|
|
Post by antigua on Jul 8, 2017 20:44:14 GMT -5
Adobe Audition has this ridiculously futuristic feature whereby you can load audio in spectrographic display, and then draw boxes over the spectrogram, and either hear the isolated audio within that frequency band (which is very interesting), or modify the amplitude of those specific frequencies, for that specific time frame. It's basically a graphical representation of an FFT filter that makes visual analysis and manipulation of audible content fun and easy. I've recorded a strummed and picked E chord on a Fender FMT HH with Seymour Duncan Pearly Gates pickups, using the neck pickup, and then I recorded the same with a Strat that has an SSL-5 in the neck position (producing a thick, dark, humbucker-like tone). In the sound clip below, it could be seen based on the spectrogram that the SSL-5 neck actually produced a lot more harmonic content than the PG neck (most especially above 1kHz), and since it's easier to subtract data than add it in, I decided to see how close the SSL-5 neck would sound to the PG neck by altering the amplitudes of the SSL-5 until it looked very similar to the spectrogram of the Pearly Gate neck, which is seen in the center of the graphic below: Here's is the sound file that is produced by the audio seen below.  Listen to the sound file, and observe how similar the manipulated SSL-5 neck (middle) and PG neck (beginning) are (or are not) to the non-modified SSL-5 (end). If you look at the spectrograms above, you might think the first two sound samples are the same, and the third is unique, but the first sample is actually unique, and the second sample is the third, having it's harmonic amplitudes carefully manipulated to make it closely match the first, as close as I could get it. In this case, the difference has less to do with the RLC values and more to do with the differences between the guitars themselves and the position of the pickup with respect to the string's length, and the comb filtering effects of the humbucker. According to Tillman, the comb filtering should be mostly north of 4kHz, but nevertheless, for whatever reason, I see lower amplitudes above 1kHz with the humbucker. This experiment could be set up better, for example a single guitar could be used instead of two different guitars, and two pickups could be found that have very close RLC values, so this experiment is just a quick sampler, what I can do with the time I have available to do it. It demonstrates what it demonstrates. Something that is obvious to some, but not all, is that if you were to ever to make a device that made one pickup position sound like another (neck -> bridge, bridge -> neck) based on harmonic manipulation, you wouldn't be able to apply a fixed EQ, because the harmonics you have to "knock out" or magnify / synthesize, would be different for each specific note. The device would have to "content aware" and decide which harmonics to alter based on the fundamental frequency they derive from. The bigger picture with this is that I'm trying to make some inroads into the mystery of why a PAF sounds "clear" with a low resonant peak, while a Strat pickup is said to sound "dark" if it shares that same resonant peak. For example, the HSS Strat is popular because a lot of guitarists thing a hot single coil, such as an SSL-5, in the bridge sounds inarticulate compared to a humbucker bridge. Some possible theories to toy around with are that knocking out harmonics by way of comb filtering results in a more "clear" sound, and that relates to this test. It could also be that the lower Q factor of a humbucker results in a softer treble response that comes across as "clear". There might be a human explanation: when people plug in a Les Paul, they turn the treble up on their amp, changing the tonal frame of reference, but when you load a hot pickup into the bridge of a Strat, it's often accompanied by bright, low output pickups in the neck and middle, so you set the treble on your amp to favor the neck and middle, making the bridge seem dark by contrast. Truth be told, I have this Strat with SSL-5s in all three spots, and if I turn of the treble on the amp, it's clean tone is perfectly satisfactory, very "piano like". |
|
|
|
Post by reTrEaD on Jul 8, 2017 22:56:04 GMT -5
Something that is obvious to some, but not all, is that if you were to ever to make a device that made one pickup position sound like another (neck -> bridge, bridge -> neck) based on harmonic manipulation, you wouldn't be able to apply a fixed EQ, because the harmonics you have to "knock out" or magnify / synthesize, would be different for each specific note. The device would have to "content aware" and decide which harmonics to alter based on the fundamental frequency they derive from. I would guess the 'content awareness' would need to go one step beyond being based on the fundamental frequency. For instance E4 (329.63 Hz) can be played different different ways in standard tuning. 1st string open 2nd string 5th fret 3rd string 9th fret 4th string 14th fret 5th string 19th fret 6th string 24th fret The sensing location of a given pickup will be at a different fraction of the vibrating length in each of those six examples. That's not to say it would be impossible to also assess the present harmonic content and adjust accordingly but that would seem to be a rather tall order. |
|
|
|
Post by Charlie Honkmeister on Jul 8, 2017 23:26:56 GMT -5
That's the Guitar Nuts spirit, Antigua!
Nice post! You touched on several good points and again, great persistence on experimenting.
Really, on the difference between a HB in the Strat bridge and a SC , I would think you're right in that the lower Q of the humbucker plays a role. The lower Q, (and not everybody thinks about this), also means the rolloff after the resonance peak is not as steep. So the highs aren't getting rolled off as fast and apparently the slightly higher level of harmonics in the 2 KHz to 3 Khz range, where the ear is most sensitive, even if they're only a couple of dB different, could maybe make the sound a bit more articulate for the humbucker as opposed to a "fat" SC. Don't know, just something to think about.
