dandyguitar

[dandyguitar]

Tube amp trio, 10 years on

Yeah I'm not dead, I just bought a house and took a few years off from making music-stuff so I could focus on house stuff

Anyway

Last year I spent... countless hours designing some guitar pedals that I still haven't built yet... but in the process, learned a lot of new quirks of vacuum tube behavior. And while designing & prototyping those pedals (which I will finish the next time that I get the urge to solder), I realized that some of those weird quirks explained why my amps never sounded quite the way I wanted to sound.

For example:

Did you know that the gain factor of a triode changes depending on the impedance it has to drive? And that's in addition to whatever signal loss you're going to be seeing when you consider its own somewhat high output impedance? For example, a 12AX7 common-cathode gain stage loaded with 50k only generates a little more than half as much gain as it would with a 500k load, and you're still looking at a 38k anode impedance running into that load. So it produces a smaller amplified signal and then attenuates it more. Consider the final preamp triode in the classic Fender AB763 circuit: at a glance it looks like its output just runs through some mixing resistors on its way to the phase splitter, but then you spot that pesky Tremolo intensity pot hiding further down in the schematic. Oh, and those mixing resistors and the other channel's anode resistors all form a big complicated network of AC paths to ground, acting in parallel with the more obvious parts of that load... and right before that triode is the big signal-attenuating reverb filter. So that last stage isn't spitting out a great big signal and slamming the phase splitter like I thought it was in 2014: it's taking a very manageable signal, spitting out a small signal, and then throwing even more away on its way to the splitter. No wonder my Fender-inspired build always seemed too dirty and never did the spanky-clean funky stuff properly.

I rebuilt the entire pre-amp in the Fender-y amp and now it's just dandy. Added some more options too, so now it convincingly covers the entire Fender clean spectrum, certain Dumble settings, and the Mesa Mk1, Lonestar, and Blue Angel pre-amp circuits. Very flexible. Lots of mini toggles.

Another example:

I was - and remain - power tube agnostic when it comes to BS like "American power tubes sound glassier and cleaner than British power tubes!" No they don't. That's how the preamps and speakers are voiced. It has nothing to do with the power tubes.

Buuut

That "they're all the same, just different power levels" attitude made me miss that different models of tube have different levels of input headroom.

For example, it turns out that a 6v6 starts to distort at about the same level of input signal as an EL34 does, which means that the power tube distortion start at an appropriate level for my Fender and Marshall styled amps when they're cranked up loud, but an EL84 offers much less clean input room before it hits its maximum output and distorts. So the Vox styled amp, with the 6v6 output section I gave it back when there were no trustworthy manufacturers for EL84s, was completely incapable of pushing its output tubes into the kind of saturation demanded by Queen. So I ended up rewiring the power transformer to nerf the voltage, replaced the output transformer to an EL84-friendly value, and popped in new sockets to use the right tubes. Luckily JJ makes good EL84 tubes now and so the reliability problems are solved and the amp sounds right again.

Okay, one more example that has nothing to do with the tubes:

When I chose my speakers for these amps I based my choices on what kind of aftermarket speakers people were putting into their amps at the time, and alnico blues were extremely popular. So sure, why not? Jensen alnicos it was. Fast forward ten years and I've realized: that's not what came in Fender amps in the 60s, and that's part of why my Fender-ish never sounds properly Fender-ish. Bought me a proper cheap ceramic Jensen, wired it up in the speaker cabinet, and voila: the crisp dare-I-say "ice-picky" trebles that were always missing were there. And to think, this wouldn't have been a problem if I had originally gone with the same "we chose this one because it was cheap" speaker the things shipped with half a century ago.

. . . . .

Yeah it turns out if you want to make something that sounds like the amps you've been listening to from the last 60 years of recordings, the best way to do that is to pay careful attention to what was inside those amps instead of trying to reinvent the wheel.

[dandyguitar]

I got a lot of recurring questions about guitar wiring back when I worked as a guitar tech.  I decided it’s finally time to post some answers with helpful, informative pictures!  Without further introduction, here is part four.

Guitar Controls 4: Treble Preservation Circuits

Guitars seem to lose some clarity as the volume is turned down.  What is the best way to preserve that clarity without getting too bright as you turn down the volume?  Is there any downside to adding a treble bleed mod to your guitar?  Let’s find out.

And before we get started, I’m so sorry.  10 graphs this time?  Too many, yes, but I’m not getting rid of any.

Let’s look at five different setups:

Standard wiring (pickups and tone controls wired to input lug of volume pot when viewed from behind)

“50′s Gibson” wiring (pickups wired to input lug of volume pot, tone control wired to output/middle lug)

Ibanez treble bleed (330pF capacitor wired between input and output lugs of volume control)

Seymour Duncan treble bleed (0.0022uF capacitor and 100k resistor wired in parallel between input and output lugs of volume control)

Kinman treble bleed (0.001uF capacitor and 130k resistor wired in series between input and output lugs of volume control)

In each of the images, each line on the graphs represent one notch on a volume control from 2-10; 1 would be completely silent and wouldn’t show up on the graph.  A sonically perfect volume control would create the same frequency response across its entire range, only quieter.  Therefore, if the traces on a graph all exhibit the same shape, the volume control configuration in question will sound similar through its entire range.

The first two images show the Standard and Gibson wiring schemes, neither of which requires any additional parts.

On the standard volume control (image 1), the resonant peak basically flattens out between 7-9 and then appears again as the control is turned down below 6.

The Gibson wiring (image 2) smooths out the overall taper of the volume control between 5-10 on the dial (as can be seen in the more evenly spaced traces on the graphs), without changing the output level at 2 on the dial.  The resonant peak gets smaller between 7-9, much like the standard control, then gets slightly bigger than the max volume peak below 6 on the dial.

The second row of three images show the Ibanez, Seymour Duncan, and Kinman treble bleed responses, all of which require one or two extra parts.

