Jump to content
pan_kara

Guitars and Numbers

Recommended Posts

I've already mentioned here and there that I'm going to try to do a comparison between the Hannes bridge and a no-name cheapo hardtail. The aim it to try to see if there is any measurable sound difference between the two.

Meet the test mule:

1490855_10202204664870557_1204537610_o.j

this is an alder body with an RG shape, currently wearing the Hannes bridge and a Lace Alumitone wired directly to the output jack. The neck is from my Nylon One build (I'm currently putting some TruOil on the top of that one and I took the neck off for that).

The idea is for the tests to be objective, I'm at the moment in the process of writing the numerical analysis code and checking first results.

If this works the plan is to try mounting different necks to the body (the body was in fact meant for that) to see if I can measure a sound difference there. If this is successful, there are many other things I'd imagine I could test - stainless vs normal frets, different nut materials, loose vs tight bolted on neck etc etc

I will be collecting ideas - assuming the first test gives me any meaningful results, so stay tuned for that.

Share this post


Link to post
Share on other sites

I object your honor. This would mean that we had science instead of Woodoo. Really bad for the industry. Stop this project at once! :lol:

  • Like 2

Share this post


Link to post
Share on other sites

Excellent, I prefer science over voodoo any day! I was the kid at school who alway had useless facts about nearly anything (except sport) pitty I didn't actually learn anything useful when I was at school!

Share this post


Link to post
Share on other sites

Ha ha!

I will post sound files for sure, to complement the numbers. Not sure about the bridge part, but when I get to comparing necks I plan to record the unamplified sound of the guitar with a mic. I'm wondering how the rather obvious differences in the unamplified tone of the different wood combinations etc carry over to the sound from the output jack - "95% of the sound is pickups", "70% tone comes from the neck, the rest from the body" and all that stuff.

Share this post


Link to post
Share on other sites

Just to let you know I havent been idle, I figured out which software to use (GNU Octave), learned to use it and wrote a bunch of scripts for analyzing wav files. I'm still establishing a baseline before I try do to comparisons of anything, but I think I'm close to having this part done.

Anyway, here's the frequency spectrum of a low F, you can nicely see how its overtone spectrum is modulated by the bridge pickup's pattern of frequency cancelations. I had some problems with the lace pickup and I have tons of 50 Hz hum for some reason (also visible on the plot), but I'll anyway be re-recording everything later..

spectrum1_zps2d08554d.png

and here's the time dependence of the first few harmonics from the same note

spectrum2_zps032e6293.png

I believe these plots will carry all the information I need. The slopes for each harmonic should be independent of

- pickup type and position

- picking strength (as long as I get no fret buzz)

- picking position (how close to the bridge etc)

I'm in the process of verifying this now.

  • Like 1

Share this post


Link to post
Share on other sites

I am almost scared of visiting CERN and sharing whisky with you. My observations will be short and inadequate!

This is great food for thought though. How does this feel from an absolutist scientific viewpoint? Correct and controlled?

**clap**

Share this post


Link to post
Share on other sites

I'm just trying to answer a few questions with the tools and methods that I happen to know, thats all. I was expecting someone would have done a similar thing long time ago, but so far I found nothing.

So far I'm seeing more or less what I've been expecting, when I'm not I'm checking where my method is wrong. Which is the classic road to observing what you wish to observe :P

  • Like 1

Share this post


Link to post
Share on other sites

as long as the science agrees with the voodoo its ok if it doesnt then its junk science.

+1 don't mess with my voodoo. Dinosaurs do not have feathers.

Share this post


Link to post
Share on other sites

Just to let you know I havent been idle, I figured out which software to use (GNU Octave), learned to use it and wrote a bunch of scripts for analyzing wav files. I'm still establishing a baseline before I try do to comparisons of anything, but I think I'm close to having this part done.

Anyway, here's the frequency spectrum of a low F, you can nicely see how its overtone spectrum is modulated by the bridge pickup's pattern of frequency cancelations. I had some problems with the lace pickup and I have tons of 50 Hz hum for some reason (also visible on the plot), but I'll anyway be re-recording everything later..

spectrum1_zps2d08554d.png

and here's the time dependence of the first few harmonics from the same note

spectrum2_zps032e6293.png

I believe these plots will carry all the information I need. The slopes for each harmonic should be independent of

- pickup type and position

- picking strength (as long as I get no fret buzz)

- picking position (how close to the bridge etc)

I'm in the process of verifying this now.

