|
Post by antigua on Jul 26, 2017 13:13:37 GMT -5
Here is a version with the amplitude reduced. Does this still appear as a square wave? hb_wiring_sound_test_reduced_dB.wav (353.04 KB) If it does appear as a square wave in anyone's DAW, I'd appreciate a screen shot so that I might infer from it how my DAW differs from yours. Also, I have to mention that by looking at the wave form in its entirety, some of the mystery can be solved from that evaluation, so in looking at it, your impartiality will be somewhat reduced. Therefore in the screen shots above, I only displayed a portion of the wave form.
|
|
|
Post by lunaalta on Jul 26, 2017 13:23:16 GMT -5
Reducing that same sound sample won't take you back to pre-normalisation, whichever kind of normalisation is used. An fresh, unchanged signal would be ideal?
And your input path is....?
|
|
|
Post by antigua on Jul 26, 2017 13:40:59 GMT -5
Reducing that same sound sample won't take you back to pre-normalisation, whichever kind of normalisation is used. An fresh, unchanged signal would be ideal? And your input path is....? Look, it's a fair amount of work to go back and redo it. Can you post a screen shot of Audacity displaying the 6dB peaks, or square wave you're describing? Have you loaded that reduced amplitude wav and confirmed the problem exists there as well? At this point I'm still entirely taking your word for it, as the screen grabs I posted above show no clipping, and no peaks that exceed 0dB.
|
|
|
Post by lunaalta on Jul 26, 2017 14:08:36 GMT -5
Of course the problem remains with your file, even when you have reduced the level. There is no way you will remove the distortion in your file! Just as you cannot refocus an unfocused photograph....... The distortion came about during recording, I'm guessing. It took me 3 minutes to record a strummed G chord with my Strat! If you are gonna offer a test situation, the parameters for your recording should be laid out, i.e. how you recorded that sound sample. Simple, really..... Regardless of what is shown on any DAW, the sample is distorted. Sorry about that. How about explaining how it was recorded, type of input used, etc. How about stating the type of normalisation used..... Wikipedia explanation
Normalising to near 0dB can often introduce 'intersample effects' (red arrow), in which case, clipping will occur during reproduction. And so, distortion........ Your 0dB sample seems to be causing this effect, at the very least.
|
|
|
Post by antigua on Jul 26, 2017 14:25:11 GMT -5
Of course the problem remains with your file, even when you have reduced the level. There is no way you will remove the distortion in your file! Just as you cannot refocus an unfocused photograph....... The distortion came about during recording, I'm guessing. .. If your answer is that you can't distinguish between the three due to apparent technical shortcomings, I'll note your response as such. I can't go into it further than that at this time. This test took quite a bit longer than three minutes to conduct.
|
|
|
Post by lunaalta on Jul 26, 2017 14:32:42 GMT -5
LOL
Actually, there are differences heard, even on your distorted sample.
You give me the impression as someone who tried to use good practices when investigating. Perhaps you don't feel this exercise warrants that....
It appears I have hit a brick wall limiter..... ROTFL
I'm done with it. I have better things to waste my time on.
|
|
|
Post by antigua on Jul 26, 2017 14:33:45 GMT -5
Though I think the amount of clipping is minimal, it's common for electric guitars to encounter some amount of gain in situ. I think there is merit to electric guitar listening tests even with some gain present. Again, I do hear the difference in the audio sample as is, my question is merely whether or not anyone else hears it, having not been prompted with what they are listening for. To reiterate what this test is about; a neck humbucker has been recorded in parallel, and split to the coil closer to the bridge. The harmonic amplitudes (timbre) between the two wiring modes differs due to comb filtering. Parallel wiring removes or reduces harmonics are various frequencies, while these harmonics remain intact when wired in split mode. The strum of the split mode was copied, and then modified one harmonic at a time until it resembled the harmonic makeup of the parallel wired strum. So there are not three strums in the file 1) parallel strum, 2) split strum, and 3) split strum modified to resemble the parallel strum, but they are not necessarily presented in this order in this audio file: www.echoesofmars.com/misc/hb_wiring_sound_test.wav so that it can be seen whether anyone can tell the strums apart.
|
|
|
Post by JohnH on Jul 26, 2017 15:28:15 GMT -5
OK I had a go, with decent headphones.
I'm not hearing any of the distortion noted above, nor seeing in the waveform either. The first and second clip have one transient that gets very close but not actually clipped.
