What will make my design unique is that there will be digital "pots" to control the various tweaks of the circuit which would otherwise by done with several knobs. The reason I'm doing it this way (which is quite a bit more complicated) is to have the ability to load presets quickly along with a song, having the FX board set itself up so I'm not tweaking it up manually (which rarely works out for me it seems)

  After playing around with one on the Wish board I was impressed with it's clarity compared to the old BBD (Bucket Brigade) echo chips. It truly has 44k of RAM and as long as the echo is inside of .3 seconds it sounds good. Some DIY'ers are trying to figure out how to increase the delay time, perhaps by ganging them together which works in theory, but needs some pretty fancy circuitry that quickly makes the whole thing futile.

  On of the main drawbacks of this chip is the VCO resistor control is just that. There's no clock in, and the VCO input won't clock. It's likely filtered and a ramp style Osc. that uses steps of the ramp to execute internal workin's. The chip is great for my application to be a reverb/presence/triplet type FX.

  The Show-In-A-Box will have 2 FX channels, one echo, the other reverb. The echo can also feed the reverb, which is what I do often with VSTs in a DAW. The diagram tot he right is a basic idea for a decent reverb. Part of the design was kinda stolen from somewhere, so if you're that guy cheers! It appears to be a bit overbuilt but does reflect (oops pun) how a reverb works. Each one of these 5 chips has a slightly different echo rate, fairly tight to only 50 ms or so.  I have some feeding into the left output channel and some into the right more so it'll have a 3-d effect.

FX Control:
 The resistors in circles are actually CDS cells acting as potentiometers. These will be illuminated by LEDs driven by a NON-PWM circuit (below right) which I'm pretty excited about really. I call it a SASH, or "Sample-And-Sorta-Hold" circuit! Because of the frequency constraints of PWM outputs in such large numbers, it's impossible to drive the LEDs this way. The CDS cells will pick up the PWM "tone" and put it into the oh-so-sensitive audio circuitry like mad.

A proposed design for FX-2

 This method isn't perfect if you're looking for linearity, and will tend to be inverse log10 type of curve, but that can be compensated with the PIC-table-DAC curve. (I've done them before but they're a bit of a pain to get linear output) For FX levels it's definitely enough precision.

 If everything audio wasn't going to 50 bit digital processing, someone may have even created this inside of a chip (would be sweet!) but that'd in a different universe I suppose.

Update May 7th 2014: As can be seen in the diagram, deciding on which CDS controls what is getting clearer. You can see the DAC (12 bit) feeding into the analog de-mux. In order to use the entire range over 5 volts, the LED needs to be pulled down some. 5v -(.6 +.6 + 2v) = 1.8v max. So the cathodes will need to be on the negative side by 3.2V. It should be adjustable though as it depends on the LED.  This board must be kept away (Or shielded) from the sensitive FX audio board so the uP noise doesn't get to it.  The distortion circuitry (below) will be driven by another 74HC4067, but that may also reside on this board now as I've changed the circuit away from analogue filters (for now hehe!)  With this much control, especially over the echo or reverbs' VCO, a lot of amazing effects can be created. For example, ramping the VCO with do a pitch shift at the output. If controlled by a MIDI file, notes could be made to "portamento" into the next.  Introducing a small square wave at the VCO can cause an octave shift. I played with that for quite a while. Great for "space music" ! If a large complex wave is introduced, the original input is so different that a new instrument is created. Even a generic MIDI synth like the VS1053 I'm building into this project, sounds great!  Kept in sync it really twists things nicely, sorta like the Kieran Fosters' VST dBlue Glitch. Sweet!

----------------------FV-1------------------------

FX Update! June 9th 2015

 Unfortunately (and fortunately) I have found a better solution than the PT2399 for the FX board. (Even after I made up all the boards!) It's called a FV-1 DSP by Spin Semiconductor
This chip is entirely unique that it is completely programmable (using their free IDE software and an external i2c EEPROM) and even comes with 8 reverb and FX built in!
 It was designed by a fellow named Frank, and the Founder of Alesis / inventor of the ADAT, Keith Barr (who sadly passed away in 2010). The ASIC (App.Specific) programming is a little hard to grasp at first, but gets easier as one progresses...totally different from PIC or Arduino.