The Ibanez method (left image, second row) preserves the resonant peak a little better than the Gibson method, and it maintains the same dynamic range as the standard volume control again.  However, its resonant peak ends up about 2.5x larger than the max volume peak as the volume control is turned down below 7.  Thus, this treble bleed ends up being significantly brighter and thinner sounding at low volume settings than at maximum volume.

The Seymour Duncan method (middle image, second row) manages to keep roughly the same intensity of resonant peak through its rotation, though the resonant frequency drops by about an octave.  Also, it has a huge effect on the taper of the volume control.  The Seymour Duncan volume control stays loud longer, but has a sudden huge drop-off between 1 and 2 on the dial.  It’s so much louder that 2 on the Seymour Duncan dial is about as loud as a 5.5 on a standard dial.

The Kinman method (right image, second row) keeps the resonant peak at about the same frequency, but the resonant peak grows about twice as large as it is at max volume, and a significant amount of mids are enhanced at the same time.  This results in a treble bleed circuit that is fatter sounding than the Ibanez, but brighter than the Seymour Duncan.  Its dynamic range is still in the same ballpark as the standard control, unlike the much louder Seymour Duncan control.

But wait - what about side effects?  It turns out that treble bleed circuits can have unintended consequences, such as making your tone control borderline useless at any volume setting below 10...  Read on.

The third row of two images shows the Standard and Gibson wiring schemes again, but this time with the tone control turned down half way.

The standard volume control (left image, third row) shows what we would expect of a guitar with its tone control turned down a bit and the volume being gradually lowered: smooth response (mostly evenly spaced lines) with no resonant peak and treble frequencies rolled off.

The Gibson method (right image, third row) is pretty much the exact opposite of what we want: the resonant peak, which we tried to knock off with the tone control, distinctly reappears as soon as the volume control drops below 10.  Its response is almost the same with the tone rolled down as it is with the tone all the way up, but with the overall output lowered by -5dB.  The tone control is much less effective at volume settings below 10 with the Gibson wiring method.

The fourth row of three images shows the Ibanez, Seymour Duncan, and Kinman methods again, but with the tone control turned down half way.

The Ibanez method (left image, fourth row) responds properly to the tone control until the volume gets down to about 6 on the dial.  Below 6 it develops a resonant hump, but it’s much wider and less pronounced than the frequency response was with the tone set to 10 (take another look at the original response of the Ibanez control to remember how huge the treble spike got).  In other words, it has more treble than the ideal standard volume control, but the tone control is still knocking about -9dB off the resonant peak and eliminating a noticeable amount of treble as it should.

The Seymour Duncan method (middle image, fourth row) responds properly to the tone control and barely develops much of a treble hump at all, but the dynamic range of the volume control is still skewed very much toward the “loud” side of things.

The Kinman method (right image, fourth row) rolls off its treble resonant peak but keeps its hefty upper-mid hump.  Much like the Ibanez method, its response is far from the standard volume control, but the tone control is still rolling off treble and smoothing out the tone as it should.

Conclusion

What recommendations can we take away from all of this?

The standard volume control loses some treble between 6-10 on the dial, but regains it at lower settings.  The tone control works properly all over the dial.  However, some people find this arrangement too dark (especially when playing with a dirty amp where a little extra treble keeps things crisp as the guitar’s volume is turned down) and would rather have their treble frequencies boosted at low volumes for a brighter, smaller sound.

The Gibson method boosts those treble frequencies a bit at low volumes, but it makes the tone control pretty useless at any volume setting below 10.  Personally, I do not recommend it because it doesn’t actually improve treble response by more than about +2dB, and I can’t stand losing the use of my tone control.

The Ibanez method creates a huge resonant peak as the volume is turned down.  This can be pretty harsh with bright pickups (such as single coils) and is best suited to very dark pickups.  Some players may find it too shrill.  On the other hand, the tone control still works and the dynamic response of the volume control isn’t compromised.

The Seymour Duncan method provides fairly even frequency response with or without the tone control engaged, but it compromises the dynamic range of the volume control.  On the other hand, this would be great for players who mostly play clean or low gain amps and need finer control over the “medium/loud” end of the control than the “soft/whisper quiet” end of the control (which is usually only useful to clean up a super distorted signal).  For example, it is well suited to archtop guitars for jazz.

The Kinman method not only boosts treble frequencies, but also gradually blends in significant amount of extra mid-range.  The tone control still works  This keeps the pickup brighter than the standard control without being as shrill as the Ibanez method.  This is a good choice for players who use a lot of distortion and need the extra clarity in the upper mids without a shrill resonant peak as they roll down their volume control.

In summary, each method has its pros and cons.  Changing the values of the capacitors and resistors used for the Ibanez, Seymour Duncan, and Kinman methods will change their frequency responses slightly, but their basic behavior will stay the same unless you make a radical departure from the standard values.

Note: this test simulated a 7k underwound PAF-style humbucker, 500k audio taper volume and tone pots, and a 0.022uF tone capacitor running into the input stage of a blackface Fender amplifier through a moderately priced cable.  Frequency response will obviously vary with different pickups, controls, cables, and amplifiers.

[dandyguitar]

I got a lot of recurring questions about guitar wiring back when I worked as a guitar tech.  I decided it’s finally time to post some answers with helpful, informative pictures!  Without further introduction, here is part three.

Guitar Controls 3: Potentiometer Size

What effect do the size of your volume and tone controls have on the sound of your pickups?  Are large potentiometers (pots) brighter?

Let’s look at four different setups:

Cyan: 1M volume and tone controls (sometimes recommended to brighten up dark pickups)

Blue: 500k volume and tone controls (common in humbucker-equipped guitars)

Green: 500k volume with 250k tone (rarely seen)

Red: 250k volume and tone controls (common in single-coil equipped guitars)

The first image shows these four control setups with volume at 100%, 75%, and 50%.  With our volume control set to 100%, we see that the larger the controls, the larger the spike at the resonant frequency.  That resonant peak accounts for much of the treble character of the pickup.  Note, however, that the treble content beyond the resonant peak is eventually the same for all the controls: super-high treble is mostly unaffected by the pot size.