Listening, they are very similar, they could certainly all sub for each other in a playing situation.
I'm guessing the 2nd is the real parallel hb, the others I don't know. the 3rd has an unusual wavering of the wave form too with three clear humps.
|
|
|
Post by antigua on Jul 26, 2017 15:34:18 GMT -5
OK I had a go, with decent headphones. I'm not hearing any of the distortion noted above, nor seeing in the waveform either. The first and second clip have one transient that gets very close but not actually clipped. Listening, they are very similar, they could certainly all sub for each other in a playing situation. I'm guessing the 2nd is the real parallel hb, the others I don't know. the 3rd has an unusual wavering of the wave form too with three clear humps. Thanks for your feedback. Looking at the waveforms is sort cheating . When I post the answer I'll also post a graphic again showing which harmonics had to be modified.
|
|
|
Post by JohnH on Jul 26, 2017 16:01:28 GMT -5
Ive got some recordings of a bridge hb series, parallel and split that I did last weekdnd but havnt had time to make a post of yet. Your middle clip sounded slightlh more similar to my parallel one.
Another thing, definately cheating, and I only thought of it after shutting down tbe pc: Do we get a clue to the parallel hb by looking at exactly 60hz?
|
|
|
Post by antigua on Jul 26, 2017 16:29:26 GMT -5
Ive got some recordings of a bridge hb series, parallel and split that I did last weekdnd but havnt had time to make a post of yet. Your middle clip sounded slightlh more similar to my parallel one. Another thing, definately cheating, and I only thought of it after shutting down tbe pc: Do we get a clue to the parallel hb by looking at exactly 60hz? I didn't think about the 60Hz, that should be a tell if you see it. I suppose with saying why the waveform is a tell, it might be hard to intuit, but there's a semi-surprising fact about that which I'll mention when revealed. I would like to get my hands on your sound samples, just as a second point of contrast. I'd like to see if I can make one mode sound like another via "DAW magic".
|
|
|
Post by JohnH on Jul 27, 2017 7:55:18 GMT -5
Hee is my test, a Bridge humbucker with series, parallel and slug coil (neck side) www.jocidapark.com.au/circuits/strathbtest.wavIts the pickup in my avatar picture. This is the spectrum And the same calculated from your tests measurements for a PAF Pro, which is very similar to my pickup The recordings were done with a 10' cord into a buffer, then via a neutral mixer into the PC line-in and recorded with Audacity. The recordings and the test and calculated plots were all normalised for volume. Originally the series one was about 5db higher than the parallel, which was about 0.5 db more than the split. If you look around 4 to 6 kHz, the main differences are evident, mainly due to the RLC effects. The calculated plots were encouraging, showing quite good matches to reality, suggesting that combining theories of string vibration, electrical effects and Tillman is enough to capture the main effects and don't miss too much else that is significant.
|
|
|
Post by wgen on Jul 28, 2017 8:12:00 GMT -5
Though I think the amount of clipping is minimal, it's common for electric guitars to encounter some amount of gain in situ. I think there is merit to electric guitar listening tests even with some gain present. Again, I do hear the difference in the audio sample as is, my question is merely whether or not anyone else hears it, having not been prompted with what they are listening for. To reiterate what this test is about; a neck humbucker has been recorded in parallel, and split to the coil closer to the bridge. The harmonic amplitudes (timbre) between the two wiring modes differs due to comb filtering. Parallel wiring removes or reduces harmonics are various frequencies, while these harmonics remain intact when wired in split mode. The strum of the split mode was copied, and then modified one harmonic at a time until it resembled the harmonic makeup of the parallel wired strum. So there are not three strums in the file 1) parallel strum, 2) split strum, and 3) split strum modified to resemble the parallel strum, but they are not necessarily presented in this order in this audio file: www.echoesofmars.com/misc/hb_wiring_sound_test.wav so that it can be seen whether anyone can tell the strums apart. Well, it's really difficult. Just after a straight push of the "play" button in the soundfile with headphones, I'll try: 1) split modified strum, 2) real split strum, 3) real parallel strum....but really it could be everything
|
|
|
Post by stratotarts on Jul 28, 2017 16:06:29 GMT -5
Just an update - I tried loading the sound file with the 3 strums in it again. I may have chosen the wrong option in Audacity. It looks fine now, I see no evidence of distortion. It's really easy to see clipping and there is none. I am listening on headphones and I don't hear any distortion either.