 If you're trying to decide on an FX chip, get it! I had to get mine from Germany (Das MusikDing), but it was worth the wait/$25. Here's why:
It can do simple or complex (EQ'ed/filtered) reverb with all of the advanced features that a VST may have, without the latency.
It can do real on-the-fly pitch shifting with a minimal delay, stepped as chromatic notes, or sliding (like off a pitch wheel)
It can do FM and Chorus FX, it's main purpose, very well.
Left in and Right in can be split and are completely independent in the programming, so one could do a pitch-shift and a reverb (quasi-stereo) at the same time if looped back through an LPF.
It can do distortion algorithms and had I known of the FV-1 then, the distPIC (below) PIC would've been controlling an FV-1 instead of processing audio directly.
It can do all FX, tremolo, FMing, filtering (LPF/HPF/Notch/BPF with Q factors), synthesis, and even sideband (I think I've figured out how) inversion.
 A pretty amazing chip!

As can be seen in the image, it's an SOIC sized chip, runs on 3.3V (a bit inconvenient), and is open source. It has integrated 16 bit ADC's / DAC's (sigma delta) which makes it easier to design around, 24 bit logic internally, runs off a 32KHz clock xtal (or VCO), all the hard stuff is done!  A bit of a disappointment is the 3 pot inputs, 4 or 5 would have been nicer, but there are ways around that. The EEPROM interface, allowing an extra 8 "programs" to be accessed, is limited to only 8 as the selector switch is 0-7 and internal/external programs. Right away I decided to circumvent this limitation by using a PIC chip to emulate a EEPROM. As it will control the selector/internal/external inputs on the FV-1, it will be able to send appropriate 512 byte sections of the PIC's program space.  For the SIAB project I will use the "Auduino" controller (FX Sash board) which is already made at this time, and some resistors acting as level shifters. These have 12 bit resolution so can be fine tuned if needed. The knob 0 input will also serve as a tempo input that will be used for certain FX I plan to program. Echo in sync can be desirable as well  right?  The 4066 switch (Tap Tempo) has been replaced by a simple NPN transistor, but could be hacked back over to a spare 4066 switch as 2 are already being used to switch out the feedbacks, just to be certain! The CDS photocells can go in excess of 1 Megohm but still a very small amount will trickle through, and depending on the program running in the FV-1 could cause issues. The 4066's are of course decoupled at both ends, being single-supply.
As can be seen in the diagram below, the outputs are also using CDS cells to control level to the Main Bus and Aux Bus. These are almost just a secondary level control as the FV-1 can do all of that from a Pot as the "Wet" control, but if I want to pass the FX output through the Post-EQ (return EQ) first, the Main Bus will need to "disconnected". The Post EQ is yet another TDA-7718, which makes 8 of those total in the SIAB. (One in the DistPIC Module, this one, and 6 input channel EQ's) The inputs are directly from the mix of those TDA's LR/RR outputs so any input channel can be fed into the FV-1 at any level, even separately (I plan on running some FX as FX-1 & FX-2 as before)
Final FX board design using FV-1 IC, a PIC18F2539, and  LM833 super low noise dual op-amp.
What's really amazing about this chip is the low noise / high frequency response! With a 32 KHz Xtal (resonator) it can still sample at that because it has internal frequency multipliers (runs at 32 Mhz I think) and with a sample rate like that, the nyquist thingy dictates that it's useable up to 15 KHz or so, which is plenty. If a higher freqency Xtal, say 44 KHz was used, it'd be at CD sampling rate, but delay memory would be somewhat less. The delay memory at recommended sample rate is 1 second. Plenty!  The PT-2399 (top of page) is fine for guitar and background reverb FX but 1) For reverb several are needed+echo, and that's at 30mA per, so total draw at 5V would be 210mA (7 of them). A lot! 2) The cut-off is around 4 KHz max, so not great for 'verbing sizzling vocals. I tried, it just wasn't there.   As can be seen to the right, the board layout is a nice compact vertical (off the "Auduino" mother board) that will have a right angle pin header. The 24 pin PIC18F2539 is through-hole mounted as there is more room on that side (clears surrounding boards) for the required socket so it can be romoved for programming.   Programming the DSP EEPROM emulation memory will be easily done as the flash PIC's have "self programming" of program space, almost the same as their built in EEPROM space. This chip has 24K so I'll likely use 20K for the FV-1 programs. That makes a cool total of 48 individual FX. Some of those may be duplicated with changes to what the 3 "Pots" do.   The only thing I haven't decided on is whether to send hex files via RS-232 to the "Auduino" pro-mini chip to send on to the PIC via it's USART RX, or to have the ATMEGA (Main Hub) read the programs off of the SIAB SD Card, then decode & send the data to the "Auduino", which in turn relays it to the PIC.  The code (an image ASCII HEX created by a Macromedia Flash routine) could also be be directly pasted into PIC MPASM during program time, which is a bit cumbersome. So many choices!   Finally, you may have noticed a "Width" CDS in the diagram above. This basically mixes the left and right channels using an external SASH drive from the FX Sash board. Kinda crude, but it'll do the job. Using 4 op-amps to do this is just a waste I think!
	A small issue that needed to be addressed (well not small if ignored!) is the logic voltage of the FV-1 vs. the PIC: 3.3V & 5.0V. I was thinking about using the diode method, but because of uncertainty on how the FV-1 pull-ups would react and how it handles the ACK, I found this nice level shifter using tiny FETs. Wow, why didn't I think of that?! Now I'm going to re-design my SD Card reader board as this could double the speed.  Anyway, I think it was posted by LadyAda from Adafruit, so kudos you guys! ( I bought one of your "Trinkets" to celebrate! ) The BSS138's were purchased for this specific purpose, which I almost NEVER do, so that's saying something.... it'd better work, I bought 10!  The inputs (switches and pots) are just using resistors as voltage dividers because speed isn't crucial at all with those. The PIC selects "external" then the address (0-7) of the program. This causes the FV-1 to try reading the I2C bus, so the PIC must be a Slave.