With the volume control set to 75% and 50%, the result isn’t quite the same.  At 75%, the 1M pot drops below all the others (in terms of output) and all four arrangements more or less lose their treble peak, which results in the pickup sounding darker.  With the control set lower yet at 50%, a smaller treble peak has developed at a higher resonant frequency, but it’s still not as large as it was with the controls set to 100%.

Conclusion

The larger the pot size, the higher the resonant peak at maximum volume (and the brighter the guitar sounds).  Pickups with a low resonant frequency (like high output, high DCR humbuckers) might sound better with a 1M control.  pickups with a high resonant frequency (like Strat style pickups) are often paired with the smaller 250k controls to tame their treble frequencies a bit.  None of the controls do a good job maintaining a consistent resonant peak as the volume is turned down.

Side Note

Why did I include the 500k volume / 250k tone combination?  Well, the smaller the tone control, the finer the control you have over the low end of the control.  The second image shows a 500k volume / 500k audio taper tone control, while the third image shows a 500k volume / 250k audio taper tone control. Each line on these charts represents 1 notch on a tone control marked from 1-10. The smaller tone control lowers the resonant peak by about -1dB compared to the matched 500k controls, but the “tone control minimized” setting has the same frequency response.  Meanwhile, the 250k pot devotes more of its sweep to the dark end of the control, evening out the behavior of the control a bit more.

Conversely, a 1M tone pot will give very poor control over the low-frequency end of the control compared to a 250k tone pot.  It helps brighten up the guitar, but the control becomes more difficult to dial in at the bottom end of its rotation.

Side Note Conclusion

If you can get away with a smidgen of treble loss, a 250k tone pot will give you much finer control than a larger pot.  If you can’t handle the treble loss, you could try a switched pot that takes the control out of the circuit when it is turned all the way up to 10, giving you the best of both worlds: no treble loss from the tone control when it’s maxed out, and finer control over treble content when it’s engaged.

Note: this test simulated a 7k underwound PAF-style humbucker, variously sized audio taper pots, and a 0.022uF tone capacitor running into the input stage of a blackface Fender amplifier through a moderately priced cable.  Frequency response will obviously vary with different pickups, controls, cables, and amplifiers.

[dandyguitar]

I got a lot of recurring questions about guitar wiring back when I worked as a guitar tech.  I decided it’s finally time to post some answers with helpful, informative pictures!  Without further introduction, here is part two.

Guitar Controls 2: Tone Capacitor Size

What effect does capacitor (cap) size have on a tone control?  Is there a difference when the control is all the way up at 10?  Let’s look at the frequency response of five different cap sizes:

0.100uF (some vintage Fenders use this)

0.047uF (many modern Fenders use this)

0.022uF (most modern humbucking guitars use this)

0.010uF (I use this in my guitars)

0.005uF (I’ve never actually seen this used, but it’s interesting to see!)

The first image shows the tone control set to 10.  As you can see, the different sizes of cap all show such similar frequency responses that they basically sit on top of each other on the graph.  In other words, capacitor size makes no measurable difference when the tone control is all the way up.

The second image shows the tone control set to 6.  Comparing the largest cap (green line on the graph) to the smallest (magenta line), the smallest cap has about +2dB higher output at 1KHz, making it sound slightly clearer.  The three most common cap sizes in production guitars (green, blue, and red lines on the graph) are almost identical.  The treble rolloff is basically identical above 3KHz for all five sizes of cap.

The third image shows the tone control set to 3.  At this point we start to see a noticeable difference between the different caps.  The tiny 0.005uF cap begins developing a resonant peak at 900Hz, the 0.010uF cap has a couple decibels more treble than the more common cap sizes, and the largest 0.100uF cap is loading down the pickup’s signal through its entire frequency response.  Again, the treble rolloff is basically identical above 3KHz for all five sizes of cap. 

The fourth image shows the tone control set to 1.  With the tone control set to its minimum, all the caps (except the 0.100uF) develop a resonant peak somewhere in the midrange of the guitar.  Smaller caps develop larger resonant peaks at higher frequencies.  This peak gives the guitar its “honky” sound, like a wah pedal somewhere in the middle of its sweep.  The higher frequency and larger resonant peak of the smaller caps is what makes them sound brighter with the tone control rolled all the way down.

Assessment of Results

0.100uF, 0.047uF, and 0.022uF caps sound pretty similar all the way from 3-10 on the dial, and only show a significant difference between 1-3:

With the tone control rolled all the way down, the 0.100uF tone cap is attenuating frequencies well down into the playable range of the guitar - it’s not just rolling off overtones!  The open high E string is already at -3dB, and it’s down -27dB by the time you get to the 24th fret.  This is way too much treble attenuation to be useful.  The 0.047uF tone cap isn’t much better: it’s still down -20dB at the top notes of the guitar.

At the other end of the spectrum, the smallest 0.005uF tone cap shows no attenuation at the top notes of the guitar - in fact, its resonant peak is focused right in the top octave of the instrument.  The peak is even taller than it is with the tone control all the way up at 10 (though at a lower frequency), giving those notes a noticeable boost.  This could be useful if you enjoy a “cocked wah” lead tone.

Sitting right in the middle, the 0.022uF tone cap has a resonant peak at about 500Hz.  The 0.010uF tone cap’s peak is at about 750Hz, which is similar to the frequencies that usually get the most push by a certain green overdrive pedal.

Keep in mind: all these frequencies depend as much on the DC resistance and the inductance of the pickups being used.  A higher output humbucker will move all of these frequencies lower and a single coil pickup would move them all higher.

