Official Luthiers Forum!

Thanks for coming by Sotos, and as said earlier, I certainly commend you for doing this - it's a lot of effort put forward purely for the sake of sharing knowledge, which is admirable.

Still, if you are welcome to critique, I offer these thoughts to consider. From my experience in attempts to accurately replicate identical string stimulus in sequential tests, as well as many others who have embarked on similar tests, I will propose this thought -

If you were to repeat this test with a few small changes, I believe it could be very revealing.

First step would be to do the test blind, meaning that the player must be kept ignorant as to which nut/saddle were in place at any time. Then would be to repeat the installation and playing of each material twice. This of course doubles the already heavy investment in time, but I think is essential to have more than one sample of each.

Next would be to make the test randomized. Put each nut/saddle package in a hat, and have one person randomly draw and install a set before passing it to the player. They then put them back in the hat, randomly draw another set, and when a given set has been used twice they are set aside. Keeping records (secret from the player), this is repeated until the hat is empty and each set has been sampled twice.

Then the tracks can be reviewed and separated in to two sets which contain one sample of each. One set is labeled 1-5, and the other A-E, randomly ordered and kept secret from listeners.

Then as a listener, your job would be to match samples 1-5, with their corresponding match in group A-E.

My proposal is that you would find it difficult, if not impossible, to match the samples up with results any better than statistically random. This would not indicate that the materials make no difference at all, but rather that the minor inconsistencies in playing were great enough to outweigh any changes which may have been caused by the materials alone.

This would in turn mean that the samples could not be of valid use as demonstration of differences in materials, because the differences listeners perceive are more a result of the differences in playing at each stage, than from the materials themselves.

So again, excellently organized and produced video, but I believe it may underestimate the complexities in controlled testing. Before you can demonstrate effects as results of a particular changed component, you first must reliably demonstrate that your methods accurately represent effects influence only (or at least primarily) by the component in question. Reliability of the test methods must first be established, or else there is no reason to believe the test results to be reliable.

_________________
Eschew obfuscation, espouse elucidation.

You are spot on Steve! ; )

https://youtu.be/zJoyLEOEGQM

I always enjoyed this clip.

_________________
Wake up and smell the rosewood!

I often find myself saying that (whatever) doesn't matter much most of the time, but when it matters it can matter a lot. The image is of 007 tied in a chair, with the water in the room rising. It doesn't matter how deep the water gets until it reaches his nose.... ;(

A good example of this is swapping out tuners. It turns out that there's a 'neck' resonance that can, when it's tuned just right, couple with the 'main air' resonance, and when it does it can change the sound of the guitar quite noticeably. Since the headstock moves a lot when the neck vibrates, the mass out there has a big effect on the pitch. Usually that resonance is too low to couple well, but with a short, stiff neck and a light head you can sometimes get it to work: it's more common to see this on Flamenco guitars with pegs than anything else. At any rate, the pitch match has to be pretty exact for the effect to come into play, so when things are 'close' fairly small changes in the head mass can alter the timbre of the guitar a fair amount.

So; somebody posts on a group like this saying that he swapped out his tuners, and the sound of the guitar improved (or otherwise) a lot. Somebody else tries it and it doesn't work for them, so they come back on to say that the whole notion is bunk. The OP defends himself, other folks criticize one or the other of the people posting, and it all descends into acrimony. They're both right in terms of their own guitar: in the end, where you end up depends very strongly on where you started out. Since every guitar is a little bit different you can't expect any two random instruments, even (and maybe especially) two of the same make and model to work in exactly the same way.

Add to that the fact that our senses are generally most sensitive to small changes at the threshold of perception; just the sort of thing that is hard to measure with microphones and such.

So, yeah, many of the things we disCUSS on these lists are 'small' details, and as such may not be important. OTOH, the difference between 'good' and 'great' can often be in the details. But, again, it's like 007's plight; most of the time any one particular detail probably won't make a noticeable difference, but that one time out of a hundred when it works can be really nice. Those of us doing the research are trying to figure out better ways of knowing when a particular intervention is likely to be useful, so we don't waste as much time fooling around. Given the complexity of the things, we're still a long way from where we'd like to be.

Several thoughts here, and all are prefaced with my enormous respect for your research, Alan.