There might be an effect also with a tube amp or amp sim where maybe a bit more low and low-mid frequency content of the humbucker could generate a tiny bit more distortion in the same 2 to 3 KHz range and even above, even when playing so-called "clean," that gives a perception of articulation and presence. That's the principle of the Aphex Aural Exciter - it generates a bit of high harmonic distortion which gets mixed back into the signal, and the audible result is more bright presence and "aliveness."
I'm also suspecting that the presence peak on classic guitar speakers may also interact with some of this, but you're starting in the right place.
It would be very interesting to chase this further with both a real tube amp, and a good amp sim like Amplitube 3, and then see how things pan out on this using the whole chain.
-Charlie
|
|
|
|
Post by antigua on Jul 8, 2017 23:47:57 GMT -5
Something that is obvious to some, but not all, is that if you were to ever to make a device that made one pickup position sound like another (neck -> bridge, bridge -> neck) based on harmonic manipulation, you wouldn't be able to apply a fixed EQ, because the harmonics you have to "knock out" or magnify / synthesize, would be different for each specific note. The device would have to "content aware" and decide which harmonics to alter based on the fundamental frequency they derive from. I would guess the 'content awareness' would need to go one step beyond being based on the fundamental frequency. For instance E4 (329.63 Hz) can be played different different ways in standard tuning. 1st string open 2nd string 5th fret 3rd string 9th fret 4th string 14th fret 5th string 19th fret 6th string 24th fret The sensing location of a given pickup will be at a different fraction of the vibrating length in each of those six examples. That's not to say it would be impossible to also assess the present harmonic content and adjust accordingly but that would seem to be a rather tall order. I'm not real sure how the harmonic content varies by fret position, although it would be easy to figure out by recording it and looking at the spectrogram, as was done here. The remarkable thing, though, is that it apparently doesn't change that much. For example, the neck pickup sounds like the neck pickup, no matter where you play on the neck. The harmonic balance changes a bit, but the overall timbre is maintained. One thing that does change is the output level. When doing amplitude tests I noticed that playing a bar chord at the 7th or 12th fret produced a higher voltage than open chords. Very counter intuitive, but the explanation is that playing higher up the neck causes the pickups to become "centered" on the moving string where greater displacement occurs. Of course you lose sustain in the shorter length of string, so the notes are initially louder, but die off more quickly. |
|
|
|
Post by antigua on Jul 9, 2017 0:19:03 GMT -5
That's the Guitar Nuts spirit, Antigua! Nice post! You touched on several good points and again, great persistence on experimenting. Really, on the difference between a HB in the Strat bridge and a SC , I would think you're right in that the lower Q of the humbucker plays a role. The lower Q, (and not everybody thinks about this), also means the rolloff after the resonance peak is not as steep. So the highs aren't getting rolled off as fast and apparently the slightly higher level of harmonics in the 2 KHz to 3 Khz range, where the ear is most sensitive, even if they're only a couple of dB different, could maybe make the sound a bit more articulate for the humbucker as opposed to a "fat" SC. Don't know, just something to think about. There might be an effect also with a tube amp or amp sim where maybe a bit more low and low-mid frequency content of the humbucker could generate a tiny bit more distortion in the same 2 to 3 KHz range and even above, even when playing so-called "clean," that gives a perception of articulation and presence. That's the principle of the Aphex Aural Exciter - it generates a bit of high harmonic distortion which gets mixed back into the signal, and the audible result is more bright presence and "aliveness." I'm also suspecting that the presence peak on classic guitar speakers may also interact with some of this, but you're starting in the right place. It would be very interesting to chase this further with both a real tube amp, and a good amp sim like Amplitube 3, and then see how things pan out on this using the whole chain. -Charlie I'm leaning away from the lower Q playing a big role in humbucker "clarity", because if I attenuate the SSL-5 past 2kHz, to approximate the effect of eddy current attenuation, or even roll back on the tone control, I don't get a burst of clarity, I just get a slight loss in brilliance. Even though I tweaked the harmonics in the middle sample in about twenty places, if you look at the first and third sections, you can see that the most obvious difference is that the humbucker has much less amplitude above 1kHz. The 1kHz cutoff is almost exact, too, if you refer to the frequency ruler at the far right. Simply bringing down everything above 1kHz by about 6dB does to most towards making the SSL-5 sound like the humbucker. Although, you can see that the SSL-5 sample is a lot more harmonically dense, and that the harmonic emphasis differs in more ways that just by amplitude. For example, the humbucker has a rather strong harmonic at 2.31kHz, but the SSL-5 showed much less amplitude there. Even though Tillman's comb filtering suggests that the frequency cancellations should be occurring in around 4 to 5kHz for coils that are half an inch apart, this looks to me like comb filtering with emphasis that starts as low as 1kHz. I have more experiments in mind that can clarify these things more. One I'd like to do next is record a humbucker, then remove one of the coils, but keep the coil in circuit so that the RLC values remain mostly the same. One input is removed, and so to is any comb filtering caused by the dual coil configuration, which isolates differences owing to comb filtering, and eliminates nearly all other variables. |