|
|
|
Post by antigua on Jul 29, 2017 14:47:28 GMT -5
wgen got it right, this first sample was split modified, second was split original, third was parallel. To clarify, the split is to the coil closer to the bridge. Here is a copy of the sound file with the split modified in the middle: www.echoesofmars.com/misc/par_split_mod_split_orig.wavWith the parallel and split modified side by side, you can really sense how alike they sound, BUT, here's the really cool thing, if you look at the amplitude waveforms at the top of the screen shot below, the split modified and split original still look almost exactly alike, while the parallel wave form looks fairly different. The harmonic modification was 90% removing or reducing harmonics, so the "split modified" wave form looks smaller than the "split original", but retains the overall shape. If you load the wav file linked above into Audacity, or whatever, you can see this for yourself. The reason, I suspect, that the waveform stays the same, is because most of the modified harmonics are of low amplitude, and removed and amplified in whole, retaining the same overall balance in the time dimension, but it goes to show that the timbres can be made to match while the amplitude by time remains mostly unaffected. The parallel waveform has a "lumpy" look to it, where as the split mode has a more noisy, linear appearance. I'll have to record some strums again and see if this is something that always happens, or something that just happened in this case. I composited three screen shots together, so that a taller vertical range could be seen: As was the case before, and as is suggested by Tillman described comb filtering, most of the differences are in the upper harmonics, with most of the obvious differences above 4kHz. To my ear, the parallel strum and the split modified to sound parallel are nearly indistinguishable. I'm not positive I hear any difference at all. My own conclusion is that the underlying difference between parallel and splitting to the coil closer to the bridge is comb filtering, resulting in numerous band stops throughout the audible content. Splitting to the neck coil probably results in two samples that a little easier to tell apart, because the neck coil by itself sits below the node of the 3rd harmonic node, reducing 3rd harmonic movement, relative to parallel mode, which receives 3rd harmonic movement via the coil that is closer to the bridge. Therefore, parallel mode is not strictly subtractive, at least not relative to the neck coil.
|
|
|
Post by JohnH on Jul 29, 2017 16:30:05 GMT -5
V interesting. Just to be clear, are these tests on a neck pickup? - I'm assuming so but I wasn't sure.
|
|
|
Post by antigua on Jul 29, 2017 16:37:58 GMT -5
V interesting. Just to be clear, are these tests on a neck pickup? - I'm assuming so but I wasn't sure. Yes, neck position of a Tele copy. It's this same setup, but with the humbucker assembled of course.
|
|
|
Post by JohnH on Jul 29, 2017 16:49:10 GMT -5
OK thanks. I expect a similar approach applied to a bridge humbucker would apply, but maybe starting with a different perceived difference between Parallel and split, with a different set of variations required due to the coil positions. Do my tests look like they might fit into the pattern? or are they anomalous? When I listened back to mine, of the raw signals as posted, the parallel one sounded more like a good SC substitute that the neck-side slug coil, which surprised me.
|
|
|
Post by antigua on Jul 29, 2017 17:21:24 GMT -5
OK thanks. I expect a similar approach applied to a bridge humbucker would apply, but maybe starting with a different perceived difference between Parallel and split, with a different set of variations required due to the coil positions. Do my tests look like they might fit into the pattern? or are they anomalous? When I listened back to mine, of the raw signals as posted, the parallel one sounded more like a good SC substitute that the neck-side slug coil, which surprised me. Since you played riffs, there are various notes that each produce a different spectrum of harmonics: I cranked up the contrast in order to make the lines easier to see. I see a couple interesting things. 1) If you look at the parallel vs slug coil, sections two and three, you can see omissions in parallel that are present in the neck side coil. There are some obvious differences at 3.2kHz, 3.4kHz, 4kHz, and 4.6kHz. Note that the same gaps exist in the series side. Comb filtering. 2) parallel mode has content reaching up to 8.2kHz where as the split mode cuts off at 7kHz. That could be due to RLC differences, but I think the real reason is that the bridge-side coil picked up the higher amplitudes in that frequency range, while in parallel, than did the neck side coil alone. And of course, you see the series mode take a dive a 4.4kHz. As far as thinking parallel sounded more single coil to your ears, I think it's because on a Strat of Tele, the pickup is angled, so it get some neck-side harmonics and some bridge-side harmonics, where as your neck side coil is all neck side. The takeaway, and I think this is become increasingly apparent, that "true single coil tone" has more to do with coil position and harmonic distribution, and less to do with RLC issues.