 

To the right is a photo of the "Auduino" mother board, (sitting on Gena's workboots!) and an arrow pointing to where this board will be. A fairly tight space with not much on either end for room. The height is over 2" though, which won't be a problem for clearance beneath the panel, but will need to be braced on it's own (so it doesn't "unplug" from vibration) being so much higher than the other boards. The SASH boards aren't yet joined together in this photo. As with the MySynth II the CDS/LED pairs will project through the board through holes in black foam tape. The LED inside board is painted black to minimize stray light from the adjacent LED and the outside world.  Because the FV-1 uses 4 SASH circuits, and the 4 PreAmp SASH LED/CDS pairs are on another board, the CDS board is much shorter (closest in photo right).

"Auduino" motherboard with some SASH cards installed. The small board is a TDA7718 channel

 Above is the final "Auduino: SASH control configuration. It can no longer change as the boards are done! The AUX taps are for secondary outputs (can be back room or stage monitors, or both!)
"PLY" on the panel is now "Pan Level Y" and AGC is "Aux Gain Control" , both for the AUX1 output. The Aux2 output is controlled by the MainOut EQ board.

 Once again everything (the board draft) has been done is Flash MX. I don't know how I'd live without it. Maybe one day I'll switch over to Eagle, but not today!

The project (and all projects) is on hold until after summer, but check back often in September 2015 for more! (June 2015) Cheers, Sandy

 Distortion PIC

 This next section will cover the progress on a distortion board I call the DistortionPIC. I did some experimentation with a PIC16F74 and a DAC08E that's been sitting in my parts drawer since the beginning of time. After hooking it up to the PIC and writing a program to basically feed the ADC input through to the DAC, I was amazed that it still works!