I like what you are doing. However, with your three hypotheses, there is a little to be said. A guitar string being a solid material is subject to some non-linearities when it comes to picking strength. This can affect the shape of the slope of the harmonics because the damping characteristics at the beginning of the note will not be the same at the end.

Also, while it is true that position will not affect the slope, when you are at a position that has a lower amplitude of a certain harmonic, your effective SNR will be low, so you may see unusually short decay times. Same can be said for pickup type, as certain types may emphasize or de-emphasize the presence of certain harmonics.

However, on the whole I think you have a very good approach. Octave is a good choice. I use matlab all day every day doing acoustics work, so I know it well and love it dearly :) One thing you can do for your 50 Hz hum if you can't get rid of it is to do something like a 6th order Bessel high pass filter on the signal (Bessel is ideal because of its maximal phase linearity and small overshoot, tradeoff is that it has a slower slope, so you need more orders). Or you could do a linear phase FIR, but be careful about how you choose it. Good job and keep the results coming!

  • Like 1

Share this post


Link to post
Share on other sites

thanks Ripthorn for the comments. Octave is indeed proving to be good for the job. I asked around and got a bunch of suggestions from people, but I'm glad I settled on this one. It was trivial to learn and works very well.

For the 50Hz hum... I'm having a problem with the wiring of the alumitones. I had just one in the bridge, wired straight to the output jack. Now I've added another one in the neck position (and a 5-way switch) plus a mini-toggle for the coil split. For whatever reason the hum is gone when in split mode, which is ok for me since I decided that I prefer this mode - the first dip (and the following ones) in the freq response of the pickup is shifted higher in this case. So I'll be doing the tests in split mode, but I don't understand why the humbucking mode has more hum.. the wiring of these pickups is strange, and there's little documentation online.

If you look at my previous plots you can see some of the harmonics have higher background than others - these are the ones polluted by the 50hz hum - 177, 349. Now I get my decay times by fitting a straight line - if I did that in these cases I would get incorrect results since the constant hum would bias the fit. So I need to cut off the signal below a certain amplitude - this is the case in these plots in fact, but dut to the 50 Hz hum the background still can be seen in the ones I just mentioned.

BTW I've changed my method, I'm now using FIR filters to pick out the individual frequency bands and then I draw the envelope plots. This is much slower than just looking at FFT bins, but I can have more narrow filtering so the cross-talk between different harmonics is gone now.

I'm checking all the fits to make sure that they're fine. In split-coil mode I get very good results. For some frequencies I do have poor fits due to low initial amplitude (the pickup is filtering some frequencies out), but this doesn't give me shorter decay times, just less precise fits (I have a shorter time range to fit to). The 1049 Hz line in my previous plots is an example of that - the slope is lower than the previous ones, but it hits bottom after 5 seconds because the frequency is damped by the pickup.

This is clearly visible in the results since I'm collecting results from several plucks of the string and then averaging them, for these frequencies I'm having much more scattered results.

I'm combating this by combining results from the bridge and neck PUs - currently by recording them separately, but I plan to wire a stereo output so that I can record them simultaneously and then intelligently combine (probably by simply adding the amplitudes).

Overall its pretty clear by now that the pickup location does not affect the results, besides affecting the precision of the measurement for some of the harmonics. Otherwise the results are very nicely compatible.

Share this post


Link to post
Share on other sites

Short update:

I've established a basic workflow, recording a stereo wave file with the bridge and neck pickups simultaneously. Looking at the results it is very clear that the pickup choice (position) does not affect the measured slopes of the different harmonics of the strings - but using two pickups gives me the option of using the signal from one when the signal from the other is gone (low) due to phase cancellations on the string.

I've recorded a bunch of test samples with the Hannes bridge and swapped it for a cheap no-name with some very light metal saddles. I've recorded some samples with that - as soon as my PC finishes the analysis I'll be able to compare the two and get a first "real" result - see if there are any statistically significant differences between the bridges and what is the size of the differences compared to the precision of my measurements.