|
|
|
Post by ms on Jul 9, 2017 6:28:50 GMT -5
In this case, the difference has less to do with the RLC values and more to do with the differences between the guitars themselves and the position of the pickup with respect to the string's length, and the comb filtering effects of the humbucker. According to Tillman, the comb filtering should be mostly north of 4kHz, but nevertheless, for whatever reason, I see lower amplitudes above 1kHz with the humbucker. Tillman's figure 13 for a humbucker with an assumed 2.5 inch aperture shows sows significant loss by 2 KHz on the low E string. Of course, it is not right since the aperture is wrong. The actual response would be obtained by using a pole width aperture on each of two closely spaced coils. |
|
|
|
Post by wgen on Jul 9, 2017 8:54:32 GMT -5
That's the Guitar Nuts spirit, Antigua! Nice post! You touched on several good points and again, great persistence on experimenting. Really, on the difference between a HB in the Strat bridge and a SC , I would think you're right in that the lower Q of the humbucker plays a role. The lower Q, (and not everybody thinks about this), also means the rolloff after the resonance peak is not as steep. So the highs aren't getting rolled off as fast and apparently the slightly higher level of harmonics in the 2 KHz to 3 Khz range, where the ear is most sensitive, even if they're only a couple of dB different, could maybe make the sound a bit more articulate for the humbucker as opposed to a "fat" SC. Don't know, just something to think about. There might be an effect also with a tube amp or amp sim where maybe a bit more low and low-mid frequency content of the humbucker could generate a tiny bit more distortion in the same 2 to 3 KHz range and even above, even when playing so-called "clean," that gives a perception of articulation and presence. That's the principle of the Aphex Aural Exciter - it generates a bit of high harmonic distortion which gets mixed back into the signal, and the audible result is more bright presence and "aliveness." I'm also suspecting that the presence peak on classic guitar speakers may also interact with some of this, but you're starting in the right place. It would be very interesting to chase this further with both a real tube amp, and a good amp sim like Amplitube 3, and then see how things pan out on this using the whole chain. -Charlie I'm leaning away from the lower Q playing a big role in humbucker "clarity", because if I attenuate the SSL-5 past 2kHz, to approximate the effect of eddy current attenuation, or even roll back on the tone control, I don't get a burst of clarity, I just get a slight loss in brilliance. Even though I tweaked the harmonics in the middle sample in about twenty places, if you look at the first and third sections, you can see that the most obvious difference is that the humbucker has much less amplitude above 1kHz. The 1kHz cutoff is almost exact, too, if you refer to the frequency ruler at the far right. Simply bringing down everything above 1kHz by about 6dB does to most towards making the SSL-5 sound like the humbucker. Although, you can see that the SSL-5 sample is a lot more harmonically dense, and that the harmonic emphasis differs in more ways that just by amplitude. For example, the humbucker has a rather strong harmonic at 2.31kHz, but the SSL-5 showed much less amplitude there. Even though Tillman's comb filtering suggests that the frequency cancellations should be occurring in around 4 to 5kHz for coils that are half an inch apart, this looks to me like comb filtering with emphasis that starts as low as 1kHz. I have more experiments in mind that can clarify these things more. One I'd like to do next is record a humbucker, then remove one of the coils, but keep the coil in circuit so that the RLC values remain mostly the same. One input is removed, and so to is any comb filtering caused by the dual coil configuration, which isolates differences owing to comb filtering, and eliminates nearly all other variables. As already ms pointed out, shouldn't a correct pickup demo with the Tillman model be done with about two 0.25 aperture widths, close to each other but with a blank space of about other 2 pole pieces of the same aperture of the first two, in between them? I think it was already been discussed from you in the past about the correct aperture widhts. It's been a quite long time I tried, but I seem to remember that with a humbucker in respect to a single coil in the same position overall, I got a huge comb filter around 3 kHz with the humbucker demo. Also, at 1 kHz the humbucker seemed to already drop around 2,3 decibels in respect to the single coil. But at 4 kHz, the humbucker seemed to "recover" from the comb filtering, so that at this frequency the response should be "flat", and this i think is very important because has much of the "attack" and "cut" of the tone, while still isn't rolled off from typical guitar speakers. Also, good bright paf usually maintain a flat response at 4 kHz, or, if they roll off, they roll of just a decibel...so they still should have some clarity, I mean, if compared to a Duncan Custom for example, which is an overwound type instead, and roll off around 6 decibels @ 4 kHz. I don't know if this makes sense, but a single coil with the same resonant peak and Q of a bright paf, could maybe have a more similar harmonic content if "scooped" around 2.5/3 kHz with a narrow q filtering Edit: when I tried this I was focusing to the bridge pickup position, not the neck one |
