Hexaphonic Pickup Project

[Cycfi]

Of course, if your goal is recreating a new kind of high fidelity DI-ed sound, experimenting with lower samplerates may not be useful.

That's more in line with what I am thinking of. I wouldn't even think of (nor afford :-) ), six amplifiers and effects chains. It seems obvious to me early on that DI and digital signal processing is the way to go with hex processing in mind. There's no other practical way. Of course it is still possible to route the mono-mix back a guitar amp, but hmmm...

[curtisa]

OK, I understand better what your game plan is now.

Well, if what you're proposing is something far more radical and unique than simply a hexaphonic pickup driving independant analogue signal paths, why limit yourself to "standard" guitar components? No need to constrain yourself to 1/4" plugs and sockets - how about ADAT lightpipe over opticfibre? Dante over ethernet? Raw serial bitstream from your onboard ADC via multicore cable?

Whatever system you decide to employ will also dictate how the data can be transmitted from A to B. As interesting as wireless sounds, I'd personally prefer to use something more mechanically robust, especially over a "public" radio method such as Bluetooth. Dante sounds promising as the cables can be picked up just about anywhere, even rugged-ised versions aren't that hard to find. Serial bitstream direct from the ADC is another option, using whatever connectors fit the bill (multi core XLR, Sub-D) - the limiting thing here will be trying to sucessfully transmit the data through potentially long lengths of cable, something that proprietry systems are designed to avoid Maybe three channel pairs of SPDIF over multicore is a possibilty here?

Using a cable also allows power to be delivered from the host to the guitar. Wireless dictates the use of batteries. Installing so much processing devices into the instrument that requires multiple sets of batteries, or short run times before recharging will make the instrument more annoying than inspiring to end users.

[Cycfi]

OK, I understand better what your game plan is now.

Well, if what you're proposing is something far more radical and unique than simply a hexaphonic pickup driving independant analogue signal paths, why limit yourself to "standard" guitar components? No need to constrain yourself to 1/4" plugs and sockets - how about ADAT lightpipe over opticfibre? Dante over ethernet? Raw serial bitstream from your onboard ADC via multicore cable?

Yes! Yes and yes! That's what I'm talking about :-) No 1/4 plugs and sockets here, for sure. At any rate, some folks might find the just the pickup usable as-is.

Whatever system you decide to employ will also dictate how the data can be transmitted from A to B. As interesting as wireless sounds, I'd personally prefer to use something more mechanically robust, especially over a "public" radio method such as Bluetooth. Dante sounds promising as the cables can be picked up just about anywhere, even rugged-ised versions aren't that hard to find. Serial bitstream direct from the ADC is another option, using whatever connectors fit the bill (multi core XLR, Sub-D) - the limiting thing here will be trying to sucessfully transmit the data through potentially long lengths of cable, something that proprietry systems are designed to avoid Maybe three channel pairs of SPDIF over multicore is a possibilty here?

I'm inclined to agree! In as much as wireless is a sexy proposition, and I'm often lured into looking into the current state of the art every once in a while :-), I think that a wired connection is still the way to go. A means for wireless transmission should be just an option.

I'll look more into Dante. Serial bitstream is cool! A bit related to that, lately I'm interested with this idea of using class D amplifiers for sending differential analog: Inexpensive analog isolation using a digital isolator.

That said, the easiest solution I can see right now is to simply send through USB. That means ADCs on-board plus a microcontroller at the minimum. USB can provide the power just fine. USB has sufficient data rate and is getting better all the time. Plus, you can connect to any computer without additional hardware.

Using a cable also allows power to be delivered from the host to the guitar. Wireless dictates the use of batteries. Installing so much processing devices into the instrument that requires multiple sets of batteries, or short run times before recharging will make the instrument more annoying than inspiring to end users.

Nod. Agreed!

[Prostheta]

The most wonderful thing about this is that being open source it will only attract more and more collaboration! I'd love to see this snowball.

[curtisa]

That said, the easiest solution I can see right now is to simply send through USB. That means ADCs on-board plus a microcontroller at the minimum. USB can provide the power just fine. USB has sufficient data rate and is getting better all the time. Plus, you can connect to any computer without additional hardware.

USB was going to be my other suggestion, although it would limit you to only being able to use the instrument with a PC/Mac. Then again, maybe this shouldn't be considered a limiting factor?