Conclusion

The largest tone caps are too dark to be useful when they’re turned down to zero.  The smallest 0.005uF tone cap would provide a nice, smoothed-out, mid-rich tone with high output humbuckers, the 0.010uF provides the similar response for low output humbuckers, and moving up one more size (0.022uF) will provide a similar response with Strat and Tele style single coil pickups.  Larger capacitors will tend to provide a dead, lifeless response with the tone control rolled all the way down.

Note: this test simulated a 7k underwound PAF-style humbucker with 500k audio taper pots running into the input stage of a blackface Fender amplifier through a moderately priced cable.  Frequency response will obviously vary with different pickups, controls, cables, and amplifiers.

[dandyguitar]

I got a lot of recurring questions about guitar wiring back when I worked as a guitar tech.  I decided it’s finally time to post some answers with helpful, informative pictures!  Without further introduction, here is part one.

Guitar Controls 1: Tone Control Taper

Which is better to use for a tone control?  A linear or audio (logarithmic) taper potentiometer (pot)?  Let’s look at the frequency response of the two controls:

The left image shows the response of a linear taper pot.  Each line refers to a notch on a knob numbered 1-10.  Between 10 and 3 on the dial it reduces the intensity of the pickup’s resonant peak (2.5kHz in this case) without rolling off much extra treble above that frequency.  Almost 80% of the control’s rotation is devoted to this subtle change.  Between 3 and 2 it suddenly eliminates the resonant peak entirely and rolls off an extra ~3dB above the peak.  The final chunk of the tone control’s rotation from 2 to 1 rolls off a whopping -24dB of treble and introduces a new resonant peak in the mids turned all the way down to 1.  This is a pretty drastic shift for the final ~20% of the control’s rotation: a mild resonant treble peak down to almost completely eliminated treble content with a resonant peak in the mids.

The right image shows the response of an audio taper pot.  Again, each line refers to a notch on a knob numbered from 1-10.  Between 10 and 6 on the dial it reduces the treble resonant peak to nothing.  Between 6 and 3 on the dial it gradually introduces up to -12dB treble rolloff.  Between 3 and 1 on the dial it completes its total treble rolloff and introduces the mid-range resonant peak.

Conclusion

The audio pot is better if you want a more even response spread over the entire range of your control’s motion: a half turn of the knob flattens out your treble resonant peak by -6dB, another quarter turn rolls off about -12dB treble, and the final quarter turn gives you fine control over the final -12dB and the introduction of the mid-range resonant peak.

Compare that to the linear control: the first half turn of the knob barely reduces your resonant peak by -2dB, the next quarter turn reduces it by another -3dB, then the final quarter turn reduces your treble by -25dB.  Not smooth.

In my opinion, the audio taper tone control slays the linear taper control.

Note: this test simulated a 7k underwound PAF-style humbucker with 500k pots and a 0.022uF tone capacitor running into the input stage of a blackface Fender amplifier through a moderately priced cable.  Frequency response will obviously vary with different pickups, controls, cables, and amplifiers.

[dandyguitar]

Belle Gen. 5: Body coats

In the last post, I “filled” the pores with black grain filler.  I say “filled” because although every pore got some filler in it, they are not filled perfectly flush to the wood surface.  Some filler always gets pulled out of the pores as I’m wiping the excess off, so it is not yet perfectly flat.  Therefore the clear coat will have to fill in what remains of the open pores and then be sanded back to leave a perfectly flat finish.

It’s basically the same process no matter what kind of film-building finish you’re using (whether it’s shellac, lacquer, varnish, or water-based): apply lots of coats of finish until the film is thick enough that you can sand everything flat (i.e., sand down to the level of the low spots, such as the pores) without cutting through the film and hitting the wood at any point.  Different finishes will have different limitations on how to apply the multiple coats in order to get them to bond properly, but the overall concept is the same no matter what the finish.

I like to use shellac because it’s non-toxic (aside from the alcohol in which it is dissolved), I can apply it in my kitchen, it dries and cures quickly, and there’s no limit on re-coat time because each application of shellac bonds completely with the layer below it to form a single solid film.  It is extremely forgiving to work with.  Its resistance to scratching, liquids, and solvents is not as high as a good varnish or water based finish, but it is a big step up from the other easy-to-use kitchen-safe finishes (i.e. oils).  (Bob Flexner’s book is a great resource for this kind of information.  This is not an affiliate link, I just really like the book.)

To build a thick film of shellac would take a

long time

using traditional French Polish techniques, so instead of applying the shellac with a rubber I will use brushes.  The few marks that the brushes leave will be sanded out at the same time the finish is leveled with sandpaper (as mentioned above).  I use Golden Taklon wash brushes because they hold lots of finish and they leave a pretty smooth result (top photo).

After applying a few applications of shellac to the bodies, the finish is starting to build (bottom photos).  The bodies are getting glossy and the clear coat has enhanced the blackness of the grain filler (which tends to dry a bit grey).  It will take at least a week of brushing on shellac once or twice a day to build up a thick enough coat to confidently sand the finish flat without having to worry about “burning through” and hitting the wood below, which would wreck the dye job.

2024 Edit:

Shortly after this post I was looking at my finances and crunching numbers and decided, “time to buy a house.”  Everything switched gears and I put my guitar projects away for several years so I could focus on finding and buying a house, then developing a dirt-patch of a yard and building and finishing the basement and then COVID happened and yada yada yada and ANYWAY it’s 2024 now and I’ve actually been back to work on my guitars for a year and a half but I’ve been very lazy about adding updates to this blog because it has mostly been Other Guitar projects instead of the ones you see here.

These two guitars are sitting quietly on a shelf, mostly varnished up, nearly ready for final polishing.  I’ll get to them eventually, and when I do I’ll post the finished results.  For reals.

