $1500 unit you build with $1.50 in components that still doesn't work as well in the end.

The $79 unit not only works flawlessly, but there's a LOT more going on there than with those $1.50 programmers.

When I got started with AVR processors, I bought the ATSTK-500, the ATSTK-502, and the JTAG ICE. I think it came to about $500, but the thought of replacing the functionality of these tools in a DIY context for less than $20,000 :shock: is highly laughable. I'm being quite serious about that too.

Oh, and the Atmel stuff runs from their environment which is called “AVR Studio”. Download it free at http://www.atmel.com

When I got heavily into the DSP56001, I bought a NeXT Cube because it came with the best DSP56K tools available and it had the chip on the motherboard. It still makes a good bookend. :lol:

On DSP....

Notice the name is DIYStompBoxes and not DIYAnalogStompBoxes. I'm going to try adding some digital in there because it's a new set of tools that does different things. There have been several DIY DSP sites out there, but they all seem to deal with non-stompbox uses. There is quite a bit about effects, but mostly in different contexts. Google around and you'll find lots.

DSP will be showing up more and more in the months to come. These DLFO chips use a combination of sequencers and DSP techniques to do their work, but I have a DIY DSP project in the wings that will come out as soon as I can get some extra cash (I'm really broke right now). It uses a couple Alesis DSPs for a wide variety of multi-effects at an amazingly low cost for what you get. The cost is less than $20 (in singles) for a complete A/D-DSP-D/A subsystem that handles reverbs as well as filters and other effects; and that's 48K, 24 bit . If you don't want to wait for me, Alesis-Semiconductor sells evaluation boards for their DSPs, but they cost a couple hundred.

When it comes down to it, programming microcontrollers takes some real study if you haven't worked in that area before, and DSP programming is quite non-trivial. It can be very hard to understand and amazingly complex when you consider that the "program" is really more of a machine description of a piece of calculus. I'm hoping the availability of DIY DSP effects hardware to serve as a starting point will help people to understand DSP and make useful effects without so much pain.

One place to start learning is to look at the PDF files here: http://www.stanford.edu/~dattorro/ (don’t read anything here if you have a headache )
Audio Engineering:
http://www.stanford.edu/~dattorro/EffectDesignPart1.pdf
http://www.stanford.edu/~dattorro/EffectDesignPart2.pdf
http://www.stanford.edu/~dattorro/EffectDesignPart3.pdf
Save them to your hard drive. Part 1 caused a big stir when it was published by AES because it may have let out some Lexicon reverb trade secrets. AES does not offer it as a re-print. Part 3 is about digital oscillators.

One neat thing about DSP in filtering is you use the same type of math you would for figuring out resistor and cap values, they just turn into coefficients instead parts values.

In closing, my micro-rant on digital emulation: DSPs are really good at doing chaos math, but distortion I feel is about entropy rather chaos. DSPs don't understand entropy except via statistical correlation. DSPs are good at making digital distortion effects, but when it comes to emulating those screaming red-hot electrons in a tube or the behavior of other materials, DSP just doesn’t cut it (at less than ultra-computer levels, once you can throw teraflops at the problem with enough samples entropy and chaos math emulating entropy, all bets are off).

DSP may be a good desert topping, but it makes a lousy shoe polish.

Take care,
-Peter

[Paul Perry (Frostwave)]

I'm not a programmer, but here is a fun thing with PICs and perhaps other micros: you can use the external clock as a master time control.
(yeah I found that out by putting 8Mhz crystals in a ckt that wnated a 20MHz clock.. everythign worked perfectly except the midi, of course.. took me "too long" to spot what was happening there :oops:
just struck me as a fun idea to have a variable 'master' clock!