#1 How do we know that the breaking strength of magnet wire is consistent within a percent or so? In my experience of engineering materials and cross sections, calling it to 200-300% is calling it pretty close.

#2 What force, and what angle of attack are we talking about? I agree that "break angle" is mostly a misunderstood concept. But I hear differences on guitars with shallow break angles when they are plucked very forcefully near the bridge at an angle nearly parallel to the top. The extreme case is playing a very loud harmonic: you can hear the string moving across the saddle, and restoring its position. The obvious solution is a notched saddle. I'm not aware of any downside to that. John Gilbert's individual steel notched saddle pins may have contributed to his reputation.

#3 No sort of recording and digital analysis (or recording and playback) will yield meaningful results. One thing that acoustic research has demonstrated is the ability of the human ear to distinguish between "Live and Memorex". An acoustic instrument, played live, generates a sound field that can't be captured by microphones, or reproduced by speakers. Why? Because an instrument radiates its sound directionally in ways that are distinctive. The only "played back" sound that has fooled educated listeners is the sound of a violin "played back" through a "slave violin", used as a speaker.

I got the breaking strength measurements myself, using a jury-rigged setup involving a triple-beam balance. The results were quite consistent over a few different runs, though. I've also been told by other folks in the business that this is the most consistent method, with one researcher saying it's good to within a dB or so. Basically, the quality control for that stuff is really tight, simply because it would be impossible to make it if the material was at all inconsistent.

My tests of break angle involved using the wire break to 'pluck' the string such that it was moving vertically with respect to the soundboard of the guitar. With no (initial) crosswise motion there was no chance for the string to roll across the top of the saddle. This is known to be a problem, and I have no data to set a limit on how much break angle you need to avoid it. That has to be one of the things that determines how much break is 'enough'. The point of my study was that once you get past 'enough', more doesn't help.

" No sort of recording and digital analysis (or recording and playback) will yield meaningful results."

OTOH, we already know that no sort of live playing will yield meaningful results either. Joe Curtin pointed out that you can't learn anything about a violin by handing it to a good player, because part of the competence of a good player is that they will be able to get the sound they want from any reasonable instrument. By the same token, you can't tell by handing it to a poor player, because they can't get a good sound out of anything. There's a feedback loop that causes players to alter the way they play different instruments, and you can't get them to not do it. That insight prompted him to work with Gabriel Weinrich to devise the 'inverse bow' test, which is similar in may respects to the 'slave violin' that you mention. It might well be possible to adapt such a method for guitar tests, that would allow for 'live' comparisons using 'canned', and therefore controllable, drivers.

OTOH, while we all agree that current methods using recordings are less than perfect, is that a good reason to abandon them? The listening tests I conducted using recorded plucks to look at break angle and string height off the top gave clear and consistent results that fit well with the way the guitar works. Granted, the subtlety of a live performance was missing, but do we always need that? Let's be careful not to make the perfect the enemy of the good, while, at the same time, acknowledging that the results we have are provisional. I'm going to stick by my data until somebody comes up with something better, and I hope they do.

I listened a couple of times, particularly to the second section. I felt like the more natural materials sounded best, i.e. the fossil ivory, the bone and the graphite. I wasn't all that enthused over the Tusq. To my ear, (and of course this is subjective) the bone had the most warmth and clarity, while the graphite had more subtle nuances and seemed to help overtones. The fossil ivory to my ear was similar to the bone, but perhaps less warm. And none of this matters because nobody else really understands my terminology in the same way that I do. This was by far the best comparison that I've seen though, and my eyes stayed pretty fixed on the signal graph in the upper left corner, trying to get a sense on how the were different. I would love to see a set of fixed snapshots of the signal overlaid with all the materials for that given moment, color coded of course, to see which one was which, and to help determine what it was that we liked. We do this at our office when designing speakers. We do sweeps of the drivers to get a sense of what their spectrums are, and then load them into a few slightly different cabinets and do more testing and comparison of the graphs. Eventually we figure out which ones give us the desired results and continue from there.

_________________
"I want to know what kind of pickups Vince Gill uses in his Tele, because if I had those, as good of a player as I am, I'm sure I could make it sound like that.
Only badly."