For Those Who Struggle with Traditional IRs

You might want to reference the laser interferometry studies done by Polk Audio as they researched how to make speakers more highly pistonic in motion and thus more linear in nature. In fact, speakers that are mostly pistonic over their operating range are rare, and the larger they are, the rarer they are. Since guitar speakers are most often 12" speakers, you can expect good pistonic motion to be the exception rather than the rule. Annular and radial (flower petal mode) breakup is far more common than is pistonic movement among speakers that are operating near their power limit. This can be demonstrated as easily as applying a light dusting of flour to the cone of a speaker that is facing upwards in the vertical direction and running it. (So as not to cause the flour to spill.)
It's true that speakers aren't pistonic. They can never be, as long as the speed of sound in paper is finite.
 
Well, as the video shows, yes, pistonic motion is possible but the speaker has to be engineered for it, and that pistonic motion is likely to be restricted to a specific frequency range. And power range as well. Its rigidity is only going to hold until you put a certain amount of energy into the cone.
 
My issue with IRs is that they're a two dimensional snapshot of a 3 dimensional phenomenon. If you take IRs at different volume levels, you get different responses. Speaker behavior is NOT linear over the full power handling range of the speaker.


A properly detailed IR would involve a series of sweeps taken at calibrated levels from threshold of audibility to the maximum output point, which, if we were to use the industry standard measurements as employed by the audio industry for sound reinforcement applications, is the 1 dB compression point, which is where the actual speaker output lags behind an input level change by 1 dB.


And then, the IR curves would be analyzed and formulas derived to express the non-linearities. Those formulas would be embedded in the IR and processed by equipment which is built to handle them, thus yielding a dynamically accurate IR.


There are not less than nine non-linearity factors embedded in the design of any conventional cone-and-magnet dynamic speaker. You can't express its behavior to a high degree of accuracy with a simple frequency sweep no matter how high the resolution is.

We need an advanced, more 3 dimensional type of IR. At least I think so.
Quite a few people have asked me how cabinet volume affects an IR. I was curious and performed a test where I captured an IR with the cabinet quiet enough to have a conversation over, an IR at my normal volume, and an IR with the cabinet at an excruciatingly loud volume that made me nervous about potential speaker damage. For the biggest chance of hearing the difference, here's a clip comparing the "Quiet" IR to the "Loud" IR switching in real time to hear the differences between the two extremes without any audio gaps. Aside from the cabinet's volume, the only other adjustment made was the mic pre's level so that both examples hit my DAW's input at the same level.




Personally, I don't hear a difference big enough to justify capturing the same mic placement at various volume levels, and the difference you may be hearing probably has more to do with how hard the mic pre is being driven rather than how loud the cabinet is. I have no bias one way or the other on how loud to run a cabinet for the most accurate capture and I hope this comparison is helpful to anyone curious about it.
 
What I'm proposing is a means of capturing the dynamic cone breakup behavior, or more properly its harmonic effects, which will be different at high volumes as compared to low volumes.

It stands to reason that the harmonic content generate by cone breakup modes will be greater when the cone is being driven harder, thus exciting those break modes more strongly.

To cite an example, the cone rasp that is part and parcel of the tone of your classic Celestion Greenback only seems to appear when the speaker is being pushed hard. I've experienced it for myself. And it's not always about the power amp being pushed to clipping, as it happens whether the amp is a 25 watt amp or a 500 watt amp running at 25 watts.
 
What I'm proposing is a means of capturing the dynamic cone breakup behavior, or more properly its harmonic effects, which will be different at high volumes as compared to low volumes.

It stands to reason that the harmonic content generate by cone breakup modes will be greater when the cone is being driven harder, thus exciting those break modes more strongly.

To cite an example, the cone rasp that is part and parcel of the tone of your classic Celestion Greenback only seems to appear when the speaker is being pushed hard. I've experienced it for myself. And it's not always about the power amp being pushed to clipping, as it happens whether the amp is a 25 watt amp or a 500 watt amp running at 25 watts.
Did you hear any of those elements in the clip I posted? They would have been audible when the “loud” IR was selected. I heard more of a difference in the low end from more room reflections than anything noticeable in the top end. An artist I’m working with is used to hearing his Greenback loaded cab with a cranked Plexi. He told me he and three of his most trusted guitar playing friends couldn’t hear the difference between the real cranked cab and the IR. He also said he prefers using the IR because he can push his amp even harder and doesn’t get the nasty cone cry (undesirable sound from driving a speaker too hard) he gets with the real cab.