Have you seen miniDSP products? Check these out, only just found them today, looks really promising for your application. I could imagine pairing one them up with some kind of multichannel ADC and straight away solving the hardest part of the signal chain - getting the signal into the host. Not too expensive either.

[curtisa]

The most wonderful thing about this is that being open source it will only attract more and more collaboration! I'd love to see this snowball.

Indeed. I wish my DSP knowledge were advanced enough to participate in this more. While I have built a 24bit 96KHz ADC in the past, it was really nothing more than joining a bunch of IC datasheets/application notes together, and it has been a long time since I've done any serious electronics work at this level.

[Cycfi]

That said, the easiest solution I can see right now is to simply send through USB. That means ADCs on-board plus a microcontroller at the minimum. USB can provide the power just fine. USB has sufficient data rate and is getting better all the time. Plus, you can connect to any computer without additional hardware.

USB was going to be my other suggestion, although it would limit you to only being able to use the instrument with a PC/Mac. Then again, maybe this shouldn't be considered a limiting factor?

No, I don't think so. It is easy to build a USB board with DACs for each channel for those who want to get the raw (unprocessed) audio. Another idea is to use the Raspberry PI, as a stand-alone audio processor. Its 700 MHz core is pretty capable and with Linux as the underying OS makes it quite easy to program with. Also, I have a hunch that its GPU can be repurposed as a DSP processor (I'm pretty sure you've heard about GPU computing).

Have you seen miniDSP products? Check these out, only just found them today, looks really promising for your application. I could imagine pairing one them up with some kind of multichannel ADC and straight away solving the hardest part of the signal chain - getting the signal into the host. Not too expensive either.

Oh, that is pretty cool! Thanks for the link! I've been keeping a keen eye on XMOS. They're pretty much focused on audio and a lot of their clients are actually audio related (e.g. Sennheiser).

Anyway, at the moment, I am looking at the lower-end. I've got this new stuff from digikey which includes the C5000, ultra-low-power DSP Booster Pack from TI. Its TMS320C5535 fixed-point DSP chip with two multiply-accumulate (MAC) units capable of up to 200 Million Multiply-Accumulates per Second (MMACS) and a tightly-Coupled FFT Hardware Accelerator. At $10 a piece (per 100 at Digikey), these DSP elves are very capable. There's a lot you can do at that price point already.

Edited April 9, 2013 by Cycfi

[Cycfi]

The most wonderful thing about this is that being open source it will only attract more and more collaboration! I'd love to see this snowball.

Indeed. I wish my DSP knowledge were advanced enough to participate in this more. While I have built a 24bit 96KHz ADC in the past, it was really nothing more than joining a bunch of IC datasheets/application notes together, and it has been a long time since I've done any serious electronics work at this level.

I'd really want everyone to be able to hack on this. My goal is to make everything modular such that one can simply connect things like they do with stomp boxes. And, for the bare-metal hackers out there, there's the source code and schematics to play with.

PS> There's a parallel discussion about the Hex-pickup project at the Seven-string.org forum. I'll be cross posting to both forums. I'm pretty sure cross-pollination would be very welcome. I have the same reply to something similar there :-)

[Cycfi]

Here are some (mp3) wave samples of the initial prototype of the pickup along with some FFT graphs of the harmonic spectrum taken from the uncompressed, raw (aiff) samples. Stratocaster and Les Paul samples are also provided for comparison.

[Prostheta]

Sounds good. Quite a hi-fi sounding pickup which is always a good basis for subsequent processing. Would probably sound quite good on a longer scale also to achieve more of the low order harmonics associated with pianos.

[Cycfi]

Sounds good. Quite a hi-fi sounding pickup which is always a good basis for subsequent processing. Would probably sound quite good on a longer scale also to achieve more of the low order harmonics associated with pianos.

Thanks! I'm quite pleased in fact. I'll try 1000 turns or lower next for comparison.

[Prostheta]

Two instruments spring to mind when it comes to being active and using very low turn counts/DCR. The Gibson RD Artist incorporated a large and perhaps overly complicated Moog circuit which was quite gainy on top of the low output pickups. The other is the Aria Pro II SB-Integra which was a basic PJ bass with a gainy pre-amp (~x10 if I recall). Both fine sounding instruments with much clarity. There will be many others, however these two are making me think more and more about the idea of low turn count pickups.

I doubt you will need to consider winding hotter however definitely do it. For science.

