R.G's Low-Fidelity / Telephone / Noise Pedal - Noise section debug

[R.G.]

I think many folks on here are starting to pick up on PIC stuff now.  I think the draw back is the learning curve plus the cost of programming tools.  If you think about all of the resources available purely for guitar pedals, it seems odd that no one has offered the paint by numbers route for PIC programming yet. 

It does, doesn't it?

[PRR]

> PIC 10Fxxx .... A 31-bit feedback shift register repeats in 2.147 billion steps

Cool! Wudda liked that back in the day. And you can't beat the parts-count/cost.

But George said "nice smooth white noise ...  Maybe more snap crackle and pop?" Which opens a big can of worms about psycho-acoustics and un-randomness. But we know what he means. Most of the interesting "hiss" is not pure random.

Mathematicians have labored to reach maximal randomness in pseudo-random machines. All possible combinations equally. What George's ear seems to want is less uniform randomness. Big chunks of curd. Grit in the groove.

Obviously we can mis-wire the shift register. Get hundreds of 1s in a row more often than pure-random. But how to be sure you don't get thousands of the same bit in a row (which would be silence)? It's a puzzle my uncle might love. He'd point out that his brain-time is worth more than many alternate hiss sources.

Trying to specify "good snap crackle and pop" is hard. But I know good hiss when I hear it. If I can hear it, I can record it. And if I don't have the exact-right hiss, a digital audio editor can paste and mix and mangle to taste. A layer of plain hiss, a dollop of off-station AM radio, sprinkled with the lead-in groove from the copy of Live Dead which survived The Party Of 1975.

Which would be absurd in days of tape and razors, but is now just a bit more expensive than a PIC. (And can be loaded with stone-age minds.) Here's a fancy kit to stash 2 minutes of audio and repeat for 100 years. Here's another with 6 tracks. I know they come VERY much smaller and cheaper, toy-level.

[PRR]

Futurlec has a list of recorder chips:
http://www.futurlec.com/ICSFOthers.shtml
"Special Function - Voice Record/Playback"

$6-$12 for the chip. $6 includes a Meg of RAM. High-cut at 3.4KHz to 1.7KHz, trade-off with time. A full-range hissbox would need more. A telco FX box would be fine at 3KHz or probably less.

However the in-stock chips seem to be awful fancy. Some need a CPU to get going at all. We just want a RECord button, and auto-play otherwise. So it may come down to dissecting stuffed bears.

Photo-frames with recordable audio:
Circle of Friends ... space for 8 photos and 8 separate voice messages! ...each 7-8 seconds long $30
Dad Greeting Card Voice Frame, 10-second message $25
Recordable Postcard - Clearance Price: $3.97:
...high quality, 10-second recordable voice chip
To record a message, click switch to 'ON'
Hold down the 'REC' button while recording
When finished, click switch to 'OFF' to keep your message saved
Push 'PLAY' to play it back

[chi_boy]

Well, you guys have blasted way past my abilities.  There are some very nice ideas here, but for me at least, they are for the most part impossible to pursue.  On the analog side though, I did make some personal progress.

I played a bit with the transistors and did find a difference in the 2N5088's I had on hand.  The Central Semi's required an R21(22k) of 220k to work and didn't work with a 100k.  I really should have tried some between values, but did not.  I switched in some Fairchild 2N5088's and R21 was functional at 68k.

I did play around with swapping in a couple of different transistors and I did hear at least a minor difference for one of them.  The others were all the same.  The only one that was a little different was the 2N3904 (R21= 68k).  It seemed to have a low rumble that the others lacked.  It sounded almost like two different tracks on top of each other.  The white noise track (hiss) was the same as the others, but the additional track was low and rumbly, (1/F noise?).  Pretty much like this:

http://www.ciphersbyritter.com/NOISE/ZENER1.WAV

Note:  This is NOT my recording. 

The final noise section I am settling in on is the 2N3904 transistors with R21 at 68k.  I also changed C9 to 100uF as suggested.  I think this may have allowed some more rumble, but it is a little subjective.

On the filter side I swapped out R10 and R11 from 56k to 15k.  From what I calculated this raised the lower frequency 300Hz to 1.1kHz.  The result is a great LoFi A.M. radio sound, complete with static.  I fully plan to box this one up.

More "character" in the noise section would be fun, but I don't really think it's a necessity.  Although I still find myself obsessing over it.  I can't explain why.

If it was within my capability I might change a thing or two about the circuit now that I have played with it a bit.