|
|
|
Post by reTrEaD on Jul 9, 2017 9:00:53 GMT -5
I have more experiments in mind that can clarify these things more. One I'd like to do next is record a humbucker, then remove one of the coils, but keep the coil in circuit so that the RLC values remain mostly the same. One input is removed, and so to is any comb filtering caused by the dual coil configuration, which isolates differences owing to comb filtering, and eliminates nearly all other variables. That seems a worthy experiment but which coil will remain in place? I would guess the results will be significantly different between leaving the 'bridgemost' coil in place vs leaving the other coil in place. Tangent:HSS vs SSS Fender Standard (Mexican) Stratocaster wiring is slightly different between these two but end up with much the same result. SSS selects tone controls by switch position. HSS has the tone controls wired in parallel with the Neck and Middle pickups. Looking at it in terms of loading (when all controls are max'd) we have a somewhat less desirable situation in regards to the loading when the SSS bridge pickup is selected alone. There is no tone control connected. In a situation where we are already severely altering the balance between fundamental and upper harmonics, that would seem contrary to what we would want. A HB with two coils in series would see a more 'traditional' load in that circumstance. Instead of two 500k pots (volume and tone) in parallel, it sees the 250k volume pot. I'm sure there's much more to this that just the loading, but I think it's a contributing factor. |
|
|
|
Post by stratotarts on Jul 9, 2017 9:30:52 GMT -5
That is an interesting listen. I do wonder what results you could achieve by global, simple equalization. For example, a parametric centered on the resonant frequency to bring down the Q of the SSL. I looked to see if you had the SSL-5 in your database, but it look like not. I'm not claiming that you could get such a close match, but I'm always looking for the simple way - even if it turns out to be a flop. I'll try playing with Audacity.
|
|
|
|
Post by ms on Jul 9, 2017 9:35:33 GMT -5
In this case, the difference has less to do with the RLC values and more to do with the differences between the guitars themselves and the position of the pickup with respect to the string's length, and the comb filtering effects of the humbucker. According to Tillman, the comb filtering should be mostly north of 4kHz, but nevertheless, for whatever reason, I see lower amplitudes above 1kHz with the humbucker. Tillman's figure 13 for a humbucker with an assumed 2.5 inch aperture shows sows significant loss by 2 KHz on the low E string. Of course, it is not right since the aperture is wrong. The actual response would be obtained by using a pole width aperture on each of two closely spaced coils. Here is what I predict for the hum bucker harmonic response (that is, the part due to coil location and aperture), two coils spaced by 11/16", each with an aperture the width of a pole piece. Of course, this is just one factor. I would expect most of the harmonic roll off to be the result of string stiffness (resistance to bending)  |
|
|
|
Post by stratotarts on Jul 9, 2017 10:35:20 GMT -5
I loaded the sound file into Audacity, did a spectral analysis and just used visual cues to develop an equalization curve:  Here is the result: composite_equalized.mp3 (92.71 KB) To clarify, I applied the equalization directly to part 3, the SSL-5, so all parts should sound similar. This was a quick hack 5 minute job, I think it could be improved. |
|
|
|
Post by reTrEaD on Jul 9, 2017 10:37:00 GMT -5
One thing that does change is the output level. When doing amplitude tests I noticed that playing a bar chord at the 7th or 12th fret produced a higher voltage than open chords. Very counter intuitive, but the explanation is that playing higher up the neck causes the pickups to become "centered" on the moving string where greater displacement occurs. Of course you lose sustain in the shorter length of string, so the notes are initially louder, but die off more quickly. Counter intuitive, because we expect the amplitude of displacement for a given input energy (picking) would be less as we move the termini closer together. You identified one of the mitigating factors. The relationship of the pickup to the location of maximum amplitude. The other would be the vertical proximity of the pickup to string. As we fret farther toward the bridge we not only change the horizontal location of the terminus, we also make a slight change in the 'height' of that terminus. The effect of that change in height is small, but significant in decreasing the gap between the pickup and string. |
|
|
|
Post by antigua on Jul 9, 2017 14:15:16 GMT -5
Tillman's calculator www.till.com/articles/PickupResponseDemo/ doesn't really elucidate much with respect to this scenario, which is why I didn't share this before, but I will now since it came up. Below is a neck humbucker spaced as my Fender FMT is spaced, and the apertures are set to half an inch, the width of the coil. You can see that fall off is not dramatic until 10kHz. If you change the "open string frequency" the curves change, but it mostly follows this pattern, whether it be 82 Hz or 330Hz for the low and high E strings.  Here it is again with pole piece sized apertures, you can see that it's mostly unchanged:  If you set it up to resemble the Strat, with one read point a 6.375" from the bridge, it indicates that there are similar valleys, but no drop off at 10kHz.  To my eye, it you look at the first and third sections, the two un-altered samples, the one on the left, the humbucker, shows evenly spaced gaps in the harmonics, where as the SSL-5 on the right shows dense harmonics and little apparent filtering. What's happening with the humbucker does appear to be periodic filtering beginning around 1kHz, but I can't seem to "curve match" that with the Tillman tool. |
