Audio Effects DSP Board

[racerg]

Doubt he ever will...not really an open source project...seems like it's taking a loooong time...

[markseel]

Yeah it's taking a long time!  What started as a hobby and an already busy life has turned into full product development ... and a busy life :-)

The hardware is a simple design and is straight from the data sheet.  But you're right this is not open source yet.  This has become quite a passion and something I'd like to work on full time. I'm not looking to get rich but I'd sure like to be able to make a living doing this. So there is a hardware cost for now.

If you consider the complexity and feature set of the usb audio and midi support, the completeness of hand optimized DSP, the text and examples and html interface support, and a carefully vetted design to enable end user algorithm development then I really believe that there's a ton of value here.

With high enough volume these boards are sub $50 (maybe $40?) and full aluminum pedals are sub $250 (approaching $200?). Look at the USB and MIDI features, size, flexibility, cost, finishing, and performance figures and compare to OWL, ToneCore, Spin, Teensy.  Check those other platforms for audio quality and effects support - effects with studio quality digital signal paths. FlexFX are high quality and free and are implemented with attention to DSP design rigor.

The last couple of months has been mostly kick starter campaign building and flexfx bug fixing. They've both come a long way. Today is the final push to wrap up kick starter media and to submit it for review by kick starter staff.  Many prototypes for hardware, boards, and effects have been made to prove out the flexfx concept and design. Contract manufacturers are in place for all aspects of production.  So close!

Thanks to all who have supported this by reading my posts, asking questions, and making suggestions. And thanks for being patient :-)

[EBK]

Even if you did publish every detail, I definitely wouldn't want to build one myself anyway (too much work), but I'd be happy to buy one someday. For now, I'll keep reading and cheering you on. 
I wish you great success with this!

[markseel]

Here's the KickStarter preview link:

https://www.kickstarter.com/projects/632112529/68367638?ref=499075&token=26791363

Feedback/comments welcome :-)

[markseel]

Sigh ... KickStarter rejected my campaign since it doesn't have a video.  I received some good feedback from some other folks (thanks!) also suggesting a video.  So I have to do that next. 

And I'll add more prototype photos (especially the internals to show that his is real :-).

So I'm going to shoot some video of some locals here in Minneapolis playing through some prototype units.  I'd like to get clips of these being used and then stitch the clips together.

If anyone here is up to shooting a short video and then receiving a FlexFX unit for free let me know.  If the video can look like a quick pedal review that'd be awesome.

I can ship you a prototype unit (as seen in the KickStarter campaign photos).  The prototypes are fully functional.  Hmmm, or you could keep the prototype rather than receive a production unit later.

Anyone interested?

[EBK]

I'm still in awe at how nice the prototypes look. 
I would love to help out, but I still haven't learned how to film a quality demo video for my own hobbyist stuff.  (Maybe I should start a thread asking for tips on that....)

[markseel]

Thanks EBK :-)  The production units will look better with the final finishing and tighter tolerances.

I don't want to imply that this has to be a professional video.  Just trying to get some footage.

[EBK]

Just out of curiosity, if you don't mind sharing, how much does it cost to produce the precision-machined, blue anodized and laser-engraved aluminum case?

[markseel]

I get them machined for $38, anodized for about $5, and laser engraved for about $5, all in quantities of 100.  I'll try to get the total down to around $40/case all in with larger volumes.  But even still it's sort of a ridiculous cost for a pedal that's sub $250 (closer to $200).  It doesn't leave me much room for profit.  I like how the knobs, USB and power jacks end up being custom fitted into the case.  A bit risky though since any part change results in a minor CAD change for the case.  But I'm really kind of stubborn on having them done this way rather than using the conventional powder coated and/or painted cast aluminum (e.g. Hammond) cases.  My drive is to create a lasting piece of hardware and that's why it has to be flexible (programmable) - so that if someone wants a different effect then just reprogram - and that way the expensive case is still usable/valuable.  I'd rather throw away software than hardware :-).  If you want to start your own line of pedals I could make the cases brand-free (no FlexFX or knob labeling) and supply an OEM version that has the electronics and case ready to go for you to customize with software and anodizing/engraving/painting/adhesive_labels, etc.

[markseel]

OK sort of hypothetical question here ...

What if I pivoted this pedal design from three knobs, USB/MIDI, and HTML control to a design with 6 or 8 knobs with USB/MIDI just there in case you want to use it for config/data loading or audio in/out (both can run off of 9V power jack - which is usually much lower noise and USB power).

The first design was predicated on the vision of three knobs for basic adjustments (volume, tone, preset selection) and preset configuration via USB/MIDI/Browser.

The second theoretical design would be the conventional approach ... lots of knobs for effect control but no USB/MIDI/HTML preset management since there's not really any presets.  You could still use USB/MIDI/HTML I suppose to configure or upload data or something but since there's a bunch of knobs the effect would look more conventional during typical use.

The first design puts emphasis on customization/tweaking via USB/MIDI/HTML while the second puts emphasis on conventional lots of knobs pedal usage.  If I pivoted there would be minimal delay to the project since the ADC for the pots is the same product family.  Would just have make minor case mods for the additional pots.

Feedback wanted!  What's the most interesting path for folks out there?