[dandyguitar]

Belle Gen. 5: Filling the grain

Swamp ash has large, open pores.  It takes a lot of shellac to fill them up and leave the wood smooth and level.  It’s quicker if you fill those pores in with some sort of grain filler.

Traditional French Polish pummice methods won’t work.  If applied before dye, the dye won’t penetrate the wood properly.  If applied after the dye, it burnishes most of the dye off the surface as it abrades the wood and fills the pores.  Therefore I will use TIMBERMATE (center photo).  This is the same grain filler as we used at [unnamed boutique bass guitar manufacturer] when I worked there.

Before applying the grain filler, the entire guitar needs to be sealed with shellac.  (Freshly sealed wood with no filler can be seen in top left photo.)  Tinted grain filler has dye in it and it will tint bare wood, but it won’t penetrate shellac.  The layer of shellac has to be thin enough that it doesn’t fill the pores (and it leaves them with nice crisp edges to catch the grain filler as you smear it in), but thick enough that when you don’t burn through the shellac and nick your dye job as you clean the excess filler off.  You can see that the body is a bit shiny in many of the above photos due to the shellac sealer.

I chose black grain filler because I thought it would look pretty rad on #13.  I also figured it would provide some interesting contrast for #12 (above), and I just wasn’t sure what other color I could pick that wouldn’t clash with the red-to-yellow burst.

THE PROCESS

Mix some water in with your Timbermate until it has the consistency of warm peanut butter.  (Not shown.)

Smear some grain filler onto the sealed wood.  I use old cotton t-shirt material  wrapped around a couple cotton balls as a rubber.  Don’t be afraid to lay it on thick!  Really work it into those pores.  (Top center photo.)

Wipe off the excess with an old cotton t-shirt before it dries too hard.  (Top right photo.)  Some filler will remain in the open grain, making it darker.  It also means you won’t have such deep holes to fill with shellac later (although I find that it never fills quite perfectly flush to the surface).  There will still be a thin film of filler on the surface, leaving it looking a bit blotchy and hazy.  It will be cleaned off later.

As you wipe it on, you will notice the pores filling up and getting darker.  In the bottom left photo, the close half of the body is filled and the far half of the body is still unfilled.

Once the filler is completely dry (I wait 24 hours), clean the thin film of remaining excess filler off the surface with 400-600 grit abrasives.  I like foam-backed abrasives because they tend not to leave scratches like the corners of a folded piece of sandpaper will.  In the bottom center photo, you can see the body has been sanded clean with 600 grit, but the control cover has not been cleaned yet.  Notice how much dirtier it looks.  Take special care NOT TO BURN THROUGH YOUR SHELLAC because if you hit your wood you’ll leave a light spot in your dye job!

Finally, all the sanding is done!  The surface is clean, the pores are filled, and the wood is nice and smooth-ish and ready for lots of body coats of shellac!  (Bottom right photo.)  The grain filler doesn’t quite show up as black yet, but its color will darken up once coated with shellac.

Note: the maple is a closed-grain wood.  It does not require filling.  There are no pores large enough in the wood to catch any grain filler even if you tried, so it’s basically ignored through this entire process.

[dandyguitar]

Belle Gen. 5: Five color coats

With #13, I was aiming for a rich golden brown color in order to take the best advantage of the roasted curly maple top.  With this guitar, on the other hand, I wanted to pay homage to the instrument that inspired these two guitars in the first place: the classic Les Paul.  The earliest LPs were all gold-tops, and then they switched to a cherry burst in 1958.  The factory hadn’t started exclusively searching for figured maple yet - they just used any old maple they could find in the appropriate dimensions - so many of the earliest cherry burst LPs exhibit fairly plain or irregular wood figure.  Thus, I decided this classic burst would be an appropriate match for the for the unusual quilt/birdseye maple top I used.

The front is more orange overall than I wanted.  The maple itself had a bit of a pink tint to it and some red dye probably bled in a bit.  I probably would have had to bleach it if I wanted to make it a really bright yellow.  Hrrrrm.

Two coats of Yellow-R, one coat of Red, one coat of Red/Black/Van Dyke, and a final coat of Yellow-R and Red/Black for final color blending.  The wood is still wet in this photo.  Note the matching cover plate!  (The other guitar has one too.)  The grain isn’t a perfect match, but at least the color is the same.

[dandyguitar]

Belle Gen. 5: 4 color coats and some oil

After applying the first reddish undercoats of dye, I went back to the guitar with some sandpaper (specifically, a 400 grit foam-backed sanding pad).  My goal was to lighten up the center section (for more contrast with the burst), sand back any areas that got too dark, and sand back any end-grain that was sealed too well to absorb color (so it would absorb more dye in the next coat).

After that, I repeated the process with a brown dye mixture (Van Dyke brown mixed with a little red) and yet again with a deeper brown (almost the same mixture as before, but with an extra drop of red and one drop of black).  Once that was all dry, I gave it one final touch-up with the sandpaper again to clean up any needlessly dark blotches and improve the blend from dark edges to light center.

A common French Polish technique is to begin with a coat of drying oil before applying any shellac.  I think the oil adds some translucence to the top fibers of the wood, much like grease will make paper transparent (bottom photo).  Whether or not that’s what actually happens to the wood, adding a coat of drying oil before shellac adds visual depth to the wood and increases its chatoyance.  However, applying oil to an open-grained wood like Swamp Ash is problematic: the oil in the pores tends to cure slowly and often leaches out for weeks afterward, which is a pain.  Plus, I think it makes ash look kinda splotchy, and I don’t like that.

Therefore, I only oil the maple top of the guitar.  Any drying oil will do, but those that cure quickly are your best bets, such as polymerized oils.  I like to use Tried & True Varnish Oil, which is a blend of polymerized linseed oil and processed pine resin.  The important thing (in my mind) is that it contains no solvents or heavy metal driers, neither of which are healthy to work with (and neither of which I want underneath my later coats of shellac).

