TB11 Preamp Project

[MglMatador]

Yes, that's TB11, it goes to eleven!

I was wondering if there was any interest in developing an 'open source' preamp available to all of the TB community. Given the interest in the Tillman design and discussions centering around other designs I thought this might make an interesting project.

Here's what I was thinking in terms of requirements:

1) Must fit into a jazz bass cavity - size to be somewhere around 1 inch by 2 inches.
2) Single or dual battery operation
3) Volume/Volume/Bass/Treble controls - I was thinking a Baxandall feedback tone control that can give mid boost with only two tone controls. Switchable midrange frequency in the 500Hz to 1kHz range.
4) Overall gain of about 6db.
5) Fully buffered pickup gain controls, to eliminate pickup-to-pickup interactions and give a true 'blend' using a virtual-earth mixer stage.
6) Using low-noise high-slew rate Burr Brown op-amp stages, which are about the best you can get for audio circuits at a reasonable price.
7) Current drain around 1-5mA to give reasonable battery life (most 9V batteries are in the 500mAh range).
8) Switching jack to deactivate preamp when cable is unplugged.

Here is a block diagram that I came up with:



Bass/treble could be stacked to fit the standard three-holes of a jazz bass.

So we would come up with the design, prototype it, and if there is interest create PCB's.

Anything else come to mind? Thoughts?

[FunkMetalBass]

3-band.

[line6man]

Well, I'm not a fan of opamps, but I tend to feel that way for technical reasons as opposed to how well they work in practice.

I like the buffering for each pickup, but I'm not a fan of the 6dB gain. I would prefer unity.

[A440Hz]

I'd love to be a part of this project. I've been toying with the idea since I'm a DIYer and I have the Jones for an Audere or Aguilar, but just don't want to drop the cash for that. I'd rather have a PCB and some parts and put them together myself, and it'd be great for us to work together on something that's useful for lots of folks in the future who don't mind the smell of burning rosin.

I contend that having a Vol-Vol setup as the only option might be a little restrictive. Lots of folks love that setup, as it has its own advantages and its own history. But I personally like a Vol-Blend setup, and there are a lot of other folks who do. There's no reason why both can't be done as part of this project, albeit with different schematics. Having a PCB that can be configured either way would be a bonus, if a way could be found to do that elegantly.

Things I'd like to see (IMHO!), with my pile-ons noted as +1:

• Modularity and customizability
• Having each pup be buffered up front: +1
• Active summing of pickups (either with V-V or V-Blend): +1
• Customizable midrange frequency
• Option for Audere-style impedance switching? (not a huge priority, but very cool)
• 9V/18V option: +1
• Low current drain: +1
• Burr-Brown: +1
• Switching jack: +1; this is a must-have
• Baxendall is cool, but could there be "auto-gain" compensation? I'm wondering if the Bass/Treble pots could be stacked, with one of the pot elements on each knob used as a gain control. I'm not talking compression! I'm talking a gain compensation to make mid control easier. It doesn't have to be +12dB on the bass is -12dB overall gain; maybe something like +12 -> -3. Something subtle. Same with the treble control. And in the opposite direction for bass and treble cuts.
• active/passive switching is a plus for dead battery situations or just for vintage goodness
• passive tone is cool, but not mandatory (pre-buffer)--could be stacked with volume control
• Trimpot for active gain

The gain compensation thing is something I'm not sure about, but if there's no mid control, it would seem like a worthy pursuit. I'm pretty used to a mid control myself.

In my side job now with AudioOrchard.com, we're taking a piece-wise approach to the final product, so this doesn't have to be perfected on the first iteration. Getting something that's really workable on the first try, and then tweaking and adding features--that's a great way to go, IMHO.

[Stealth]

Originally Posted by MglMatador

I was wondering if there was any interest in developing an 'open source' preamp available to all of the TB community. Given the interest in the Tillman design and discussions centering around other designs I thought this might make an interesting project.