|
|
|
Post by antigua on Jul 9, 2017 14:31:37 GMT -5
I have more experiments in mind that can clarify these things more. One I'd like to do next is record a humbucker, then remove one of the coils, but keep the coil in circuit so that the RLC values remain mostly the same. One input is removed, and so to is any comb filtering caused by the dual coil configuration, which isolates differences owing to comb filtering, and eliminates nearly all other variables. That seems a worthy experiment but which coil will remain in place? I would guess the results will be significantly different between leaving the 'bridgemost' coil in place vs leaving the other coil in place. Tangent:HSS vs SSS Fender Standard (Mexican) Stratocaster wiring is slightly different between these two but end up with much the same result. SSS selects tone controls by switch position. HSS has the tone controls wired in parallel with the Neck and Middle pickups. Looking at it in terms of loading (when all controls are max'd) we have a somewhat less desirable situation in regards to the loading when the SSS bridge pickup is selected alone. There is no tone control connected. In a situation where we are already severely altering the balance between fundamental and upper harmonics, that would seem contrary to what we would want. A HB with two coils in series would see a more 'traditional' load in that circumstance. Instead of two 500k pots (volume and tone) in parallel, it sees the 250k volume pot. I'm sure there's much more to this that just the loading, but I think it's a contributing factor. I can try removing both coils one at a time. I don't suspect it will make that much of a difference, but it's worth a try. Personally, I agree that a bridge humbucker sounds better than a HOT single coil, the likes of an SSL-5, I think the voicing is a lot different, going beyond differences in Q factor. From what I've observed here, I think it does have to do with harmonic omissions. This seems to be like "treble mud" where too much harmonic activity in the high end leads to a lack of clarity. It's a well known concept with regard to bass (you never hear anyone play a chord on a bass guitar) but it seems that it can also be true of treble. |
|
|
|
Post by antigua on Jul 9, 2017 14:41:03 GMT -5
I loaded the sound file into Audacity, did a spectral analysis and just used visual cues to develop an equalization curve: Here is the result: To clarify, I applied the equalization directly to part 3, the SSL-5, so all parts should sound similar. This was a quick hack 5 minute job, I think it could be improved. I should measure the SSL-5's, it's just a pita to open up and test Strat compared to most other guitars, and they're tapping pickups, so it adds to the complexity. I tester the SSL-4s, which are substantially similar. Do you hear anything noteworthy from having applied the equalization? I'm not sure what I'm listening for. I'm going to do that humbucker with coil removed test, ASAP, I think that will be some good data to work with. |
|
|
|
Post by wgen on Jul 9, 2017 15:41:42 GMT -5
Tillman's calculator www.till.com/articles/PickupResponseDemo/ doesn't really elucidate much with respect to this scenario, which is why I didn't share this before, but I will now since it came up. Below is a neck humbucker spaced as my Fender FMT is spaced, and the apertures are set to half an inch, the width of the coil. You can see that fall off is not dramatic until 10kHz. If you change the "open string frequency" the curves change, but it mostly follows this pattern, whether it be 82 Hz or 330Hz for the low and high E strings. Here it is again with pole piece sized apertures, you can see that it's mostly unchanged: If you set it up to resemble the Strat, with one read point a 6.375" from the bridge, it indicates that there are similar valleys, but no drop off at 10kHz. To my eye, it you look at the first and third sections, the two un-altered samples, the one on the left, the humbucker, shows evenly spaced gaps in the harmonics, where as the SSL-5 on the right shows dense harmonics and little apparent filtering. What's happening with the humbucker does appear to be periodic filtering beginning around 1kHz, but I can't seem to "curve match" that with the Tillman tool. Sorry, I really wouldn't like to distract you from other more appropriate testing, but, do you find the Tillman demo gives you the same results if you set 82 Hz as the fundamental, 0.2 aperture as you already did, but in the bridge position humbucker instead of neck? I'm saying this just because when I tried I'm pretty sure I used the demo program with these specifications and I got a scoop @around 3 kHz just like ms showed, maybe little less scoop than that, but very similar overall |
|
|
|
Post by JohnH on Jul 9, 2017 17:16:44 GMT -5
I think there are a few watchits when looking at Tillman plots.
They don't account for the relative harmonic balance within the vibrating string based on picking position. This creates an extra filering and comb effect. The plots presented have a basis of equal harmonic amplitudes which when you work through the maths, is equivalent to the super-thin tone picking like 5mm from the bridge. But this issue can be addressed (GuitarFreak includes it). What this means is you can look at a simple Tillman plot and compare it to another relatvely, but it will over-emphasise high harmonics- even without thinking of electrical interactions on top.
The maths does support the observation that output is higher if you fret higher. This is particularlg true for a bridge pickup where the sin term that determines the fundamental ampljtude is based on a greater angle as the pu to bridge distance becomes a greater % of the free length.
