LFO DSP

[Peter Snowberg]

I started sketching out the final implementation notes for a high quality digital LFO last week and even had thoughts of offering it as a preprogrammed chip for the DIY crowd that would just need a resistor ladder D/A and some pots added to make it useful.

If I end up offering this chip, it will be a commercial product just to keep the code under wraps from other boutique makers (sorry). With Guyatone now using a digital LFO, I guess it's finally time.

Feel free to make any comments or suggestions, but I may use them in the chip.


This is what is planned so far:

- Different chips with different capabilities
- low power
- up to 8 analog adjust inputs with 10 bit resolution
- resistor ladder (parallel), HC595 serial, or other outputs
- output widths from 1 to 16 bits
- multiple outputs available including inverted and phase shifted
- LFOs modulating LFOs
- Sine, Asymmetric Sine, Cosine, Square, Triangle, Hyper Triangle, Asymmetric Triangle, Sawtooth, Reverse Sawtooth, Integrator, and LFSR Noise waveforms
- Wavetables can be stored in external E2PROM (limited by internal RAM or EEPROM speed)
- 32 bit internal resolution
- Wide range of mixing oscillators
- Modulate wave speed and/or amplitude via external 0-5 VDC input
- Modulate wave speed and/or amplitude via half or full wave rectifiers on the audio
- Mode Switching via regular switches and analog pots
- FET switching driver-debounce-flipflop (if the pins are there)
- status LEDs available via direct out or HC595 serial
- LFO DSP chips will be based on Atmel devices
- Preset store/recall to EEPROM
- tap tempo (Thanks to MikeB of Prophecysound Systems)

I'll be using these chips in my own pedals, so as they become available I’ll let you know the details. If there is interest before that point, I might make them available faster.

Pricing will be minimal above part cost.

Take care,
-Peter

[mikeb]

Possibly an option for tap-tempo? That's what I've got under development in mine ....

Mike

[Peter Snowberg]

Cool ! Nice idea. Thanks.

Your Infinitphase is just jaw dropping! Beautiful pedal!!!

Take care,
-Peter

[ExpAnonColin]

That sounds excellent, Peter.  I'd definitely give them a try if you could keep the cost below $20.

-Colin

[Peter Snowberg]

DIY Community Only Pricing:
The cost of the most basic DLFO units should be about $8 while the more complex chips with internal 2Kx16bit wavetable storage and I/O for LOTS of switches and options should be about $21. A mid range should be roughly $13.

The $8 chips should give most people everything they ever wanted in a universal LFO and more. The big problem is configuring controls.

With so many options, I may just make a web form for specifying custom parts. There might be a one time cost of a few dollars for a new part, but all reorders would be a standard price.

DIY pricing will be available exclusively to users here and in DIY quantities. Chips will cost commercial vendors a little more, but not much.

All parts will be available in TQFP with more basic parts available in DIP28 & DIP40.

Take care,
-Peter

[Peter Snowberg]

Additional Features:

- MIDI Conrol
- Serial Control
- Clock Analog Muxes with any waveform (just use parallel 4 bit output + a mux or two )

[ExpAnonColin]

Nice, Peter, I can't wait.  Where's the datasheet  :lol:

If I make a design I like with it and want to sell it through NSFX, does that still count as DIY?

-Colin

[drew]

Nice work Peter... I'd like to check one out. Are you gonna make a rudimentary board to go with it, in the same vein as a "prototype kit" that's available for a lot of commercial chips? (if you did, I'd almost certainly get a couple to use with pedals/synths.)


drew
www.toothpastefordinner.com

[Peter Snowberg]

Thanks for the comments.

If I make a design I like with it and want to sell it through NSFX, does that still count as DIY?

If you're selling that's no longer DIY, but don't worry, the commercial licenses will be similar for low volume. I'll program limited numbers of chips, but any real volume would be done by license and somebody else would burn the chips. If you're just selling a few pedals, you're still pretty much DIY.

Are you gonna make a rudimentary board to go with it, in the same vein as a "prototype kit" that's available for a lot of commercial chips? (if you did, I'd almost certainly get a couple to use with pedals/synths.)

Hmmmm I hadn't thought of doing that. I'm really just making some chips I'm already going to be working with available to others and offering to make custom versions too. I'll sure have app notes showing how the chips could be integrated into any situation via opamps, optos, JFETs, MOSFETs, or BJTs, and I will do an ExpressPCB layout for an evaluation board that could be used in the real world, but I don't think I'll go much beyond that. It's all pretty textbook microcontroller D/A stuff. Hook up power, some pots, optionally some switches, and some resistors to form a D/A. Level shift with an opamp or transistor, and use the signal.

My intent is to create something with as much analog feel as possible in both operation and interfacing (except for MIDI or serial interfaces )

Take care,
-Peter

[troubledtom]

peter do you really know .......what you're getting yourself into .... this could be very fun . DANN GREEN is also on the diy forfront ,he's been doing this stuff for years.
  he's a friggen great designer.
              wishing all the pedalpals the best,
                       - tom  

ps: the work you do will never have an end. prepare fro it. : )

[drew]

Peter- ok! Couldn't tell whether you were doing a PIC/similar thing or an actual fabbed chip. Originally I thought you were making a new die/etc and was intensely impressed...