Much interest from my side - I see this as a way to learn many things, Master. I am weak in the way of Theory, and weaker in the way of Practice.

I do have a few questions before you begin, though:

Originally Posted by MglMatador

3) Volume/Volume/Bass/Treble controls - I was thinking a Baxandall feedback tone control that can give mid boost with only two tone controls. Switchable midrange frequency in the 500Hz to 1kHz range.

This one confuses me the most. I know the background between two-band control and the fact it's mostly implemented as a two-band shelf (in most preamps anyway). How would you create a switchable midrange frequency and get a midrange boost with only two controls? Don't most tone controls only produce a mid-boost when the bass and treble are rolled off?

Also, would it be complicated to produce a Bass-Mid-Treble variant of the two-band Baxandall EQ you're proposing?

Originally Posted by MglMatador

4) Overall gain of about 6db.

Could the gain control be implemented by using a trimpot or such to set anywhere from 0 to 6 dB?

Originally Posted by MglMatador

6) Using low-noise high-slew rate Burr Brown op-amp stages, which are about the best you can get for audio circuits at a reasonable price.

Which one were you considering, the OPA13x?

[MglMatador]

I contend that having a Vol-Vol setup as the only option might be a little restrictive. Lots of folks love that setup, as it has its own advantages and its own history. But I personally like a Vol-Blend setup, and there are a lot of other folks who do. There's no reason why both can't be done as part of this project, albeit with different schematics. Having a PCB that can be configured either way would be a bonus, if a way could be found to do that elegantly.

So a volume-blend would be: blend to mix the two pickups, then overall 'master' volume for the blended signal?

A few other notes:

1) Op-amps - there's no reason op-amps can't sound good if you select the correct specs. GBP isn't one of them. I think with a reasonable slew rate, rail-to-rail operation, and having a low output offset it should sound pretty good.

Discrete designs are possible but I think it would be very challenging to fit it on such a small board, especially with active tone controls. But we can try!

2) Volume controls - I always prefer to control volume by controlling gain in the feedback path, rather than using a volume pot that always changes the output impedance of the pickup as the pot is turned. However with a buffer so close to the front-end either style may work (especially if we want to support an active-passive switch).

3) Impedance switching is a great addition, and it pretty simple to support I would think.

Perhaps we can start by coming up with some designs for the volume/buffer section.

4) 3-band - the beautiful think about the mirrored Baxandall is that you get mid-boost without having a mid control: you get it by turning down both the bass and treble. However any options are on the table at this point.

Perhaps we can start by coming up with some designs for the volume/buffer section? Requirements:

a) Variable input impedance switching
b) Somewhere between 0db and 6db of front-end gain
c) High input impedance (> 1MOhm)
d) Low(ish) output impedance (< 1kOhm)
e) User control for volume (however that is implemented)

[MglMatador]

Which one were you considering, the OPA13x?

5mA max per channel is probably too much for an on-board design (unless you were willing to utilize a phantom-powered approach with a custom cable): but you need to burn power to get the 20V/us slew rate of those bad boys.

[line6man]

Originally Posted by MglMatador

5mA max per channel is probably too much for an on-board design (unless you were willing to utilize a phantom-powered approach with a custom cable): but you need to burn power to get the 20V/us slew rate of those bad boys.

Ha! There is no way I would ever fool around with 5mA current draw on an onboard preamp. That's insane!

[A440Hz]

Volume-Blend: yes, it's blend for the two pups and overall master. The blend pots are specially tapered so that each pickup is full-on for 50% of the rotation, and audio taper for the other 50%.

How about discretes for the buffers (FETs) and opamps for tone controls?

What about bootstrapping for increasing input impedance without adding noise from huge shunt resistors?

Options for Vol-Vol or Vol-Blend: if we go active-passive switching, using high enough values for either mode would be beneficial. Otherwise, we'd have to find some dual-ganged pots with, say, 25k on one wafer and 250k or 500k on the other--not easy to find, but possible to make. But DIY-made pots might make the project too daunting for some. Who knows. Why not just use 250k or 500k values for the pots? Yes, higher values add more noise, but it's probably not a showstopper.