I work such things based on assigning a constant force to the picking as if a flexible pick will release consistently after bending a certain amount as it flicks past a string. I also assume all string tensions are equal (which is roughly true in a balanced set). From there, deflection and harmonics can be calculated, but it is still assumption.
|
|
|
|
Post by stratotarts on Jul 9, 2017 21:23:14 GMT -5
I loaded the sound file into Audacity, did a spectral analysis and just used visual cues to develop an equalization curve: Here is the result: To clarify, I applied the equalization directly to part 3, the SSL-5, so all parts should sound similar. This was a quick hack 5 minute job, I think it could be improved. I should measure the SSL-5's, it's just a pita to open up and test Strat compared to most other guitars, and they're tapping pickups, so it adds to the complexity. I tester the SSL-4s, which are substantially similar. Do you hear anything noteworthy from having applied the equalization? I'm not sure what I'm listening for. I'm going to do that humbucker with coil removed test, ASAP, I think that will be some good data to work with. Sorry, I should have been more clear and produced a better example. I made another one, this time it conforms to your format. The first part is the HB, the middle is the modified (equalized) SSL, and the end is the unmodified SSL. Listen to the first two parts to see whether you can tell the difference. What I hear (but other ears are definitely better qualified) is little difference between the first two, and a brighter sound in the last part. composite_equalized.mp3 (92.71 KB) |
|
Deleted
Deleted Member
Posts: 0
Likes:
|
Post by Deleted on Jul 10, 2017 7:20:32 GMT -5
One thing that does change is the output level. When doing amplitude tests I noticed that playing a bar chord at the 7th or 12th fret produced a higher voltage than open chords. Very counter intuitive, but the explanation is that playing higher up the neck causes the pickups to become "centered" on the moving string where greater displacement occurs. Of course you lose sustain in the shorter length of string, so the notes are initially louder, but die off more quickly. The higher the fret the shorter the height of the string from the pup's pole. Might be from this. |
|
|
|
Post by stratotarts on Jul 10, 2017 8:25:23 GMT -5
Okay, here's some fun... I compared the spectra: (I reached the attachment limit, will post later) I created three separate sound files, ABC. I flipped a coin to determine the assignment of names. The three files are as follows: The challenge, of course, is to identify them! pickupA.mp3 (31.08 KB) pickupB.mp3 (31.48 KB) pickupC.mp3 (31.48 KB) |
|
|
|
Post by stratotarts on Jul 10, 2017 8:32:06 GMT -5
PG vs. SSL-4 spectral response:  Compensation filter:  |
|
frankfalbo
Meter Reader 1st Class
Posts: 74 
Likes: 1
|
Post by frankfalbo on Jul 10, 2017 12:01:44 GMT -5
One thing that does change is the output level. When doing amplitude tests I noticed that playing a bar chord at the 7th or 12th fret produced a higher voltage than open chords. Very counter intuitive, but the explanation is that playing higher up the neck causes the pickups to become "centered" on the moving string where greater displacement occurs. Of course you lose sustain in the shorter length of string, so the notes are initially louder, but die off more quickly. The higher the fret the shorter the height of the string from the pup's pole. Might be from this. Yep I suggested that back when he first noticed this phenomenon in this thread: guitarnuts2.proboards.com/post/81021/thread
It'd be pretty easy to test just by lowering the pickup to match the distance when barred vs open chords. But it's got to be pretty minuscule. |
|
|
|
Post by antigua on Jul 10, 2017 12:57:31 GMT -5
Okay, here's some fun... I compared the spectra: (I reached the attachment limit, will post later) I created three separate sound files, ABC. I flipped a coin to determine the assignment of names. The three files are as follows: The challenge, of course, is to identify them! I didn't record the PAF and the SSL-5 perfectly the same unfortunately, so you can kinda figure some it out on that basis, but that aside, A must be the PAF, B is EQ filtered and C is original -1dB. I think the knocking down of particular harmonics changes the timbre in a way that a flat EQ doesn't, because the comb filtering is "dynamic" insofar as it follows the note, whereas flat EQ filtering is not dynamic in that respect, it squishes everything at a given frequency. Isn't harmonic proportionality the basis for timbre in general? Isn't that how an electric organ can be made to sound like various instrument? The idea that these differences are centered on comb filtering would certainly explain why guitarists often have a hard time nailing down how certain changes effect the tone, because they're not necessarily changing the EQ, they're changing the voice, and that's a much more difficult thing to characterize. Words like "quack" and "piano" come up - words that describe voicing, not EQ. You can't really describe a voice, you have to hear it to know it. People say "doing this increases the highs and tightens the lows" it's like they're using EQ lingo in an attempt to describe changes in harmonic proportionality that only loosely tracks with the EQ. All of the pickup height adjustment testing I've done, from the earliest stuff last year with a '59 to the more complicated stuff a few months ago with an eBow suggest that changing the pickup height changes the harmonic balance, more than it changes the EQ, and the most recent testing suggests that magnetic string pull is responsible for shifting the harmonic makeup, by magnetically attracting the anti-nodes that are closest to the magnetic field. People say raising and lowering the pickup makes the bass more or less tight, or it makes the mids more complex, whatever, the truth is subtly changes the timbre of the signal in a way that almost defies description. Though I don't fully understand it, I'm putting stock in JohnH's claim that picking location might be responsible for bringing the filtering down to the 1kHz range, because if true, we'll pretty much have an open and shut case. There would still be a lot of worthwhile testing to do even if that explains what's happening, for example, do the SSL-5 and the PAF sound more alike if you pluck very close to the bridge? Do the harmonics align more closely under those circumstances? Is any divergence pushed to a higher frequency? |