If it's a preprogrammed microcontroller I guess it would be pretty easy to hook up on perf. I have put together some simple stuff like midi-din converters, midi-cv converters using PICs and they are pretty straightforward and simple. Of course, I'm sure that's partly due to the programming...


drew
www.toothpastefordinner.com

[Gripp]

Yes!!!! Indeed nice work Peter!
You've got a buyer here.

Best!
/ Pelle Garpebring

[Mark Hammer]

That sounds like an extremely worthwhile project/product.  I might point out that there is a huge modular analog synth community/market that would probably welcome a small "modulation design" module that could supply a wide variety of designable modulation waveforms.  I posted a versionof such a module from way back - the Elektor SEWAR - on my site, but the amount of stuff you have to incorporate to generate a wide variety of waveforms in the analog domain is costly, huge, and often less flexible than you'd think.  What you propose just makes a huge pile of sense (and hopefully a huge pile of cents!).

Obviously there are limits to what can be stuffed into a PIC-like controller, but I'd like to suggest one additional feaure that I *think* can be readily implemented.  Over the last few years, a number of people have posted queries about how one would go about "freezing" a sweep on demand, then resuming.  Seems to me that in the digital domain, it's a simple matter of reading an input (i.e., a momentary switch), latching the current wavetable data ("freeze"), then continuing the readout from the wavetable conditional on the designated startup signal (which could be release of the momentary switch or pressing it again).

For an even more generic modulation source, it would be nice to have some means to skew or stagger two modulation outputs, such that they can range anywhere between 0 and 180 degrees different in phase.  That would incorporate in-phase, antiphase, quadrature, and all points in between.

Of course, the desire is to keep it small simple and cheap, so I'll shut up now.  :wink:

[mattv]

Thanks Peter!

[toneman]

As long as were dreaming...
What about LED outputs that R Plug-N-Play(?)

To make the *look* of that stompbox from NAMM
with the 2 columns of LEDs (recently mentioned)(?)

That MoogerFooger MURF could use some LEDs also.  LOL!

All the Wierd Synth Modlules have/had a "knight Rider",
or could it be a "cylon"  eye(?)

BTW, remember the PAIA "Sheppard Generator"

No LEDs, but
8-phases of  modulation output!!
2 waveforms, 4 phases!
No fancy chips.

dream, dream, dream
tone

[R.G.]

Good going. That's exactly the kind of thing I had in mind with the ASMOP stuff.

One thing you might want to consider. I am partial to the PIC series, largely because that's what I learned on. They have an eight pin version that I think I can get to put out a PWM LFO output. For DIY, the small chip footprint is going to be a real advantage. It's simple enough to program different chips for different waveforms.

The simplicity of getting anything you can think of inside a microcontroller rapidly leads me to the onset of creeping featurism, so I have a background task running that pops in from time to time with a "How *small* can we make this?" while I'm designing.

[amz-fx]

Peter, I have a working model of a digital LFO on my workbench right now... it is not as full-featured as your plans though:

Atmel AT90S2313
8-bit output resolution
R2R digital-analog conversion
EEPROM wavetable
Sine, triangle, square, sawtooth, hypertriangle waves
32-bit maximal length LSFR digital random noise
0.3Hz to 78Hz frequency range
status LED
no tap tempo but I'm working on that...

Here's a pic of my homemade programmers:



I'd offer my assistance but I'm sure you are probably more capable in this area than I.

regards, Jack

[R.G.]

Jack,

Good to see others lining up behind ASMOP.

Do think about getting it into a smaller package if Atmel offers that. It's OK to have several different models, each of which does the one or two LFO waveforms that work in a pedal.

Also, there is quite a bit of merit in having the LFO waveform swallowed into the chip. For instance, the outputs can be just the PWM outputs directly if you're going to use a switched-capacitor or modulated resistor to control something on the effect. External LFO voltage is most useful in strict retrofitting on voltage controlled pure analog things, as in using an LFO output voltage directly into a smoothed voltage input on a VCA for something like tremolo, cross and multi fading or morphing, or multi mixing.

[doug deeper]

how do you get into this stuff!!!
i want to learn!!!
need new noise!!!
doug

[Peter Snowberg]

Thanks for all the replies and PMs everyone. This stuff is more of a hot topic than I thought.

peter do you really know .......what you're getting yourself into .... this could be very fun .

That’s the problem with electronics. I was hooked by age 5. :lol: Does Dan Green have a URL? I’ve heard his name associated with the 4ms guys.

Peter- ok! Couldn't tell whether you were doing a PIC/similar thing or an actual fabbed chip. Originally I thought you were making a new die/etc and was intensely impressed...