[Cycfi]

Two instruments spring to mind when it comes to being active and using very low turn counts/DCR. The Gibson RD Artist incorporated a large and perhaps overly complicated Moog circuit which was quite gainy on top of the low output pickups. The other is the Aria Pro II SB-Integra which was a basic PJ bass with a gainy pre-amp (~x10 if I recall). Both fine sounding instruments with much clarity. There will be many others, however these two are making me think more and more about the idea of low turn count pickups.

I doubt you will need to consider winding hotter however definitely do it. For science.

Not sure if you know this, but Les Paul actually favored low impedance pickups that are DI'd directly to the mixing board. He favored the wider frequency response and the possibility of sending the signal through long cable lengths (in a differential manner of course): The orphaned (and very rare) Les Paul Personal, Professional and Recording models use passive low impedance pickups. They didn't catch on at the time because the outputs were very low and couldn't drive guitar amplifiers.

Edited April 14, 2013 by Cycfi

[Prostheta]

I did actually. I'm a fount of obscure information :-)

Les was a much underestimated inventor. Most people think he stopped inventing after the solidbody guitar named after him. Certainly not!

[Cycfi]

I did actually. I'm a fount of obscure information :-)

Oh of course you are :-)

Les was a much underestimated inventor. Most people think he stopped inventing after the solidbody guitar named after him. Certainly not!

Indeed. Anyway, obviously I am a fan of low impedance PUs. BTW, there's this guy at the music-electronics-forum who tinkered with extremely low impedance pickups with "proprietary" active circuitry. He proclaims that "nothing else sounds quite like a 10 ohm humbucker". 10 ohms! Now that got me thinking...

[Prostheta]

10Ohm? Crazy. Nothing short of an EMP would spike noise onto that!

[Cycfi]

10Ohm? Crazy. Nothing short of an EMP would spike noise onto that!

How low is the impedance of the Les Paul Recording anyway? I read here (from the comments): Les Paul’s Favorite Les Paul – The Recording Model that it's a stacked humbucker design with 5 Ohms per coil, for a total of 10 Ohms! Can anyone here verify that?

Edited April 15, 2013 by Cycfi

[Cycfi]

Let it breathe (part 2): More Frequency Analysis

Last time we did some (very rough) preliminary tests comparing an early version of the hex pickup with a Fender Stratocaster and a Gibson Les Paul. The initial test was done using a 2000 turns AWG 42 coil and the cheapo TL072 Op Amp in differential configuration. The goal basically was to see the frequency spectrum of lower impedance coils side by side with that of the single-coil Strat and Les Paul humbucker pickups. Here we will continue with the tests, this time focusing only on the Hex pickup installed on a Strat while continually improving on various aspects of the pickup design.

Tell me what you think.

[Leo Pedersen]

Wow that's really interesting.

This sample sounds much more mellow than the previous hex on strat one. Much more humbucker-like to me. More bass and less sparkle.

It was really interesting to open the previous samples in another browser window and listen to them back to back.

This one is by far my favorite so far (ignoring the Bosendorfer of course, haha.) I really like the tone.

It would be interesting to hear this new coil design with the previous amplification circuit just to hear what a difference the coil makes without changing the circuit. Is your intention to leave the previous circuit behind because the new one is better?

I'm certainly liking the sound and I'm very excited to hear your subsequent tests with the 2000 turn coils.

[Cycfi]

Wow that's really interesting.

This sample sounds much more mellow than the previous hex on strat one. Much more humbucker-like to me. More bass and less sparkle.

It was really interesting to open the previous samples in another browser window and listen to them back to back.

This one is by far my favorite so far (ignoring the Bosendorfer of course, haha.) I really like the tone.

It would be interesting to hear this new coil design with the previous amplification circuit just to hear what a difference the coil makes without changing the circuit. Is your intention to leave the previous circuit behind because the new one is better?

I'm certainly liking the sound and I'm very excited to hear your subsequent tests with the 2000 turn coils.

I am currently working on the tests for 500 turns, 1000 turns, 2000 turns and 3000 turns so we can see the freq. response side by side. I also rigged an (core-less) 8 Ohm driver-coil placed a few millimeters on top of the pickup-coil which is then driven by a 20Hz to 20kHz sine sweep while monitoring the preamp's output response. This will give us an accurate profile of the actual frequency response curve of the coil + preamp.