The schematic to the left is pretty much exactly the way it was on the wish board. Let the fun begin!  I wrote a sloppy DSP routine to act as a low pass filter and it worked pretty good so now the confidence is up. I tried modifying the waveforms using a few tables to replace original values and came up with a nice smooth distortion eventually. This is all new to me keep in mind.  Next I tried making a delay the phases in and out by saving 128 bytes into RAM, then starting a second pointer that would vary and ouput in a loop, like a temporary formant. Got some pretty cool FX! But not what I expected.  The sample rate runs at just over 30 Khz with 8 bits. I've never run a PIC ADC that fast either. The last bit's dither was audible, but it *is* a distortion board, and there should be some form of gate on it, just like most good guitar amps have.

Before I pulled it apart I got some recording of it during various test stages. I shrunk it down to be shorter and it's here, (the completed design samples are below)   The original idea, before the ADC was to use the PIC chip as a sort of CMOS buffer, which is ideal for distortion, but that's a bit too static for my liking, so using several I/O pins (highly sensitive as an amp will do the same thing. With added advantage of being able to control the dynamics with slight delays an a few caps that can be switched in/out, it becomes more appealing.

This diagram shows a (yet another!) TDA 7718 being used primarily as an EQ. The good part of using this chip is that it can be configured to be in different parts of the circuit! Having EQ before the distortion can be desirable to break away from the strip EQ, then mix it back in at the main output with/without reverb/echo FX. The TDA chip has 4 fixed midrange frequencies: 500 Hz, 1kHz, 1.5 kHz, 2.5kHz. Plus a variable Q of .75, 1 & 1.25 on that band area. There's also a Loudness control with 400Hz, 800Hz, 2400Hz add/null, and the bass 200Hz level + 4 Q factors that might come into play. Anything below that would be useless. The "frequency" of the chip can't be changed either. The TDA 7439 (no longer avail.) had ext caps on it's EQ. Will this be enough? I don't know! Only experimenting with a guitar will tell.  The plan is to have tweakability on this board, but also a nice set of presets that will set everything "just right" to get a certain sound quality!
The EQ modes, will allow for some flexibility by the position it's inserted at. The first insert, just before distortion, will change the dynamics of the distortion for sure.   At insert 2, the left side output is fed back into the Right side input, thus doubling the EQ's peaks or notches. The EQ would sit just before the flanger in this case.   At insert 3, the EQ is placed at the output. This is also a great place to have a filter. There's  6 other obvious modes the circuit can be in, but you may have noticed the LM386 off of the SW outputs? This is to simulate a cabinet using a speaker and a mic. The audio is sent down a PVC tube, then picked up be a mic. The mic's position can be changed directionally which should change the dynamics. This might be manual or using a tiny stepper/servo. The 386 may have a filter (CDS-C) on it, not sure.   The gate in the circuit is controlled directly by the PIC. The waveforms to the right explain the various ramps of Attack and Release, and the tremolo. The 47k resistor and the 1uF cap will limit the Tremolo to 20 hz, which is way faster than I ever use. CDS cells can go to a very high resistance, >1 Meg, but after initial darkness, can take 2-3 seconds to get there. It seems to change from one to another, but keeping the impedance of the circuits using these low will speed things up and stop leakage. In a worst case scenario, the 4066 switch at the feed, or the output from the TDA7718 can be attenuated at "gate" time. I like the smoothness of CDS cells in audio circuitry, no "clicks", and if the slope is right, feels very professional! Finally, the whole thing is controlled via the FX PIC using some sort of serial interface. All of the CDS cells are controlled from the DAC on the FX PIC as well. In fact these boards might be stacked, not sure yet.

 SO everything will probably be changed again, but hey this looks pretty nice at this moment!