I, too, like the idea of the pots in the feedback loop rather than in a shunt config, but as you say, it may not matter because of the buffering. As long as the pups see a constant load (and possibly a user-switchable one!), there will not be a tone change with the volume.

Your specs look good, MglMatador.

[A440Hz]

FIYFYI: an example from Carvin.

Note that it has active-passive switching and uses a 500k volume control on the output. I worked at Fender with a guy who had worked at Carvin, and he said they used some sort of low-power op-amp. I had my Carvin bass for 10 years, and I think I changed the 9V about 4-5 times.

Finding the right op-amp is not easy, but a good compromise between performance (slew rate, THD, etc.), noise, and power can be had. My Carvin was nice and quiet.

[MglMatador]

Here is my first stab at a discrete JFET volume/buffer stage:



1) Uses the regular 500k volume pot for volume control. There is a switch to bring this down to 250k to change the magnitude of the resonant bump (like swapping in a 250k pot).
2) Input impedance - 250k to 500k
3) Output impedance - about 650 ohms <= this is why we need this stage!
4) Capacitive coupling with 4.7uF into basically infinite input impedance of the next stage = high pass at near DC (which means this can be lowered to 1uF probably)

Source-follower uses up about 3db of gain, however that can be made up for later in the mixing and tone stages.

Transient analysis of pickup output versus buffer output:



AC sweep of circuit. Red line is 500k pot, green line is with 250k 'virtual pot' with switch closed:



EDIT: Also, DC current draw is 0.16mA.

[JackANSI]

Could always just do it like EMG with a seperate board on the back of the balance pot for those who want that..

Someone could get the whole design rolling by creating a unity pickup buffer that fits nicely on a vol pot. (hey!)

I would really like to see the design done 2-sided SMT, that would leave the most room for extras and its not that much harder than through-hole to assemble.


A440Hz:
I think carvin still uses the MC33178

[MglMatador]

I would really like to see the design done 2-sided SMT, that would leave the most room for extras and its not that much harder than through-hole to assemble.

Absolutely. With 1206 sized resistors, SOT23 packages for JFETS, and SOIC/TSSOP's for op-amps there little reason to go with through-hole components.

[MglMatador]

Originally Posted by JackANSI

Someone could get the whole design rolling by creating a unity pickup buffer that fits nicely on a vol pot. (hey!)

Very cool idea! Perhaps place each buffer/volume on a 1" square board that would get silicon glued to the back of each pot? And perhaps the same for the tone network on the back of the tone pot(s)? That would be a three board solution.

[A440Hz]

The impedance switching thing--I have a slightly different idea to consider. I would put the impedance switching where you have it currently, but without the volume control there. Put the volume (or blend) at the output of the pup buffer. Make the input Z higher, like, say, 1Meg.

It would be cool to have the low-Z mode as low as 50k or 100k. Make it switchable (or user-configurable) with a trimmable gain compensation for the impedance change. Experimentation would help, because we're talking volume compensation here, which is somewhat subjective. Audere does provide a trim for the Low-Z mode, actually for both pickups individually.

If you do another SPICE run with a 50k or 100k load total on the pup, you'll see the very low-Q result. (see Audere's graph here.)

Done a little research on Burr-Brown (i.e., TI) parts. I don't see anything that competes with the aforementioned ON Semi part (spec sheet here, PDF). Low current draw (<1mA per amp), good slew rate (~1-2 V/us), low THD (0.0024% typ @ 1kHz).

Thanks for the tip, JackANSI.

Correct me if you see anything better--from any manufacturer. Oh, and the On Semi part handles an 18V supply for folks insist on that. There are TI parts that do as well, but none that I could find that had a high enough slew rate.

Again, the Carvin pre has performed very well for me in terms of sound and battery life, so their op-amp choice must be a good one.