The next step will be to seal the instrument with a few thin coats of shellac, then fill the pores, and finally go to work on the top coats.

[dandyguitar]

Belle Gen. 5: First color coat

For this instrument, I want the core of the flame maple to look a bit like Roger’s Golden Syrup, bursting to a deep brown somewhere between East Indian Rosewood and Black Walnut.  The ash sides and back will have that deep brown color and will lighten up a bit in the middle of the back.

It took a couple tries to get the right color of brown.  I discovered that a hand-rubbed dye under shellac can look quite different depending on the lighting conditions.  In dark light, it looks dark.  In bright sunlight, it seems to light up and almost glow.  A color that seemed fine in the workshop looked drastically different in the living room.

First I tried using plain Van Dyke Brown.  It showed some promise under dim light, but under sunlight it took on a fairly yellow character, almost like Ovangkol.  I tried adding Burnt Sienna to the brown and it helped a bit, but it still needed a little more warmth.  Finally, I tried an undercoat of 3:1 Burnt Sienna and Red, let it dry completely, then worked over it again with some 5:1 Van Dyke Brown and red.

I feel like the two applications of colors (instead of one application of everything mixed up at once) not only gave me the color I wanted, but allowed me to work the two colors of dye differently, letting it sit and penetrate longer with one color than with the other.  This meant that the final color varies slightly (much like real wood grain does) in the finished product, rather than being such a solid, monotonous single color as with one application of dye.

In this photo, the Burnt Sienna / Red dye undercoat has just been applied.  The dye has been wiped on by hand using cotton balls wrapped in old t-shirt material, much like in this video.  The various cavities are still dry and untouched by any dye.  The center of the guitar face is mostly just wet, with maybe a slight hint of dye in some places.  The guitar will be allowed to dry for at least 24 hours before the brown is applied over the red, or else wiping the brown on could end up lifting red dye off the wood.

[dandyguitar]

Belle Gen. 5: Yet Another Finish Experiment

After my previous “final” finish experiment in September, I moved into a new house and had to pack everything up.  I only got back to work on the guitars recently and decided that although I am satisfied with the finish process I came up with, I was not happy with the colors of my final finish experiment.

So!  I bought some red dye and started experimenting again, looking for a color combination that will satisfy me.  The results:

Yellow-Red-Van Dyke Brown burst for the spalty guitar (similar to a Sunset Burst, Burbon Burst, or Violin Burst, depending on what company’s color names you’re looking at).

Clear-Burnt Sienna-Van Dyke Brown burst for the front of the curly guitar, Van Dyke Brown over Burnt Sienna on the back... The goal is to fade from the caramel colour of the roasted curly maple in the front to a rich walnut tone for the back and sides.

Both color samples go: hand rubbed dye, polymerized oil (to seal in color and add depth), shellac.

[dandyguitar]

Effect of brace shape on stiffness:weight ratio

The goal of braces/struts/reinforcements inside an acoustic instrument (violin, guitar, mandolin, etc.) is to add stiffness to the soundboard.  They must do so strategically: different parts of the soundboard need to be stiffer than others.  They must also do so efficiently: a brace must add as little mass to the soundboard as possible, because the instrument has a finite energy budget (vibrating strings) to produce sound with, and we don’t usually want to spend that energy moving needlessly heavy braces.

The geometry of a brace has a huge impact on its stiffness.  Consider, for example, the moment of inertia for a rectangle:

I = bh³/12

...where I is the moment of inertia, b is the base width, and h is the height.  Among its other mathematical uses, the moment of inertia is what determines how stiff a beam’s cross sectional shape is.  The bigger the number, the stiffer the beam.

If you double the width of the brace, you also double its stiffness and its weight.  This linear progression means that increasing the width of a brace does not improve its stiffness:weight ratio.  Yes, it improves the stiffness of the brace, but it increase its weight at the same rate.

If you double the height of a brace, though, the stiffness increases by a factor of 8 (2³) and the weight only doubles.  You have just improved its stiffness:weight ratio by a factor of 4!  (8 stiffness : 2 weight = 4:1.)  Different shapes have different formulae associated with them, but in general, a taller/thinner shape will exhibit a better stiffness:weight ratio.

Let’s put this to work!  I have modeled a variety of beams in AutoDesk Inventor, shown in the first image.  They all have the same width (because, as already established, stiffness:weight has a 1:1 relationship with width), the same length, and the same weight.  (Making them all the same weight will show us which shapes offer the most stiffness for a specific weight of material.)  The models also include an extra 1/8″ of material on their top face to represent the soundboard itself; the brace and soundboard are glued together in an instrument, so they form a single geometric entity.  The remainder of the soundboard is omitted to keep the simulation simple.

From left to right, the profiles are:

A. Tall rectangle

B. Flat rectangle (same proportions as A, but rotated 90°)

C. Tall rectangle with face rounded

D. Elliptical

E. Pointy triangle

F. Triangle with flattened top

G. I-beam

H. Sparse truss

I. Medium truss

J. Dense truss

K. Drilled rectangle

L. Hollowed brace, as seen the guitars of a dude who pissed me off and inspired me to write this post

For this stress analysis, each brace was fixed at one end and had an upward force applied to the other end to simulate string pull (or push, in the case of an arch top instrument).  The exaggerated displacement results can be seen in the second image.  The greater the displacement, the worse the stiffness:weight ratio.

If we take the reciprocal of the beams’ deflection values, we can determine their stiffness.  (In a beam bending application, stiffness and deflection follow an inverse relationship.)  We can then normalize those values, using the least stiff member as a base point to tell us the relative stiffness of all the other shapes.  The raw data and math are not included in this blog post, but here are the results, ranked from lowest to highest stiffness:weight ratio...