Cheers! Sandy*
May 7th 2014

 

Update June 4th 2015   The above design turned out to be the best design, and I'm glad I stuck with it. The DistPIC module was completed way back in February. And I couldn't stop playing with it! Good toy. After some slight modifications, only 5 SASH's remain to be useable off-board, mostly because of the new stereo output control and gate. I've changed the schematic above to update.   Below left is the photo-etched PT2399 FX board. This yields a phased output for delays, reverb, phasing etc. If used with CH.2 output, it's stereo. The mix happens in the SOIC 4558. The photo to the right shows the whole thing put together. The top 2 boards are the SASH LED/PIC and the audio CDS side beneath it. The small bottom board is of course the PT2399 board.
If you look closely, the mess below is kinda crude. This is what happens with proto-types. This is before the other board stacks over it and hides the worst of it. It's all ok as long as it stays working! The photo below right shows the arcrylic black paint my neighbors lent me to block out the light from the LED-CDS junctions. 3 coats were needed, but inside the dark bowels of the SIAB it's probably not an issue. I had it sitting out in the studio for quite a while as I started the Sampler project just because hardware guitar distortion totally rules over a VST (I'll never go back!)
If you don't believe me I've posted some samples below. Of course those have been recorded digitally, and then turned into a muddy .mp3 so they won't sound nearly as good as the "real" thing, so I guess you'll just have to come to one of my shows!
Anyway, here's the original board designs that were etched. Don't go copying them to make your own, because they aren't very accurate. Traces were cut, circuits were added, others were removed, it's totally not the same. Maybe one day I'll amass the pages of scribbles and notes and draw up a proper schematic. If I was getting paid I would tomorrow, you know how it is.   The upper board has the 40 pin PIC16F887, (I like because can self-program, good for the 128 presets the distPIC has) encompassing the external SASH circuits, and the 16 SASH selector (4067). The right side of that board has the LEDs (602 size LEDs) facing down through drilled holes marked by the circles. The hole drilled removes the "short".   If you look hard you can find the 6 pin 12-bit DAC. I love those tiny things! (It's just above the word "DEC") unique address I2C happiness!  On the lower board you can see the TDA 7718 footprint, next to a 4066, and to it's left an 8-bit DAC. Sample rates make up for the low bits and it is after all a distortion board. To the right of the TDA is the now never-to-be-used-again-in-an-audio-circuit LM324 footprint. It turns out the 324 has a nasty noisy crossover point at zero volts (as this board uses +/- 12V) and I ended up having to offset it by +6 Volts to make it stop with the "crackle" on lower levels.   The far right board, which wasn't modified much, is for the 2 PT2399 delays, and the bottom tiny board is for the LM386 500mW amp to drive the cabinet speaker. This is mounted in the lower section of the SIAB Box. See? pretty simple when you break it all down.
What's tricky is understanding the changeable circuit configurations. The distPIC can re-configure it's circuit 8 different ways. To simplify this the user manual breaks the circuits down into their actual function by way of 8 separate diagrams. These are in the Users Manual. I've taken to creating a user manual for each module, or part of the system, because it's a lot to remember!   Creating an FL Studio DashBoard for it was a very good idea, although it doesn't explain how the circuits are configured. It does give a good visual interface.   Once a preset has been saved to the distPIC, it will configure the settings to exactly where they were. A loaded song in SIAB need only have the preset (0-127 above dashboard window) number and assign the panel knobs as the song requests.   The pedal control shown (standard MIDI CC#04) can connect with a real pedal, or be used as another preset setting. This dashboard is the most stylish one I've ever made, and yes it seems easy to control.   The only real issue is the servo. After a preset is selected, the servo won't adjust for about 5 seconds, so it's not flicking back and forth wearing out as preset knob (if present) is turned. The main problem is it creates quite a power surge which can be heard if connected to a big amp. It's a short, deep buzz for 1/2 second. If the servo isn't required (no mic movement) then the bypass button can be turned on by the song sequence before the preset is changed.   I never thought this would be an issue, but changing the way a guitar sounds during a song is amazing! For example, going from British distortion to a smooth twang for the chorus. Wow!