  • Like 1

Share this post


Link to post
Share on other sites

I've always been of the opinion that quality matters. Junk hardware makes for a more junkier instrument. That said, there's a lot to be said about perception. Whether Bridge A or Bridge B is truly any better in a perceptible manner is often impossible to quantify which is why these types of experiments are rewarding fun. Some people "feel" that something is better, and whether it is or not, that feeling often translates directly into how one feels about that instrument when playing it. Placebo effect? Certainly. Some people over-rate that voodoo though, and start marketing the sugar pills as a dead-cert way to make you into a tone god. Hell, cleaning your guitar often makes you play better even if there is no real difference between it clean and dirty. You just pour your feelings through.

I'll get my coat then.

Share this post


Link to post
Share on other sites

I've always been of the opinion that quality matters. Junk hardware makes for a more junkier instrument. That said, there's a lot to be said about perception.

That's precisely what the experiment tries to clarify, through scientific procedures... no perceptions are allowed in the geek area. XD

I think the difference in quality of materials will not derive directly in the quality of the sound in an quick experiment, but on durability through daily use... so what is important in this case is to know how long the good sound stays before get worse due to material degradation of components, in a long term.

Of course the steel bridge is always the best option because it will keep the consistence of the whole instrument more time than the one made with plastic saddles (and probably will sound much better) but cheap does not always mean low quality.

I was wondering about the strings... do you use a new set for every test?

Share this post


Link to post
Share on other sites

Completely agree with you guys. The point of this fun experiment is to see whether some of the perceived (or not) tonal differences between XXX and YYY factors in guitar building can be numerically quantified. We all know that if we replace the saddles or the nut with rubber blocks the guitar will sound dead. If we replace steel saddles with graphtech, with brass, with whatever crap this last bridge uses.. there will be a difference. Question is can we hear it and can we measure it. I am recording both single note test signals that I then use to do the analysis and just clips of me fiddling around on the guitar. I'll later run the latter through some amp sim to see if the bridges also sound different, in addition to having different harmonic decay rater.

Because yes, they ARE different. I only looked at E string data so far, the cheap bridge sustains measurably better on the 3rd harmonic and worse on the higher ones beyond the 7th or so (IIRC).

So far I'm using the same set of elixirs, I had good experience with them keeping consistent tone through time.. as a proof of principle I should probably put the hannes back on and see if the results from before get reproduced... I might do that later on, actually.

Concerning the "cleaning guitar to have it play better".. I was thinking about this a bit in the past, possibly when trying to rationalize the purchase of a Mayones neck-thru 7 :P

Its true that when you listen to a recording it may be impossible to tell if the guitar that's playing is a private stock PRS costing as much as a car or a chineese LP copy (with good pickups swapped in :P).. but there might be a difference in WHAT is played. In other words with how the player connects with the instrument.

I made myself a Strat with a one piece flame maple neck which has some insane figuring.. I keep admiring the wood every time I play the guitar. Does it sound any different? Probably not. Does it make me play any different? Probably yes... :)

Share this post


Link to post
Share on other sites

I think it does. I'm not a great player but when an instrument makes me feel "on" rather than "inadequate", there is a positive benefit. Whether that be less tension, confidence or whatever, who knows? Either way, my "players" are still only tools that need fingers and a mind to make them work. They do coddle the player and make him play better some days I think. :-)

Share this post


Link to post
Share on other sites

On the subject of experiments, I am building a single pickup explorer because I want to see what a dedicated tone machine does, but I would be interested to see you take data of the bridge pickup with and without the neck pickup installed and see if there are any changes in the harmonic decay. I know some guitarists who swear that the presence of a neck pickup has a noticeable effect, and I wouldn't know because I have only ever played two or three pickup guitars. Just a thought...

Share this post


Link to post
Share on other sites

I would love to see an analysis of the strength of harmonic overtones for a string slowly being raised in pitch to ascertain how relatively high or low tension affects tone. Essentially, how loose is too loose and when does tension become so high it is tonally too sterile and tight? Might put pay to a lot of the "gauge = tone" debaters if the "tension window for tone" is wide.

Hell, Billy Gibbons and EVH use 8s and SRV used suspension bridge cabling.

  • Like 2

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


×