|
|
|
Post by wgen on Jul 30, 2017 14:02:25 GMT -5
Thank you very much for the explanation! I was still wondering an aspect that still remains a little obscure. Let's say, leave alone for a second the differences of harmonics above 4 kHz, which, chances are, would be cut from RLC electrical properties and speaker response, grid stoppers from a tube amp, Miller capacitance etc. Honestly I was expecting to find much more of a "notch" around 3 kHz from the test, between parallel and split pickups. I was expecting to see something like ms showed in the first page of this thread..which, if you use the Tillman demo with 0.2 apertures and a blank space of 0.7 in between those, would actually simulate the coils of a humbucker. Is it correct if I say that, even with the differences that you showed and that are clearly visible, actually there isn't that much of a "frequency notch" in that 3 kHz area? Also the testing of JohnH doesn't seem to show that notch it seems, between parallel and split
|
|
|
Post by antigua on Jul 30, 2017 16:24:55 GMT -5
Thank you very much for the explanation! I was still wondering an aspect that still remains a little obscure. Let's say, leave alone for a second the differences of harmonics above 4 kHz, which, chances are, would be cut from RLC electrical properties and speaker response, grid stoppers from a tube amp, Miller capacitance etc. Honestly I was expecting to find much more of a "notch" around 3 kHz from the test, between parallel and split pickups. I was expecting to see something like ms showed in the first page of this thread..which, if you use the Tillman demo with 0.2 apertures and a blank space of 0.7 in between those, would actually simulate the coils of a humbucker. Is it correct if I say that, even with the differences that you showed and that are clearly visible, actually there isn't that much of a "frequency notch" in that 3 kHz area? Also the testing of JohnH doesn't seem to show that notch it seems, between parallel and split When you calculate a comb filter with the Tillman tool, it's assuming an open string at a particular frequency, that you're able to set.Since I'm strumming a G chord, there are several overlapping sets of harmonics and notch filters all combining together at once. I think you can approximate a fretted note by setting the scale length value of the fret, but then you have to move the pickup to compensate, it would be tedious. IMO, the differences are easier to hear when strumming a chord, because it exposes multiple differences brought about by the comb filtering, rather than just a few you'd get from a single string, I bet there are particular notes or chords you could play that would emphasize comb filtering notches really well, but I don't have time to figure that all out. The fact that harmonics are periodically missing between one sample and another, and that the one sample sounds like the other when you duplicate the harmonics differences, is good enough to make the case that comb filtering accounts for the differences between split and parallel timbres. I did see comb filtering effects as low as 800Hz, which you can see it if you look carefully at the screen shot I posted with the dots representing where I modified the harmonics. It just happens to be that there are more difference the higher you climb in frequency, which fits with the Tillman description. I think it therefor stands to reason that where comb filtering is involved, a greater frequency response from the pickup and and amplification stages will help reveal those differences. It would be easy to demonstrate with a simple EQ filter, and then do a listening test, but there's no reason at all to believe that statement is untrue.
|
|
|
Post by antigua on Aug 1, 2017 0:14:21 GMT -5
I just set up a guitar with Seymour Duncan Triple Shots and PAF replicas, so I can test series/parallel/split slug/split screw real easy, and I checked to see if the peculiar shape of the overall wave form was specific to one coil or two, but I found that there was no relationship. It was somewhat random, probably having to do with how the strings are strummed.
|
|
|
Post by wgen on Aug 1, 2017 5:26:30 GMT -5
I just set up a guitar with Seymour Duncan Triple Shots and PAF replicas, so I can test series/parallel/split slug/split screw real easy, and I checked to see if the peculiar shape of the overall wave form was specific to one coil or two, but I found that there was no relationship. It was somewhat random, probably having to do with how the strings are strummed. Thank you! So it seems that the type of difference is very well hidden, from a "visual" point of view, but still it is audible between parallel and split, even if, that difference is really slight.