   OK the moment everyone's been waiting for, the samples! Please keep in mind the audio connections where unshielded alligator jumpers for some of this ;)
   The Samples are as recorded off the mixer in Reaper (an awsome DAW software I now use in place of Sony Asscid Pro) unmodified with any post FX etc.
   There's so many modes and configurations, (21 modifiers, 15 Types, 8 configs) it's hard to understand without reading the user manual, but I'll try to explain what it is you're hearing...
 

**Digital Processing Modes** Retro bit, Retro Sample Rate, Digital Phasing, Wave Riding, Fuzz, etc etc. Lots of digital toys to modify the feed! Here is a potpourri of short samples, too numerous to list. Excuse my 14-year-old heavy metal technique, that's only where I'm at so far with electric guitar lol!	Drums Masher! The distPIC isn't only good for guitar! Listen to what it can do to drums. I announce the Type/Modifier for each. The drums are from a GM Kawai set (not the best for sure) but you'd never know it now! Some digital processing modes are in this (fairly long) sample along with EQ, delay, and various Types.	Mono to Stereo Twang The distPIC was originally to be mono, but I couldn't resist tapping off to make it stereo. The 1st strums are mono, the rest is stereo. You may notice it's on the verge of distortion. That's because this is a real miss-use of the distPIC hehe!
Noter Type This is a really basic use of "Noter". It'd be best used minimally at the end of a solo perhaps? Too much is too much with this unless you're into Thrash or Machine music. I think it's freakin' awesome! It's just notes controlling the EQ+outs.	Tremolo This is a clean sounding Tremolo. The tremolo can be sine, ramp, or reverse ramp. The tremolo can sync with the tempo as well which is sometimes desirable.	Mono to Stereo Bull Beef Distortion Another mono to stereo clip. Notice the higher dynamics sneaking through even though the distortion is very heavy? This is because of multiple paths (and serendipity!) to the output.
Noter Wawa A short sample of using Noter (or Arp) with same test beat but in Wawa Type Mode. In a song sequence there would be more variety of course. I tried a synth into this later on and OMG!	Jump The function of the Jump is to change the dynamics of EQ Bass and Midrange, but only during moments of "silence". It scans at LFO rate so can be predictable. The controls are extreme for the example.	Clean Auto-Wa I made up a little song to demonstrate the "Clean Flange" circuit config. Notice the "boxy" sound of the guitar, and the mild, almost imperceptible  flange behind it.
RR Arpeggiator With another instrument, using RR Arp (level Arp) makes a guitar almost synth like. If I knew better how to play, this would be much more intense, but you get the idea!	Retro & FM Retro bit mode (not retro sample rate) is used to tin up the jabs at the first part of this sample, then a drunken FM chorus takes over. The delay is being controlled by audio level to the ADC, always a fine effect.	EQ Arpeggiator The idea of arpeggiating the EQ controls came to me when setting the EQ in FL Studio to change with the beat. Why not just make it a Type? So it is and here's what is can do! "Noter" can do more than this, but needs a song to control it.
Noter Blast This is a mono-chord demonstration of how "Noter" can twist a pluck of a chord into something pretty cool! As I play, I can get into a groove with this, but the limiting factor was the piano playing in the background so I had to stay on that chord.  It makes for an interesting sample though.
Synth Feed I decided to pin a key down on my Korg and feed the audio in so I could play with the distPIC to see what kind of sounds I could get The first part is a "whistle" sound, which is run through the note-shifter, and the rest are full-bodied multi-octave notes. Once it sounded nice the notes change. Most of this sample is using the EQ-Arp, but has been chopped up to make it shorter as I was playing for quite a while! Sorry about the annoying drumline, I was testing for sync mainly.
Cabinet (Speaker/Mic) This short mix of samples were created while testing the cabinet feature. The servo turns the mic away from the center speaker which changes the dynamics. When mixed with electric signal, some nulling can occur. I've just changed the speaker, so not sure how that will affect the overall results. The mic had to be turned up too high and would cause feedback from the big amps.

Users Manual