1)   I really can't see an application for the noise only option in the panner.  When fully in that mode, no guitar signal passes though.  Personally I would have the "noise" as an adjustable addition to the constant guitar volume and eliminate the panning all together.
2)   Since I am still obsessing over noise, the ability to add in a second, different, noise source would be interesting.  The user would then be blending 3 sources for the final product.  Of course I haven't found that noise source yet, so this may be moot.
3)   Someone suggested adding the noise before the filter, but I can't find the post to give credit.  In listening to this circuit, I think that would be an enhancement, but would need to hear it first.  A switchable injection point would be the ultimate tweak.
4)   Adjustable filter settings would be interesting.  I did try a layout to use rotary switches to change the filter resistors.  When I breadboarded it, it worked fine, and I don't doubt it would work in reality.  I just ran out of talent when trying to do the layout so gave up on it.

Of course this all adds complexity that wasn't part of the original design criteria, and is in no way intended as criticism.  It's more like [my] talentless enthusiasm for the project.

Overall I am surprised this schematic has been around for so long and has not been picked up.  There are no layouts to be found anywhere.  It is significantly more complicated than a LoFoMoFo, but for those that don't care for that design, this one may be a good solution.

R.G. and all, thanks for the help and suggestions.  And thanks for a fun little circuit!

Cheers,
George

[R.G.]

> PIC 10Fxxx .... A 31-bit feedback shift register repeats in 2.147 billion steps
Cool! Wudda liked that back in the day. And you can't beat the parts-count/cost.

I was pretty sure I wasn't the first to think of this. Check this link from EDN, 2003:
http://www.edn.com/article/CA313057.html
Note that this is an attempt at a purist's noise generator. Any one of the 12C508s will do noise suitable for audio all by itself, and drive a pinking filter directly as well. The parallel adding ensures better frequency distribution to get a true-Gaussian signal. I doubt I can hear that well, nor can most others. And he got the same number of states for 31 bits I did! 

But George said "nice smooth white noise ...  Maybe more snap crackle and pop?"

Yeah, that takes finding a device that has snap/crackle/pop breakdown. It's much simpler to generate randomly distributed crackle noise.

To do this, one takes a white noise signal, low pass filters it to get the degree of density one likes, then uses the noise as an input to a comparator, the other input being a variable reference level. The output of the comparator can be capacitor coupled to make impulses, or fed into the trigger of a one-shot, and the level of the impulses out used as the snap/crackle/pop signal. It can be added back to the white noise, or even gate the white noise. But all this gets outside the simple circuit realm.

[marmora]

Finally finished building this circuit.
I used chi_boy's first layout (I had a PCB made before he updated it):
http://www.diystompboxes.com/smfforum/index.php?topic=73881.0

Long story short: Q2 needs to be flipped.  I should have realized this if I carefully read the schematic and looked at the transistor's data sheet.
I wound up settling on a 100k resistor to get the noise section going.

I did these mods:
http://www.diystompboxes.com/smfforum/index.php?topic=25185.msg164559#msg164559

http://www.diystompboxes.com/smfforum/index.php?topic=26214.0

Impressions:
Cool circuit, though the white noise can get grating after a while.
I have to agree with chi_boy, some "character" to the noise would be beneficial, as would a filter for the noise.
The adjustable filter for the guitar signal is excellent though.  Thanks to Mark for the suggestion.  There is some distortion depending on your settings and attack.  It sounds good though.
I tried some other things through this circuit and realized this is a good candidate for "dirtying" up your sound.  Not as obvious as distortion for example, but noticeable.
The other fun/useful thing to use this for is contrast.  Play a lick/riff/arpeggio with the effect off, switch it on and play it again, repeat.  Interesting for a quick "color" change.

Thanks again to all involved!

[chi_boy]

That's kinda funny.  I've been working on the same project.  I did Marks mods too and also changed the panner to a mixer. 

For the noise section I made a white noise generator based on this: http://www.musicfromouterspace.com/analogsynth_new/NOISECORNREV01/schematic_pg1.pdf, but converted to 9v.  It has smooth white to frying eggs.  I like it.  I also added whine based on the synth stick, but with 3 oscillators.  I also have it set to be able to switch the noise from before or after the filter of the main board.

Needless to say, the scope has crept just a little beyond R.G.'s original vision.  But it is  a hobby, not a business, so what the heck. 

Mine is in the process of being etched and drilled.  This part is new for me so I'm climbing up that learning curve, along with learning Eagle.

I'll post my layouts once I verify it all works.  For now it's all spread out on 3 breadboards.

Cheers,
George