[audioartillery]

I am currently building an enclosure around BlackAddr's ARM-based effects board. I have a thread here with a mock-up of the design. It's big and overloaded with switches and knobs since it's intended for developer usage. 6 pots, 3 switches, 2 foot switches,, a small OLED panel, and a rotary encoder knob/switch. Might add a jack for an expression pedal too.

I think 4 pots is pretty useable. Line6's pedal has 6 pots and 2 3-position switches. I found that quite sufficient. 6 is better though, I've done effects that really needed 6 parameters.

[vigilante397]

I may be the exception rather than the rule, but I'm a simple man, I think 3 or 4 is great. Every pedal on my board has 4 (or less) controls, my amp has 4 knobs, all my songs have 4 chords...

[markseel]

Yeah it's not an easy question - how many knobs that is.  Simple is good.  Flexibility/control is good.  Hmm what to do?  If I had owned pedal that nailed a sound or objective (Tube Screamer) then a few knobs for tweaking works.  But if that pedal is meant to do a wide range of sounds or variations to allow for discovery/creation (ChaseBliss Thermae) then more knobs/switches/buttons makes sense.

Anyway I went with ... 6 knobs.  PCB should be back in a week for assembly and testing.

The 3-knob KickStarter FlexFX pedal will go forward as is (still need to get that video made).  I'll follow up with the six-knob version after that.

[markseel]

Back to the schematic topic.  The XMOS specific stuff is pretty much from the data sheet but as far as the analog stuff goes ...

The DAC and ADC used for my board is the AK4420 and the AK5386.  Here's the schematic for the ADC input circuitry.  There's two unity gain high impedance buffers.  These could be non-unity gain (G > 1) with a few simple changes.

This can be used as a high impedance stereo input or a very quiet pseudo-differential mono input.  Note that the 100n input decoupling cap is on input A but is missing on input B.

For the latter feed the guitar signal to one input (A) and the guitar ground to the other input (B).  Then in your DSP code subtract B from A to give you a low noise guitar signal (cancelling out the common-mode noise from RFI or dirty power supplies where RESULT = (A+Noise) - (B+Noise) = A-B ~= A.  I've measured the input noise floor for this arrangement to be near the ADC's noise floor of -110db.  This is nice if your effects has lots of gain :-)

The left red circle represents the guitar pickup.  The right red circle represents the ADC inputs.  The circuit is a 3rd order low-pass not counting the frequency response of the op-amp.  Phase is pretty flat out to 10 kHz.  Attenuation to prevent anti-aliasing is around 72dB at 6 MHz.  Note that the AK5386 sigma-delta ADC samples at 6.144 MHz.

The signal source VS2 is used to generate power supply noise (100 mV pk-pk) for testing.  The third graph shows the pseudo-diff (blue trace) vs the single-ended (red trace) output of the circuit.  The blue one is much quieter.  Noise couples into the guitar signal via op amp supplies (not much due to op amp supply noise rejection) and via the Vcc/2 bias for the op amp inputs.  If you're using a single guitar input be sure to use the pseudo-diff approach - it's a lot quieter (blue) vs the single ended (red).

[markseel]

The potentiometers for FlexFX are sensed using the MAX11600 series of 4/8 channel ADC's.  If you use these parts then you won't have to write your own ADC control code since FlexFX includes support for this device.

[audioartillery]

Mark, what's the status on this project?  Kickstarter going up soon?

[markseel]

Kickstarter rejected the project and I can't understand why. Just said it didn't fit with their philosophy or something like that. Maybe they didn't understand it?  Too bad since the project  is pretty much finished. I just need the cash to have boards assembled is a reasonable quantity. The effects in github work well - preamp overdrive, delay/chorus/flanger combo, 15 band graphic eq, reverb (not finished), and amp/cab (using IRs) simulation. The source code is available for all of them. I did some measurements of the analog/noise performance of a complete system (module plus ADC/DAC and circuitry) and it's impressive. Just need to find a way to get this out there :-). I'll look at the other crowd source options.

[audioartillery]

That's really strange. I wonder if that was just a mistake?

What kind of quantity did you need to get buyin for to get off the ground? You were targeting $225 each right?

[markseel]

Yeah for the pedal something like $200 to $225.  If I can drive the materials and PCB cost down then I want to cap it at $200.  Not sure if I can do that or not.  That KickStarter hasn't been submitted - I need to get the demo video finished for it.  I am using the FlexFX stuff inside this pedal.

The FlexFX module KickStart was rejected even through there's plenty of kit PCB's out there that have succeeded - so not sure what they didn't like about it.  The Kickstarter module cost was to be $50 but I want to drive that down a bit - closer to $35 or $40 depending on economy of scale.

I came up with some 'kit' boards that go with the module and with those perhaps the KickStarter will look better?  These are just small expansion boards for the module to give it I/O capabilities.  One board is a guitar interface - AK4621 CODEC and low noise opamp I/O buffers with input and output being differential/balanced for low noise.  Another board is a decent quality headphone amp with two PCM5102's (one for left, other for right) configured to drive AD8397 high current OPA's using a DC coupled and fully differential signal path between DAC's and OPA's.  Boards should go for $50 each but like the module I'd want to get that price down by scaling up quantity.

[markseel]

I appealed the campaign rejection and Kickstarter OK'd it. Going to start the campaign this weekend.