|
|
|
Post by antigua on Aug 1, 2017 13:49:33 GMT -5
I just set up a guitar with Seymour Duncan Triple Shots and PAF replicas, so I can test series/parallel/split slug/split screw real easy, and I checked to see if the peculiar shape of the overall wave form was specific to one coil or two, but I found that there was no relationship. It was somewhat random, probably having to do with how the strings are strummed. Thank you! So it seems that the type of difference is very well hidden, from a "visual" point of view, but still it is audible between parallel and split, even if, that difference is really slight. Just to be clear, the only difference I can find is harmonics proportionality. I have not found a difference in the time dimension (a differences that is only perceived as the strings ring out). In the screen shot above, the shape of the split and parallel wave forms were different, but I found upon further testing that it was not a consistent difference, it was just happenstance. I'm still keeping an open mind that parallel / split might show differences in the time spectrum, but there is no physics based reason to believe that there should be. One thing that seems to be true, from a psychoacoustics point of view, is that differences in harmonics and frequency response result in a perceived difference in the attack and decay. For example, an increased high frequency response gives the sense of a "glassy" attack. In other words, in our brains, a difference in the Y dimension manifests as a perceived difference in the X dimension.
|
|
|
Post by wgen on Sept 20, 2017 3:46:08 GMT -5
Sorry, I was reading again this thread, and I was looking more closely to your spectrum analysis of the humbucker pickup. I was wondering if the Y axis really shows the dB increments which are written there. I was especially wondering if, let's say, given the 1000 Hz in the spectrum, the 4000 Hz frequencies really go down about 10 dB in respect to the 1000 Hz mark. Thank you very much anyway
|
|
|
Post by JohnH on Sept 20, 2017 15:18:36 GMT -5
Sorry, I was reading again this thread, and I was looking more closely to your spectrum analysis of the humbucker pickup. I was wondering if the Y axis really shows the dB increments which are written there. I was especially wondering if, let's say, given the 1000 Hz in the spectrum, the 4000 Hz frequencies really go down about 10 dB in respect to the 1000 Hz mark. Thank you very much anyway The measured and calculated plots are similar in their general slopes and shapes, so I think they are both reasonably correct. The real plot had no filtering nor post processing applied and the calculated plot is just based on combining a few theories about the string, pickup and electrical performance, with no particular further adjustments.
|
|
|
Post by wgen on Sept 21, 2017 12:54:04 GMT -5
The measured and calculated plots are similar in their general slopes and shapes, so I think they are both reasonably correct. The real plot had no filtering nor post processing applied and the calculated plot is just based on combining a few theories about the string, pickup and electrical performance, with no particular further adjustments. Thank you for the reply. Now that's interesting...I was now thinking that this kind of frequency response, even if amplifiers and overdrives obviously must be considered into the equation, but still, this gives some important insights when you think about the mix of a band...for example the fact that a loud snare and cymbals of an acoustic drum might produce some overtones that really come out way too much in respect to the actual response of electric guitar and bass. I also found an old thread here on Guitarnuts where you showed the response of a piezo pickup, and I thought that it wasn't too dissimilar to the passive magnetic pickups analysed here...apart from a huge mid boost around 600 to 700-800 hz that the piezo seemed to have. You were working on a circuit to create a notch around those mids, in that occasion. The piezo also rolled off the higher harmonics above 1000 hz if I remember correctly, in a similar way to the response of the magnetic pickups.
|
|
|
Post by JohnH on Sept 21, 2017 16:51:48 GMT -5
When you take out the effect of guitar amp and speaker, you see more signal in the 4 to 8 khz range, as in those plots. But piezos have more again, with significant signals higher than 10khz. You hear it clearly in the transients due to strumming etc which are maybe under-represented in the measured plots and not captured in the calculated ones.
|
|
|
Post by wgen on Sept 22, 2017 3:01:58 GMT -5
When you take out the effect of guitar amp and speaker, you see more signal in the 4 to 8 khz range, as in those plots. But piezos have more again, with significant signals higher than 10khz. You hear it clearly in the transients due to strumming etc which are maybe under-represented in the measured plots and not captured in the calculated ones. Thank you again. I now have a question about the plots...so, they are not representing the transients, are they? From the analysis of Antigua it seems that higher harmonics tend to fall much sooner than lower harmonics and fundamentals...So I was now thinking which "moment" of the strum the plots are representing...hope this is clear enough, please excuse me for my language.
|
|
|
Post by antigua on Sept 22, 2017 13:11:18 GMT -5
An interesting aspect of the response curve of the pickup is that when people thing "EQ", they think overall sound, they don't think in terms of transient, but we can see that if a pickup cuts off at 2kHz, it sounds wildly different in the transient than a pickup that cuts off at 4kHz, because there is so much instantaneous content within that band, but in the decay, most of that content is gone, and so the two pickups sound very alike from there on out. Therefore, to a large degree, the response curve is defining the "attack" of the pickup, more than it's determining the overall tone.
|
|