Just saying, a speaker isn’t the dynamic element some people may think it is. A lot of those dynamics come from the amp portion of the chain. Like you said, it happens with 25w amp as well as a higher powered amp knocked down to 25w… it’s still a 25w power section being pushed that’s causing what you’re hearing. Thankfully, I get the same power amp results using a real Marshall compared to a Fractal model. Try turning up the master volume a little higher and see if helps in getting those attributes you mentioned.
 
I respectfully disagree with the notion that you're hearing the power amp being pushed when the power amp is abundantly overpowered for the task, although it does make sense if the amp exhibits non-linear behavior within the power range in question. So say it's the 20, 25, or 30 watt rating of specific flavors of Greenbacks.

The test, then, is simple: In an anechoic chamber or at least well acoustically controlled environment, run clean sine wave sweeps of the speaker in question, using a highly linear amplifier that is in no danger of stepping outside of its most linear range of operation, and be sure that the sweep is not too rapid, giving the cone the time required to fully develop its breakup modes which will be excited at certain frequencies, and which may have a high Q and thus won't be excited if the sweep passes too quickly. Measure and plot the harmonic spectrum that results from the sweep, using a suitable laboratory grade microphone and measurement setup.

Start the sweep tests at the low threshold of audibility, capture the results, repeat the test 10 dB higher, and continue to repeat the tests at higher levels until the speaker reaches the 1dB compression point which defines its practical output limitation.

I think that you will find that if you do these tests as described, you will find that cone breakup modes have a substantial effect on the overall sonic signature of the speaker and it is dynamic across the full usable power range.

My belief is that it makes enough of a difference to warrant the development of a new type of IR that takes these dynamic behaviors into account.

The method I describe, which is beyond my ability to set up and test at this time, would validate or repudiate my hypothesis. There would be no need, and no room for, a mere opinion if you can perform this test sequence. I would be very interested to learn if I'm right, or wrong, or somewhere in between.

If I am right, then my suggestion and the resulting inquiry, and possible new developments in IR technology, would stand as my small but hopefully useful contribution to the quest of making modelling technology more faithful to the originals being modelled.

If I'm wrong, then I learn something else. That's never a bad thing. Learning should always be regarded as positive.
 
  • Like
Reactions: Rex
Well, as the video shows, yes, pistonic motion is possible but the speaker has to be engineered for it, and that pistonic motion is likely to be restricted to a specific frequency range. And power range as well. Its rigidity is only going to hold until you put a certain amount of energy into the cone.
Haven't watched the video yet, but does it talk about what kinds of problems exceeding the pistonic range of a speaker would cause?
 
I respectfully disagree with the notion that you're hearing the power amp being pushed when the power amp is abundantly overpowered for the task, although it does make sense if the amp exhibits non-linear behavior within the power range in question. So say it's the 20, 25, or 30 watt rating of specific flavors of Greenbacks.

The test, then, is simple: In an anechoic chamber or at least well acoustically controlled environment, run clean sine wave sweeps of the speaker in question, using a highly linear amplifier that is in no danger of stepping outside of its most linear range of operation, and be sure that the sweep is not too rapid, giving the cone the time required to fully develop its breakup modes which will be excited at certain frequencies, and which may have a high Q and thus won't be excited if the sweep passes too quickly. Measure and plot the harmonic spectrum that results from the sweep, using a suitable laboratory grade microphone and measurement setup.

Start the sweep tests at the low threshold of audibility, capture the results, repeat the test 10 dB higher, and continue to repeat the tests at higher levels until the speaker reaches the 1dB compression point which defines its practical output limitation.

I think that you will find that if you do these tests as described, you will find that cone breakup modes have a substantial effect on the overall sonic signature of the speaker and it is dynamic across the full usable power range.

My belief is that it makes enough of a difference to warrant the development of a new type of IR that takes these dynamic behaviors into account.

The method I describe, which is beyond my ability to set up and test at this time, would validate or repudiate my hypothesis. There would be no need, and no room for, a mere opinion if you can perform this test sequence. I would be very interested to learn if I'm right, or wrong, or somewhere in between.

If I am right, then my suggestion and the resulting inquiry, and possible new developments in IR technology, would stand as my small but hopefully useful contribution to the quest of making modelling technology more faithful to the originals being modelled.

