Femto-Verb--a DIY digital reverb pedal

[DaveTV]

I wanted to post this schematic while the circuit is still on the breadboard. I was hoping to get some advice on possible improvements/changes/obvious mistakes.

FemtoVerb Schematic

The Femto-Verb is a DIY 24-bit digital reverb pedal based around Wavefront Semiconductor's AL3201B reverb chip (Wavefront used to be Alesis Semiconductor, hence the AL prefix). The circuit also uses Wavefront's AL1101 A/D converter and the AL1201 D/A converter. Peter Snowberg has written quite a bit about these chips here on the message board and I thought I'd try building a simple reverb pedal around them.

The AL3201B is the same chip used in the Alesis Picoverb and includes 8 built-in hall, room, and plate reverb effects. The chip also has 8 other effects like chorus, flange, delay, and a pretty good rotary speaker. However, with the Femto-Verb's design, these other effects aren't tweakable and can only be used with their default settings.

The Femto-Verb's design is derived directly from the schematics shown in Wavefront's data sheets, so no points for originality here. The circuit is simplified for use with guitar and can easily be modified for stereo output (the AL3201B is intended for stereo use).

The chips are SOIC packages only, so they required being hand-soldered (tricky but not impossible). Using surfboards really helped, especially for prototyping with these chips.

I'd like to give a shout out to Peter Snowberg for giving me a lot of support with this project. He's been doing a lot of DSP development using the AL3201B and hopefully he'll go public with his designs soon. I for one would really like to check them out.

[Mark Hammer]

Holy Toledo, Dave, that is a whole lotta work on your part.  Indeed, far more than might be apparent from the simplicity of the drawn schematic (VLSI chips drawn as boxes sure make it look simple, don't they?).

My hat is tipped to you thricefold for:

a) putting in the time to get the technical info
b) taking the time to study and laster it enough to leverage it into design aspirations
c) whip up something that cold be feasible and usable by the folks here

Now, unceremoniously sweeping all the praise and tickertape aside, how does a mere mortal get hold of chips like these?

[Paul Marossy]

Can't offer any advice on the circuit, but those "surfboards" are cool!  

[Phorhas]

Applause.

Your work and those chips are inspiring... I already began to drift in thoughts...

Thank you Dave.


Dan.

[puretube]

Brother Hammer:
http://www.wavefrontsemi.com/
have fun!

[Marcos - Munky]

It's bad that I'm not ready for surface monted parts solder yet .

[jjs]

I have an Atmel Atmega16 and some Graphic LCDs lying around for quite some time. Initially I wanted to do a guitar/bass tuner, but that looks boring now compared to the new possibilities I see with this chip.  
That sure would make a cool multieffect.
Thank You for bringing me back to thinking (maybe only dreaming) about doing digital projects!

It also reminds me of the time I pulled the DFX card from my Marshall and hooked it between my guitar and the amp. Works quite well, by the way. Even thought about including this as digital effect in my High Octane build, but cancelled that.

How do the presets sound?

[Paul Marossy]

I wouldn't mind trying my hand at a digital effect just for kicks (just for the experience). I really don't know squat about digital, though...

[DaveTV]

Thanks for the kind words, everybody. This is the first time I've tried putting a circuit like this together, so please let me know if you see anything questionable about the different opamp stages.

Indeed, the chips only seem to be available through Wavefront. I phoned their sales department and told them I wanted to do some prototyping with their chips, and they were generous enough to offer me a couple free samples of each. The only downside was that it took them about a month and a half to put them in the mail, and this was only after I called them a couple times to politely remind them.

As far as soldering SOIC's go, I didn't think I was ready for it either. And the funny thing is, I was right. I totally cooked the first chip I attempted, but fortunately they sent me 2 so I had a backup.

The preset reverbs sound pretty good. The circuit is essentially a workalike of the Alesis Picoverb, so you might look there for sound samples. The other effects are okay, but I have other pedals for things like chorus and delay. I do however really like the rotary speaker sound.

[puretube]

yep - it`s a drag:
the more complex the integrated circuit,
the smaller the pack!

[Hal]

::spray paints::

DIGITAL SUX



hahahahahaha.   j/k.  awesome, Dave....but how does it _sound_ ?

and how does the Delay sound?  I'm kinda interested in a digital reverb/delay project....

[mikeb]

Very nice .... is it just me or is it amazing that such a neat device can be created with so few external parts? Obviously all the parts are designed to work together - the ADC and DAC are clocked from the DSP chip (pin 9) right?

I notice the ICs run from 5V whereas the opamps run from 9V - is there any problems with clipping of line-level signals?

Cheers

Mike

[Peter Snowberg]

Cool job Dave!      My hat is off to you!

A couple of notes....

The ADC inputs self-bias to 1/2 Vdd so if you want to silence the unused input channel, you should do it by connecting the inputs to ground via a couple of 10uF caps. If you want to be really complete, bypass those 10uF caps with a 0.1 or a 0.01uF film cap. Grounding the inputs with a DC ground will smoke the bias supply for the inputs.