[JackANSI]

Couple of quick links for people curious about the type of tone control proposed:


http://www.schmarder.com/radios/tech/tone.htm


http://www.duncanamps.com/tsc/

[MglMatador]

It would be cool to have the low-Z mode as low as 50k or 100k. Make it switchable (or user-configurable) with a trimmable gain compensation for the impedance change. Experimentation would help, because we're talking volume compensation here, which is somewhat subjective. Audere does provide a trim for the Low-Z mode, actually for both pickups individually.

I tried loading a single coil passive jazz pickup with 100k, and it sounded extremely dark to me. Is the low-Z mode meant only for low(er) impedance pickups?

It's amazing how much the bass pickup sound centers around the location and magnitude of the resonant bump.

I have a good spice model for the MC33178 so i'll see what I can come up with.

[MglMatador]

Here is version 2 of the volume / buffer section:



Changes:

1) Moved to the MC33178 op-amp as provided by A440Hz. This is a bipolar op-amp, which has a lower input impedance than a FET input op-amp, and cannot drive to the rail (it can drive up to about two collector-emmitter saturation voltages less than the rail for both sides), however with 4.5V of headroom this isn't an issue even for extremely hot pickup outputs.
2) Move the volume control to after the buffer - this allows use of low resistance pots which should be less noisy. I used a 25K but I would guess even a 10K would work fine
3) Input impedance - input current is still tiny at these voltages, so input impedance is still in the mega-ohm range
4) Output impedance - very tiny! SPICE says it's about 90 ohms.
5) Added a trim for input impedance. If the switch is open, you can adjust from 250k to 500k. With the switch closed, you can adjust from about 50k to 60k.

This circuit also does not have 3db of gain reduction (with the volume pot at max) like the discrete version (thanks to gain compensation in the op-amp). All in all I like the specs of this one better, and two channel would fit into the layout space of half of a single channel of the first design.

Transient response (at 50% volume control):



AC sweeps: top line is 500k, middle is 250k, and bottom is 50k (very close to Audere's results):



Current draw is low: there is 45uA to set the bias voltages, and if the datasheet is to be believed the op-amp takes about 0.5mA per channel.

[A440Hz]

Originally Posted by MglMatador

I tried loading a single coil passive jazz pickup with 100k, and it sounded extremely dark to me. Is the low-Z mode meant only for low(er) impedance pickups?

It's amazing how much the bass pickup sound centers around the location and magnitude of the resonant bump.

I have a good spice model for the MC33178 so i'll see what I can come up with.

Yes it does sound dark. That's the point. And that's why it's switchable. Because the Z-switch is a change to the resistive loading (I presume) and not to the capacitive, it's going to change the Q of the response. Differently than the tone control, I believe, seeing as how the tone control is shunting various amounts of signal to ground through the cap.

I'm replicating your SPICE model now to mess around with just the pre-buffer passive section (Z loading, passive tone, etc.).

This is fun stuff. I'm a digital guy these days, even though I did a lot of analog back in school. But that was about 13 years ago.

Nice work, MM.

[MglMatador]

Because the Z-switch is a change to the resistive loading (I presume) and not to the capacitive, it's going to change the Q of the response.

Looking at Audere's site, they do include a switchable cap that is basically parallel to the pickup, which will move the frequency of the resonant bump (downward for increasing C). This simulates what happens when you plug in a really long guitar cable that might put an additional 2000+ pF loading on the pickup. It all follows this equation:



Increasing C will make the first term smaller, which will lower the resonant bump (up to the point where the quantity under the square root becomes zero). You can also see that inductance dominates this equation as well.

For loaded pickups, the following equation applies:



You can see that decreasing the load resistance lowers the resonant bump as well, and also decreases Q, which can make the pickup sound less 'peaky'. You can see from the output response it's essentially flat with 50K over the entire range of bass frequencies.

With the volume controls placed after the buffer, volume / blend and volume / volume setups become pretty trivial to implement without changing the fundamental sound of the loaded pickup (no more treble loss as the volume controls are turned down as in the passive V/V/T case!).