Hexaphonic Pickup Project

[RedHouse]

...Jude is the patron saint of lost causes...

My kind'a saint!

[RedHouse]

When you debug and finalize your design, you should out the SMT parts on the bottom side of the bottom piece of flatwork, then only have a 3-wire interface to the pickup.
(power, gnd, signal-out)

Nice work BTW.

[cycfi]

Ralph Jude Barthine
(No connection to the Beatles song; Jude is the patron saint of lost causes. )

Done. Thank you, Jude.

[cycfi]

When you debug and finalize your design, you should out the SMT parts on the bottom side of the bottom piece of flatwork, then only have a 3-wire interface to the pickup.
(power, gnd, signal-out)

Nice work BTW.

"bottom side of the bottom piece of flatwork": I am not sure I am parsing that correctly. Could you please explain a bit more?

The Neo PU (the small piece with one or two coils) has the SMT at the bottom already. The Neo interface has 4 wires however: v+, v-, vref and out. This is so you can have single supply operation, dual supply operation, or pseudo-differential single supply (as it is used by the hex PU).

Thanks!

[rjb]

Done. Thank you, Jude.

Um, I don't usually use my middle name.
My screen name is rjb because rb was either taken or "not enough characters" (don't remember which).

[salvarsan]

I just remembered that EMG used the LF441 in their pickup designs as late as 2003. Now obsolete, it had modest specs, still better than a 741.
For comparison, the 741 had 0.3 MHz bandwidth, 0.435 V/us slew, and a 2 ma idle current.

Of the LF441, Ti.com sez:

• 1/10 supply current of a LM741 200 µA (max)
• High gain bandwidth 1 MHz
• High slew rate 1 V/ms
• Low noise voltage for low power 35 nV/SHz
• Low input noise current 0.01 pA/SHz
• High input impedance 10 E+12
• High gain (VO =+/-10V, RL=10k) 50k (min)

If you go over to the TI.com low-power op-amp page, the choices fall out easily.
If the LF441 was an indication of tolerable performance, you need a minimum 1 V/uS slew rate and 1 MHz gain-bandwidth produt (GBP).

[cycfi]

I just remembered that EMG used the LF441 in their pickup designs as late as 2003. Now obsolete, it had modest specs, still better than a 741.
For comparison, the 741 had 0.3 MHz bandwidth, 0.435 V/us slew, and a 2 ma idle current.

Of the LF441, Ti.com sez:

• 1/10 supply current of a LM741 200 µA (max)
• High gain bandwidth 1 MHz
• High slew rate 1 V/ms
• Low noise voltage for low power 35 nV/SHz
• Low input noise current 0.01 pA/SHz
• High input impedance 10 E+12
• High gain (VO =+/-10V, RL=10k) 50k (min)

If you go over to the TI.com low-power op-amp page, the choices fall out easily.
If the LF441 was an indication of tolerable performance, you need a minimum 1 V/uS slew rate and 1 MHz gain-bandwidth produt (GBP).

Looks good, I remember using that chip. The question is: what are they using now? Anyone?

FWIW, modern low power Op Amps are a lot better than the LF441. My current low-power favorite is ths OPA314: http://www.ti.com/lit/ds/symlink/opa314.pdf

[cycfi]

Um, I don't usually use my middle name.
My screen name is rjb because rb was either taken or "not enough characters" (don't remember which).

Ooops. Sorry, Ralph

[cycfi]

Cross-talk

One thing I need to know is how much crosstalk can be tolerated (in terms of db)? Early experiments show good separation as-is.

May I ask that question again? Anyone?

Based on some scope measurements, I get -38 db cross-talk. I measure the E and A string pickups while hitting the E string. I chose the E string because it has the heftiest magnetic mass and should be picked up easily by the adjacent (A) pickup compared to the other strings. I got -38 db from averaging 20 takes. I'm thinking that -38 db is already very good. I read in this thread that Mike Sulzer gets something like -20 db, so it seems I've surpassed that. Scope shot below:



Thoughts?

[RedHouse]

One thing I need to know is how much crosstalk can be tolerated (in terms of db)?

Wouldn't the answer to that depend on your design criteria?

[cycfi]

Wouldn't the answer to that depend on your design criteria?

Ok, it seems the context was lost, so let me bring it back. Re-quoting RJB:

I realize you intend to neutralize string crosstalk with DSP- but for hacks like me who might want to use the pickup to do nothing more than send the signals from the lower strings to an octave doubler box, I fear the pickup's inherent crosstalk may be too high.

In this post from a previous thread, it is suggested that a humbucking pair of coils be used per string- with the distance between the poles of each HB pair being "significantly closer" than the string spacing.
http://music-electronics-forum.com/t26380/#post226819

The Sixpack 1.0 uses 5mm (~0.2") Diameter magnets, with the poles spaced to match string spacing.
I'm guessing that HB pairs with ~2.5mm D magnets and coils spaced as close together as possible would be less prone to crosstalk.

In this post, a member describes a setup for triggering an octave doubler with diy low-impedance pickups, using .125"D neo magnets.
http://music-electronics-forum.com/t7149/#post56850

Back to the drawing board again?

So, IOTW, for analog processing (no DSP voodoo), is -20db good enough as Mike Sulzer noted he's getting (in the other related thread)? I am getting -38 db, so should I put the cross-talk issue to rest?

[RedHouse]

Yes I understand the context of the discussion.

There is an old axiom when it comes to "testing" as a measurement of success, and the montra is "we test to the spec".
(as your webpage says you're a "programmer", I'm sure you've encountered this in a dev cycle)

So what is your design criteria for x-talk? as you're working away here to develop something (obviously) are you saying that you don't have any design criteria to go by (or meet, or exceed).

[Steve Conner]

Given that a regular pickup has 0dB of crosstalk isolation, I think anything above 20dB would be a worthwhile improvement.

Of course in a hex fuzz context the required crosstalk performance would depend on the amount of distortion used. If you stick 20dB of gain on it and then clip it, there goes your 20dB of isolation, because a note 20dB down on the wanted one will still get through.

[cycfi]

Yes I understand the context of the discussion.

There is an old axiom when it comes to "testing" as a measurement of success, and the montra is "we test to the spec".
(as your webpage says you're a "programmer", I'm sure you've encountered this in a dev cycle)

So what is your design criteria for x-talk? as you're working away here to develop something (obviously) are you saying that you don't have any design criteria to go by (or meet, or exceed).

Well, to be honest, I'm not that concerned with cross-talk. I am confident that it can be minimized using DSP (It's not straightforward. Notice the lack of obvious correlation between the direct signal and cross-talk in the scope shot, but I believe it can be done). That being said, I also think that RJB has a very valid use-case. And for that, yes, I do not have a design criteria to go by.

[rjb]

So what is your design criteria for x-talk?

My own hypothetical design criteria is that, while comping, I can send the walking bass line on the lower strings to my imaginary octave doubler (the one with zero latency and output that sounds exactly like a plucked string bass) without the chords on the higher strings accidentally triggering the device. Each "bass" pickup signal may (or may not) pass through an individual LPF to reduce overtones before going to the octave doubler.

Considering I'll probably never get around to this project anyway, I'd say the -38dB crosstalk spec is more than adequate.

EDIT:
Forgot to mention, this would be on a flat-top acoustic guitar.