1.000 : 1 - B. Flat rectangle

4.597 : 1 - A. Tall rectangle

4.658 : 1 - C. Tall rectangle with bottom face rounded

5.405 : 1 - L. Hollowed brace

6.017 : 1 - D. Elliptical

8.599 : 1 - F. Triangle with flattened top

9.833 : 1 - E. Pointy triangle

11.005 : 1 - I-beam

11.543 : 1 - K. Drilled rectangle

13.358 : 1 - J. Dense truss

26.550 : 1 - I. Medium truss

45.191 : 1 - H. Sparse truss

Braces that are pretty easy to shape with simple woodworking tools (flat chisels, sandpaper, and a drill) are marked in italic text.  It should be pretty clear that wide, flat braces are pretty much useless for adding stiffness at a low weight.  Rounding the top of a tall, rectangular brace might make it look nicer than a plain rectangular brace, but it doesn’t improve the stiffness:weight ratio very much.  Hollowing out a brace, despite looking cool, removes so much material that it basically becomes two separate beams - it’s hardly worth the effort, only being about 10% stiffer than a simple tall rectangle.

Triangular profiles, I-beams (which can be further improved by making the middle section thinner and the entire brace taller, by the way), and rectangular braces drilled full of holes are the next step up.  They’re all about 2-2.5x stiffer than a simple tall rectangular brace, and aside from the I-beam (which can be hard to carve in a height-tapered brace), they’re all easy to implement with simple tools.

The trusses blow the beams out of the water, but they start getting pretty huge and spindly to get these kinds of stiffness.  It’s also worth noting that they are not beams any more at this point - they are trusses, which behave differently from a structural standpoint.  There is a reason that so many bridges and other large structures (sky cranes, radio towers, etc.) are built from trusses: they’re extremely efficient!

Final thoughts

If you’re extremely lazy, tall rectangular braces are the way to go.  If you’re less lazy, drilled rectangles are a good choice, but you won’t be able to shave those braces down very far once they’re installed, which is a disadvantage if you decide they’re too stiff and need to be adjusted.  If you’re slightly less lazy yet, triangles are easy to carve and extremely easy to shave down shorter to lighten up your bracing.  Hollowed braces are a lot of work for little to no gain, I-beams are frustrating to carve by hand, and high stiffness:weight trusses get so spindly that they’re difficult to implement in spruce (and almost impossible to lighten up without wrecking if you build your top too stiff).

[dandyguitar]

Rick-o-Jazz Concept

Okay.  So I also play bass.  And like most bass players, I love the way that Rickenbackers look... from a distance.  But my assessment quickly changes once they’re in my hands, and I’m NEVER happy with how they sound.  I’m a P/J player at heart and the tumpy sound of a Ric just doesn’t satisfy me.

So I decided that if I was going to ever have something as pretty as a Ric, I’d just have to build it myself.  So that’s what this is.  I won’t go into very many details except that it’s heavy on Gotoh and DiMarzio hardware, it’ll be made from pretty much the same woods as I almost always use, and I want to try carving the neck with a CNC router instead of by hand.

It’s still in the design stage right now, but that’s almost finished.  I’ve had the wood seasoning for a few months already, and I’m almost ready to put a neck blank together and start making fixtures and templates.  ETA?  Who knows?  Not before the snow falls, but definitely before it melts.

2024 stealth edit: a-hahaha no.  Body and neck blanks put together?  Accomplished.  Start shaping wood?  Not for close to a decade.

[dandyguitar]

Belle Gen. 5: Final Finish Experiment

After almost two dozen experiments, I think I finally have my entire process worked out.  This sample is made from the same woods as the real instruments and has realistic (if simplistic) carves and contours to work around for practice.  It’s not quite finished yet.  This is where it’s at in the process:

Wash-coat edges of body with 1-lb cut of shellac > Dye body > Oil maple top > Seal entire body with padded on orange shellac > Grain fill with black grain filler > Scuff off excess grain filler > Seal entire body again as before > I am here > Brush on shellac until pores are completely filled and finish is thick enough to be sanded flat without burning through (about 10-15 coats at 1-2 coats per day) > Sand flat with 400 grit > French polish for fine surface appearance > Polish/buff as necessary

Sorry for potato quality.  iPhone photos in bad light.

[dandyguitar]

And now for something completely different

A question was posed.  What do the frequency responses of different humbucker wiring methods (series coils, parallel coils, single coil) look like?  So I simulated a PAF humbucker (8k DCR, 4H inductance, 500pF capacitance per coil) with volume and tone controls into a typical Fender style tube amp input stage.

This is the result:

Not surprisingly, the series coils (standard humbucker wiring) have the highest output.  The resonant peak is about +3 dB at about 3.3 kHz (green trace on the graph).

A single coil (”coil tapped” or “coil split” as many guitarists would say), having half as much coil generating power, is half as loud.  The resonant peak is both higher in intensity and frequency: +6 dB at 3.9 kHz, producing its characteristically brighter tone (blue trace on the graph).

Two coils in parallel have the lowest output (half the output of the single coil, or a quarter of the input of the series humbucker) and the highest resonant peak of the bunch: +9 dB at 4.4 kHz (red trace on the graph).  “But there are two coils generating power!” you say.  Yes: two coils are producing roughly the same voltage in parallel.  Unfortuntately, voltage sources do not add in parallel - they only add in series.  Thus the voltage output from each coil doesn’t combine to anything larger than a single coil.  So what effect does that second coil have that reduces the pickup’s output compared to the single coil?

First, ignore one of the coils as a voltage source and look at how it’s connected: it’s in parallel with the other coil between the instrument output and ground.  In other words, each coil is loaded down by the other coil, which is just a large inductor.  The signal coil’s impedance forms a voltage divider with the loading coil and results in about 6 dB signal loss.  Thus, the parallel coils have the lowest output of the bunch (but are remain noise-free like the series mode).

[dandyguitar]

Belle Gen. 5: Finish experiments

I’ve spent the last two weeks trying out a variety of finish experiments on a variety of wood samples, trying to find the best way to finish these guitars. 