Oh no, I’m just a Bozo. I do PALs/GALs and I’m trying to start with Xilinx but VHDL is way beyond me at the moment. After that, it’s a matter of big cash. I’ll be able to do digital custom by the end of the year, but I’ll never approach anything analog custom. You need to be a wizard for that. The digital stuff is all automated. Enter your schematic, synthesize, verify, sell a kidney, and send the files to Mosis. http://www.mosis.org/Orders/Prices/price-list-domestic.html

The chips will all be Atmel and most may be ATMEGA series microcontrollers. Jack may have something planned with an AT90 series chip! We’ll have to see.

Yes!!!! Indeed nice work Peter!
You've got a buyer here.

Many thanks Pelle! Greetings to Sweden. â€" Peter Snöberg

That sounds like an extremely worthwhile project/product.
….
Of course, the desire is to keep it small simple and cheap, so I'll shut up now.  :wink:

Thanks for all your comments Mark. I hadn’t thought of sales to the synth community, but that’s a jolly good idea. Thank you. One thing that inspired this was seeing the process and availability of chips like the CEM3340 and CEM3310. Now that digital oscillation and dynamics processing is easy and relatively cheap, it might be the right window of opportunity for something like this.

A freeze switch! What a wonderful idea! I was already planning a “reset” function that would sweep to a predetermined point and start in a predetermined direction from there, but a freeze command is wonderfully elegant! Thanks! Hmmm… maybe a little deceleration on the freeze and a little acceleration on the resume. That could of course be extended to high frequencies where a moderate slow-down would be like a rotating drum slowing down. At flanger sweep speeds you would want the halt a little shorter I would think.

One of the things I want to do with this is modulate the modulation. Phase offset and modulation are in there. I was thinking of a vibrato/tremolo for a Leslie simulator that would try to address the directionality of the rotating drum & cone a little more by using two synced LFO waves. I was wondering how many Leslie simulators combine both vibrato and tremolo?

I’ll take a look at the SEWER on your site. Thanks!

As long as were dreaming...
What about LED outputs that R Plug-N-Play(?)

For some models, there will be direct LED outputs. Just use a resistor to limit to 20ma MAX! other models will use 74HC595 serial/parallel chips to drive the LEDs without requiring a huge and expensive microcontroller. I like LEDs too.

If you wanted to get a “cylon” scan of an analog mux, that can be done with a triangle wave. I suppose there’s no reason why an 8 or 16 position LED bar graph meter couldn’t just be hung with one or two 74HC595s.

I’ll have to look up the Sheppard Generator. This series will also have up to 8 outputs of any phase, but with up to probably 32 or 64 oscillators of a number of basic wave shapes or wavetable data that can also be mixed, but it sounds like that one is an analog “uncle” of this sort of design. Think of this as a POD for LFOs.

Thanks Peter!

You’re welcome. I’ll be very happy to see people post pedals with a DLFO!

Good going. That's exactly the kind of thing I had in mind with the ASMOP stuff.

Thanks for the comments RG and the background task! When you realize you can do that and that and that and that and….. well… Luckily I’ve made a living doing embedded systems so I’m aware of the firmware limitations, but when these chips start having 128K of flash in them, well…. Feature creep can be an issue.

I’ve been doing PICs for close to 10 years (in two months) and while they’re neat, other chips have caught my eye recently. I got heavily into the Scenix SX28 and did a 230.4KHz DPLL with 20ns resolution controlling a software UART that read FM0 HDLC coded serial (LocalTalk + software version of the Zilog 8530). Cool chips.

For the past year, it’s been Atmel for me. I got a JTAG ICE box and I’m loving it.

What I may do is just macroize the code and use a web form to specify custom parts with a wide range of capabilities. Select your device, I/O style, features, and configuration data. Each of these chips may have some config registers accessible via serial. A couple serial commands should be able to customize operation for any set of switches or controls.

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.

Peter, I have a working model of a digital LFO on my workbench right now... it is not as full-featured as your plans though:

That’s great Jack! Are you going to make that a project and release source etc? I was thinking of concentrating on the 12-16 bit end. It looks like your chip is what most people will be looking for I think. Very cool . It sounds like they’re very similar but the AT90 chips are cheaper and if 8 bits is enough, then why hassle with larger more expensive parts? I’m thinking more of the MEGA parts right now and of integration into my pedal designs. I really appreciate the offer of help, but since this is for my commercial ends, I must respectfully decline. I thank you much though.

how do you get into this stuff!!!
i want to learn!!!
need new noise!!!

The Atmel AVR chips are pretty cool and fairly easy to program. There are lots of web sites about getting started and the tools are pretty cheap now. Check out Digikey (get a free catalog!). Try the ATSTK500 for $79 here: http://dkc3.digikey.com/PDF/T041/0288.pdf

It takes a while to get going, but once you do, there’s nothing like DIY embedded computers. If you’re more adventurous, you could always build a programmer like Jack’s two examples above, but the Atmel ATSTK500 will make life MUCH easier.

You may also want to check out PICs. Cool chips with lots of features and a low price.


Take care,
-Peter