The unused DAC outputs can be left floating. If you ground them, 1/2 of the output stage is going to overheat as it tries to sink power into the ground connections.


It's funny to note that your FemtoVerb is almost identical to the Alesis PicoVerb, except yours has more parts, is more polished, and is MUCH better suited to guitar. The Alesis PicoVerb is made for line level applications. (read: almost useless unless it follows another effect)


I actually like soldering SOIC packages because it's faster and easier than through-hole devices. The secret is to use a pre-tinned set of pads on the board and to coat them with liquid or paste flux before soldering. If you do that you can simply re-flow solder the part in place. I start by holding the chip down with a finger and touching the iron to pin 1. The next pin to go down is the one at the opposite end of the package (pin 11 on a 20 pin chip). With the first two down, all you have to do is to touch the iron to the rest of the pins to re-flow solder them into place. Wash the flux at the end and you're all set.


It's just too bad that you can't adjust the LFO speed for the built-in effects. I guess that's what programming is for.


On clocking... The DSP uses a 12.288 MHz clock to generate an internal bit clock, a left/right clock, and a word clock. To decrease EMI/RFI emmissions, the ADC and DAC use the DSP word clock and an internal PLL to re-generate the bit clock locally. That way the clock lines run at 1/64th the bit clock frequency.

[mikeb]

It's just too bad that you can't adjust the LFO speed for the built-in effects. I guess that's what programming is for.

Wouldn't it be possible to adjust the LFO frequency in a brute force manner by driving pin 3 with an external oscillator (it says this is possible in the datasheet)? I guess then the quality of the processed signal would also vary .... but if you are looking for a small 10:1 range of variation maybe this wouldn't matter (especially for guitar) ....??!!!

Mike

[DaveTV]

Hello Peter. I was hoping you'd chime in. For silencing the ADC stereo right inputs, would I need two 10uF capacitors connected to ground, or would 1 suffice? I'll update the schematic right away with these changes. Thanks so much.

I notice the ICs run from 5V whereas the opamps run from 9V - is there any problems with clipping of line-level signals?

I chose to run the opamps at 9V because I was worried that 5V was too close to the minimum operating voltage. I probably won't use this circuit for line-level signals, so I'm not too worried about clipping so long as I keep it near unity gain. I added the gain control to the input just in case I noticed a volume drop.

Also, I just noticed that the opamps are mislabeled. They're NE5532's. I'll change that as well.

[Peter Snowberg]

Dave, I'm guessing that the two inputs use the same bias generator but I don't know for sure. In any case I would use two caps. I would also suggest giving a try to feeding the other channel with the same input signals.

Mike, according to the datasheet if you had a VCO that varied between 6.144MHz and 12.288MHz, you could adjust between 1/2 and "normal" speed while dropping the sample rate to only 24KHz.

If you go below that you start having problems with the sample buffer because it's all built out of DRAM and requires instructions in the DSP to refresh it before it forgets.

Going higher in speed hits the rated ceiling VERY quickly. 12.5MHz is the fastest the data sheet shows support for.... how fast you can push it is up to testing. You might even have to boost the voltage a little to go much higher.

[DaveTV]

I would also suggest giving a try to feeding the other channel with the same input signals.

Hmmm. I guess I never thought of that.  :oops:

Might that be something I could try with the unused outputs on the DAC?

For the time being I disconnected those pins on the schematic and have just left them floating. I'll redraw them once I figure out what works best. Thanks again.

[Peter Snowberg]

I would suggest trying two DPDT switches in there..... One to switch the right inputs between ground and the left inputs, and one to select the left OR the right output.

You'll quickly hear that the reverbs take advantage of stereo to expand the effect.  

[DaveTV]

I think I'm getting the idea, Peter. Connecting the R and L outputs on the DAC together wouldn't be good because the signals aren't the same. But switching between one or the other would give two different flavors of the same effect--and perhaps one might even sound better than the other. I'll play around with this while the circuit's still on the breadboard.

In truth, I could very easily add the second stereo output to the circuit, I'd just need to add another NE5532. Maybe I'll consider doing this. My goal was to keep things simple, and there really haven't been too many situations where I've played using a stereo setup (if you don't count the Roland Jazz Chorus I used to have--why? ). Still, it would be nice to have the option. Would having stereo outputs require any special considerations? If I primarily used only one output, would the other one need to be tied down somehow?

This has been some great help.

[mikeb]

Peter:
Thanks for the info. I'd been thinking of quick-and-dirty ways to hack in some more functionality ... hmmmm ....

All:
When I contacted Wavefront Semi about getting some ICs to experiment with I received a very nice email from Randy Yorston, the operations manager; I'll be able to get a dev board of theirs with the three ICs on it for reasonable cost. Looking forward to playing with it! :twisted:

Mike

edit : here's a link to the schematic of the dev board, it might be of interest with reference to the circuit posted here....
http://prophecysound.com/temphost/Sch-W01-rev.pdf