Here is a summary of the problems I am trying to solve:

- Ash is boring and must be colored to be interesting.  Color should ideally be transparent so that wood grain is still visible: I just want to change the color of the wood rather than mask it completely.

- Color should be even across the ash, rather than splotchy or uneven due to wood grain direction.

- Maple figure should have a high degree of chatoyance (movement of color in the finish when viewed from various angles) rather than being over-dyed, resulting in a static, dead-looking finish.

- Ash is highly porous and requires filling for a smooth finish.

- Sprayed finishes are not an option.

Here is a summary of the experiments I have conducted and my findings:

- Aniline dye colors ash and maple nicely.  Dissolving it in alcohol results in a more uniform color on the wood, while dissolving it in water results in sharper contrast between growth lines.

- Completely unsealed wood absorbs dye more readily in end-grain, such as around the edges of the guitar body.  Giving the end grain a wash coat of shellac before dying the wood greatly evened out dye penetration, but sealing the wood with glue size (a more traditional strategy) was almost completely ineffective.

- A coat of oil on paper turns the paper translucent, just as Dr Nick demonstrated to Homer when he was dangerously underweight.  It seems to do the same thing to the surface grains of wood, which enhances chatoyance.  Highly figured maple tops should therefore be dyed (if necessary for color) and oiled before being top-coated.  Polymerized oils are preferable for their extremely superior drying properties and (usually) their lack of added chemical thinners and dryers, which could react in unforeseen ways with subsequent coats of finish.  It is interesting to note that while figured maple benefited almost universally from being sealed with a coat of oil, ash became blotchy and uneven when oiled (compared to being sealed with shellac).

- Filling the ash pores via traditional French Polish methods (wash coat of shellac to act as a binder, then apply alcohol and pumice to abrade the surface and fill the pores) is only appropriate for dye-less scenarios: although very effective at filling pores and leaving a smooth, highly burnished surface that’s ready for a top-coat, it prevented dye penetration if performed before the color coats and it scrubbed most of the color off if performed afterward.

- Filling the ash pores with water-based grain filler is effective if done properly.  First: if the wood is not properly sealed first, the pigments in the filler will stain the wood, but if the wood is sealed too heavily, the filler will not lodge in the pores as effectively.  Second: most of the excess filler must be cleaned off the surface before it dries, and doing this tends to rub a little of the filler out of the pores at the same time, so it takes a few applications to get a mostly-level surface.  Third: any thin layer of filler left on the surface must be lightly sanded off (400 grit works well) before any top coats can be applied, but it must be sanded off very carefully to avoid cutting through the sealer coat and into the dyed wood below!  Ugh.

- Because the grain filler doesn’t leave a perfectly flat surface (there are still very shallow divots in the grain), clear finish has to be applied rather heavily to finish filling the pores and sanded back to flatten out the finish before final top-coating and polishing takes place.

- Shellac is the only effective finish that combines all the following requirements: a) easy and non-toxic to work with in a home environment (oils and water-based finishes meet this requirement, but fail either b or c), b) highly resistant to environmental changes (oil is terrible for this, despite common woodworking myths), and c) forgiving to work with and highly repairable (water based finishes don’t bond together fully between layers and you can see lines in the finish if you burn through one layer into another while you’re sanding; it’s also very difficult to repair).  Shellac can also be applied by wiping/rubbing (as with French Polish), brushing, or spraying.

- Shellac goes on smooth but thin when applied by French Polish techniques, or goes on smooth-ish and thick-er when applied by brush.  Because shellac can be sanded flat and worked into again with perfect adhesion between layers, it is beneficial to use a brush for bodying coats, sand the work flat after building up some finish, and then complete the finish by traditional French Polish techniques (for their glassy appearance).  Spraying is also effective but I’m not willing to invest in spray equipment.

So after extensive testing and experimentation, the method for finishing the guitar bodies will be:

1. Seal end-grain of swamp ash body parts with a wash coat of ultra-blond shellac.

2. Apply color to bodies with hand-rubbed aniline dyes dissolved in water.

3. Apply a coat or two of polymerized drying oils to the maple caps only.

4. Seal the entire body with wiped-on orange or garnet shellac.

5. Fill the ash pores with water-based grain filler, let dry, scuff off excess.  Repeat as necessary until as smooth as possible.

6. Brush on 10 coats of orange or garnet shellac.

7. Block sand with 400-600 grit paper to level the finish without burning through to the dyed wood below.

8. French polish to take out all the sanding marks and shine up the body.

Man.  This is too damn complicated.  I might have to outsource sunburst finishing for future instruments, because I know I’ll never invest in my own spray booth.

[dandyguitar]

Belle Gen. 5: Body sanding

Sanding the body is not much different from sanding the neck.  It uses the same sanding blocks and assorted tools.  The hand-carved contours are evened out with sanding blocks and 100 grit paper, then all remaining tooling marks are sanded out with sanding blocks and 150 grit.  Then the surfaces are smoothed (and the corners softened) with sanding blocks and 180, 220, and 320 grit sandpaper, with the body being lightly misted with water between each grit to raise the grain.

The control, pickup, and neck cavities are not sanded because they will not be visible in the finished product, and some (like the neck pocket) have tight geometric tolerances that must be maintained.

Speaking of geometric tolerances in the neck pockets, they ended up being a little oversize (as is typical once the neck has been sanded smooth).  To compensate, I made 1/64″ thick veneer from swamp ash and used it to shim the sides of the neck pockets (visible in left photo).  The necks don’t fight tight into the pocket (because this can cause the sides of the pocket to crack if the body shrinks in bad weather), but the gap is minimal and aesthetically pleasing.

With the necks and bodies fully sanded, it’s time to start working on finish samples!  Experimenting with finishes could take several weeks, so stay tuned: I guarantee there will be more to see before these guitars are complete.