|
|
|
Post by stratotarts on Jul 10, 2017 17:02:07 GMT -5
Okay, here's some fun... I compared the spectra: (I reached the attachment limit, will post later) I created three separate sound files, ABC. I flipped a coin to determine the assignment of names. The three files are as follows: The challenge, of course, is to identify them! I didn't record the PAF and the SSL-5 perfectly the same unfortunately, so you can kinda figure some it out on that basis, but that aside, A must be the PAF, B is EQ filtered and C is original -1dB. I think the knocking down of particular harmonics changes the timbre in a way that a flat EQ doesn't, because the comb filtering is "dynamic" insofar as it follows the note, whereas flat EQ filtering is not dynamic in that respect, it squishes everything at a given frequency. Isn't harmonic proportionality the basis for timbre in general? Isn't that how an electric organ can be made to sound like various instrument? The idea that these differences are centered on comb filtering would certainly explain why guitarists often have a hard time nailing down how certain changes effect the tone, because they're not necessarily changing the EQ, they're changing the voice, and that's a much more difficult thing to characterize. Words like "quack" and "piano" come up - words that describe voicing, not EQ. You can't really describe a voice, you have to hear it to know it. People say "doing this increases the highs and tightens the lows" it's like they're using EQ lingo in an attempt to describe changes in harmonic proportionality that only loosely tracks with the EQ. All of the pickup height adjustment testing I've done, from the earliest stuff last year with a '59 to the more complicated stuff a few months ago with an eBow suggest that changing the pickup height changes the harmonic balance, more than it changes the EQ, and the most recent testing suggests that magnetic string pull is responsible for shifting the harmonic makeup, by magnetically attracting the anti-nodes that are closest to the magnetic field. People say raising and lowering the pickup makes the bass more or less tight, or it makes the mids more complex, whatever, the truth is subtly changes the timbre of the signal in a way that almost defies description. Though I don't fully understand it, I'm putting stock in JohnH's claim that picking location might be responsible for bringing the filtering down to the 1kHz range, because if true, we'll pretty much have an open and shut case. There would still be a lot of worthwhile testing to do even if that explains what's happening, for example, do the SSL-5 and the PAF sound more alike if you pluck very close to the bridge? Do the harmonics align more closely under those circumstances? Is any divergence pushed to a higher frequency? ...and we have a winner! I guess that was too easy. I was surprised how similar the unmodified sounds were. Myself, I can hear that there is a difference but I can't convince myself that I can identify which is which. I have noted the testing you did re harmonic balance. Harmonic proportionality is still affected by an evenly distributed equalization like a low pass, on a per note basis. Of course, in such case the proportionality does not track with the fundamental and that is the difference. But in this case the signal is the product of both the string and pickup characteristics. If you compared the same pickups again with a different string like a D, maybe it would be possible to make a good guess as to what is due to the string position and what is due to the bare pickup response (or at least the part of it that is not dependent on string characteristics). I'm interested in taking details and trying to derive useful simplifications that might fit with some kind of simple circuit mod or pickup design feature. Your adjustments to the signal were many, if I understand correctly. One interesting part of that, is what rules you followed, that could be automated. Those would have to be known in order to use a DSP to process the sound, for example. There has always been the problem with guitar, of separating the signals from the strings so that the algorithms can work properly. But hexaphonic pickups have proven to be much trickier to build than normal ones. This supports a growing suspicion in the back of my mind, that the humbucking design does have a distinct character of its own, whereas you could consider a single coil design as actually having less character or colouration, at least if it is constructed and loaded in a certain way. I'm trying to lean toward a design that offers maximum flexibility and so that character of a humbucker (which I think we are attributing to phase cancellations here) is a liability. The same character is a benefit not a liability if you are actually seeking a classic sound character that already exists - e.g. a PAF sound. It just irks me because I'm also a stickler for noise reduction, which bring a level of complexity to the table when single coils are required. Again, from my point of view, and with my jaded ears, the difference is not as much as I expected. It could explain why my Firebird pickups don't sound quite the same as singles in spite of many similarities that they have in common - magnetic poles, no steel, low losses resulting in very high loaded Q. |
|
|
|