I like the sound of this one too. I can probably re-test the older circuit, but I doubt if it contributed to the 'color'. The changes in the electronics were meant to give improved CMRR (by using matched/precision resistors and better Op Amp) as well as lower the noise floor (lower noise Op Amp and lower valued resistors). I believe the pickup design with a more focused magnetic field and lower string pull (thus closer proximity to the strings) contributed a lot more to the color than anything else. If there's such a thing as proximity effect on pickups, I guess this is it (it would be easy and interesting to test that conjecture).

Thanks for your input! I really appreciate it. In the end, we can probably have a set of samples to choose from and have a poll which one you guys think is the best.

Edited April 29, 2013 by Cycfi

[Cycfi]

So... we are currently using single-coils in floating differential mode. The coils are connected to the + and - inputs of the differential amplifier without any direct path to ground. This configuration is inherently noise canceling. Should we be using double coils in humbucking configuration instead to get even better noise cancellation?

Here, we talk about the limitations of passive humbuckers and propose possible alternatives such as the Bifilar Coil winding. Read the rest here: Singles or Doubles?

[Prostheta]

I've always liked the idea of a bifilar-wound pickup or one with several "taps" throughout the wind which can be summed later in the signal chain. I presume this wouldn't be a million miles away from how a stacked pickup works, but in a more compact form? Would there be appreciable crosstalk or other interaction between the two coils due to their extreme proximity in a near-identical wind?

Excellent proof of noise cancellation. The exact same method is used in the classic 70s Aria Pro II SB-1000s and they were near silent. I have upgraded many many of them the last couple of years with pre-amps using metal film resistors and modern ICs like OPA2277s, OPA2107s, etc. and this makes the differential amplifier deadly silent. Almost to the point of it being a joke!

Is the rumble in the low frequencies just an artifact or actual noise? Truly fascinating results. You've killed the noise already....now let's see what games can be played with this thing!

[Cycfi]

I've always liked the idea of a bifilar-wound pickup or one with several "taps" throughout the wind which can be summed later in the signal chain. I presume this wouldn't be a million miles away from how a stacked pickup works, but in a more compact form? Would there be appreciable crosstalk or other interaction between the two coils due to their extreme proximity in a near-identical wind?

I honestly don't know. I haven't tried this before and to be honest, electromagnetism is by and large still a mystery to me. I am learning each day. People who did it sing praises. For example:

"Well, I got around to hooking it tonight (using a two stage differential amp made up out of 3 x TL072s) - even with very old (dead) strings on my test guitar it's sounds very 'full bandwith'. And as Rick has just pointed out ...there's something about the low end through a distorted amp sim - it struck me as being less muddy vs a high impedance pickup.

I'm quite impressed. But what has really knocked me out is the lack of noise (that'll be the common mode rejection going on -my first real experience with it ...& I'm sold!)."

Excellent proof of noise cancellation. The exact same method is used in the classic 70s Aria Pro II SB-1000s and they were near silent. I have upgraded many many of them the last couple of years with pre-amps using metal film resistors and modern ICs like OPA2277s, OPA2107s, etc. and this makes the differential amplifier deadly silent. Almost to the point of it being a joke!

Is the rumble in the low frequencies just an artifact or actual noise? Truly fascinating results. You've killed the noise already....now let's see what games can be played with this thing!

The noise level is so low I can't hear anything even with my headphones set to max. I reckon I can hear it if I boost the signal with ample gain, but I haven't done that. I'll try when I get the chance.

[Prostheta]

You're making Faraday cages look like stones tied onto sticks with horse hair. Good show old chap!

As soon as some kind of distortion is added to the output of the pickup we have a Litmus test for the pickup taste. Looking forward to the next installment.

[DarkAvenger]

I've always liked the idea of a bifilar-wound pickup or one with several "taps" throughout the wind which can be summed later in the signal chain. I presume this wouldn't be a million miles away from how a stacked pickup works, but in a more compact form? Would there be appreciable crosstalk or other interaction between the two coils due to their extreme proximity in a near-identical wind?

Unfortunately it is a million miles away. Also to clarify, there are multiple ways of wiring a bifilar coil and one is relevant here, 'bifilar series constructive' as I call it. Anyway, a stacked coil is basically adding a dummy coil to the bottom of a single coil to cancel EM noise. The dummy coil is magnetically isolated, to a degree, therefor when wound backwards relative to the coil above it does not cancel out the magnetic signal from the string's movement.