Post by antigua on Jul 10, 2017 23:04:26 GMT -5
I'm trying to lean toward a design that offers maximum flexibility and so that character of a humbucker (which I think we are attributing to phase cancellations here) is a liability. The same character is a benefit not a liability if you are actually seeking a classic sound character that already exists - e.g. a PAF sound. It just irks me because I'm also a stickler for noise reduction, which bring a level of complexity to the table when single coils are required. If you want humbucking and minimal comb filtering, I think the solution exists in tiny humbuckers, such as Bardens. I'll be the first to admit I wouldn't mistake a Barden for Fender set, but that's a mystery unto itself. Same for Lace Sensors; if it's not comb filtering that make them sound different, then what is it? The stronger magnetic pull of the AlNiCo pole pieces? |
|
|
|
Post by antigua on Jul 10, 2017 23:25:57 GMT -5
Here is another harmonic manipulation demo. I have a Les Paul with a Tonerider AlNiCo 4 PAF replica with coil splitting. The beauty of this is that in series, you will get comb filtering, but in split mode you won't, so you have an apples to apples test. The only drawback is that series mode attenuates at a much lower frequency. I have some Triple Shot rings on hand, I will install some humbuckers with those so that I can contrast parallel / split, and that will allow the creation of some spectrograms that extend up to 6 or 7kHz. So this sound sample is four strums: 1) Neck in series 2) Neck split, harmonically manipulated to match the neck in series 3) Neck split, attenuated above 4kHz, to match the EQ profile of the neck in series, individual harmonics are unchanged 4) Neck split, no modifications The split mode puts out a lower voltage, so I normalized both strums separately to make them even amplitude. Sound file: www.echoesofmars.com/misc/hb_neck_strum_composite.wav In the interest on fidelity, this is a 2.7 meg wav, in case you're on a dial up.  I've put little white dots to indicate where most of the harmonic differences existed that had to be tweaked to get a match. You can see visually that above 2.8kHz the differences become more extreme, but below 2.8kHz the differences are more subtle. If you open the images separately, it's much easier to see. I'll reserve commentary about what I hear until others have had a chance to listen to the sound sample. The 4kHz attenuation seen in the series sound sample is a lot higher than the 2.5kHz I measure when using the 470pF test load, but I'm using a lower capacitance Roland guitar cable that's only about 320pF, whoch should put the peak around 3.5kHz, and then the drop off slope is somewhat gradual past the peak for the lower Q factor, so I think that's why we're seeing the voltage drop off at a seemingly high 4kHz. |
|
|
|
Post by antigua on Jul 11, 2017 0:26:31 GMT -5
Here is the same test as above, but this time I strummed and plucked only 20mm away from the bridge, which causes the neck pickup to sound remarkably similar to the bridge pickup. Sound file: www.echoesofmars.com/misc/hb_ser_split_bridge_strum_composite.wav In agreement with JohnH's observation, it seems that plucking near the bridge caused the harmonic differences between the series and split sound sample to shift upwards. With neck strums, the deviation began around 2.8kHz and continued up from there, but with the bridge strumming, the harmonics are similar up to about 3.2kHz, then the differences become more obvious. You can see it visually in how the harmonic lines in first two sound samples look "thinned out" relative to the that last two sound sample, above 2.8kHz in the first test, but above 3.2kHz in the second test. Since these strums include the harmonic content of multiple string overlapping at once (mostly evidenced by the pulsating lines), it would be tedious to map particular harmonics to a particular fundamental, so I might try this again with single notes, to see what that yields. |
|
|
|
Post by antigua on Jul 11, 2017 1:36:37 GMT -5
reTrEaD mentioned on page 1 the fact that a note will have different harmonics depending on where along the fret board it's played, here is a 330Hz "E" played in five strings, "pluck, pluck, pluck..." In order: Open high E string, 5th fret B string, 9th fret G string, 14th fret D string, 19th fret A string:  You can see that the voltage hits a high mark in the 19th fret A string. It appears that the harmonics trend downwards as you move up the neck, with comb filtering effects creating different harmonic gaps for each string and fret. |
|
|
|
Post by JohnH on Jul 11, 2017 2:57:47 GMT -5
Apologies..i'm just having a problem understanding what tbe adobe audition plots are really showing. What is on each axis?
|
|
|
|
Post by antigua on Jul 11, 2017 3:06:05 GMT -5
Apologies..i'm just having a problem understanding what tbe adobe audition plots are really showing. What is on each axis? They're sound recordings, so the X axis is time, and there are two Y axis, the smaller display at the top is overall amplitude of the wav file on the y axis, and then the larger display below is a spectrogram, where the Y axis is frequnecy, and the amplitude of the signal at that frequency is indicated as "heat", so the blues are lower amplitudes, and the red/orange/yellow indicates higher amplitudes, so it's effectively three dimensions. All of the horizontal lines are harmonics, the lowest one is the fundamental. A really nifty thing is that you see some of those lines "pulse", they are not solid, and that means that two or more strings are producing that same harmonic at the same time, causing phase cancelling, leading to the pulses, or beating. If you look at the "E" note test, only one string was plucked at a time, so you see only solid harmonics, and no beating. This is a really effective display method, I can probably redo some of the earlier tests this way and get a more coherent result. |
|