If I'm wrong, then I learn something else. That's never a bad thing. Learning should always be regarded as positive.
What you're hearing in my test is similar, but not identical, to the situation you described. The cabinet was captured using long clean sine wave sweeps through a high-powered linear power amp in a heavily acoustically treated small room. The microphone was just a standard 57 since that's probably the most commonly used mic on a guitar cabinet. I only shot three IRs in my test, but figured comparing the quietest and loudest options side by side would show the biggest differences. The image below shows the response of the Quiet and Loud IRs overlapped to see where they differ. The blue line is the Quiet IR and the orange line is the Loud IR. I know some of the elements of my experiment are a little different than what you suggested, but hopefully it still offers something useful.

Quiet vs Loud Overlap.jpg
 
The effect of non-pistonic motion of the driver is simply the generation of harmonics that are not part of the input signal.

The shortcoming of the swept frequency plot as shown by York Audio is that it is cumulative in nature and does not represent the spectral harmonics of a driver at a single point in time. The Bode plot is not a good system for monitoring the harmonic spectrum at a specific point in time. For this application, a sonogram or spectrum analyzer operating in waterfall plot mode would be by far the superior tool.

One of the fundamental challenges of measurement is knowing what measurements to make to reveal the type of information you are looking for.

When looking for speaker generated harmonics, a Bode plot is not it.
 
Let me provide a simplified explanation of what I'm thinking, in a test and measurement approach.

Set up a speaker with a very linear amplifier and a clean sine source, and a calibrated microphone connected to a high resolution audio spectrum analyzer.
Apply a low level 1 KHz tone to the speaker and capture the microphone output via the spectrum analyzer. Record the 1 KHz level and the level of all significant harmonics above the noise floor.

Now repeat the test at higher levels. Step by whatever incremental level increases you wish, let's say 10 dB. Stop when you reach the 1 dB compression point, which may require a separate test run to determine.

At each level, record the ratio of harmonics to the primary signal. If the speaker's behavior is pistonic and does not cause it to generate additional/stronger harmonics as it is driven harder, then the ratio of the primary signal to the harmonics will remain the same.
Do this test not just at 1 KHz, but at other frequencies with the typical frequency response range of this driver as well.

By NOT doing this as as swept impulse, and using single tones, you get a simplified output that doesn't mask what we're looking for, which will tend to happen if you are making cumulative captures of a swept signal.

Maybe the simplest way of all to find the non-linear behavior is to set the signal to a high level but below the max power handling capacity of the speaker, and to sweep the signal from the lower to the upper limit, and watch the spectrum analyzer for that point at which the distortion products (the harmonics) are at their highest level relative to the primary signal. This should be the worst case frequency at which the speaker's behavior is least linear and it generates the most harmonic products.

The problem with a swept cumulative response plot is that as it sweeps, the higher level signals overwrite and obscure the lower level signatures. Only the peaks are preserved.
 
Last edited:
Please explain the measurement modes and methodology used to obtain this data.
I imported the IRs into REW and let it generate waterfall and spectrogram graphs. I did the test last year, so I’m just using what I have from that experiment. Again, it was made with a high-powered linear amplifier with long sine wave sweeps, as you mentioned in your first suggestion, ranging from human speaking volume up to the point of possible speaker failure. At the end of the day, if the difference isn’t overly apparent when listening to seamless clips, I don’t think the volume matters that much. You’ll get more variance from different capture techniques than from adjusting the volume level of the speaker cabinet.

Even if we saw a difference using a different microphone, the 57 still hears what it hears from 40Hz-15k, and there doesn’t seem to be a significant change when seen and heard between quiet and loud options.
 
OK. So there's that limitation....maybe there's some value in pursuing a new type of IR, one that is more like a speaker transfer characteristic rather than just a stimulus-response sweep, which I guess you can describe an IR to be.

I'm not in any position to do more than ask some questions that arise from my curiosity and natural desire to analyze everything.
 
...maybe there's some value in pursuing a new type of IR, one that is more like a speaker transfer characteristic rather than just a stimulus-response sweep, which I guess you can describe an IR to be.
It would have to be something more than an IR (Impulse Response). An IR represents the speaker's response to an impulse. By its nature, it cannot capture nonlinearities.
 
It would have to be something more than an IR (Impulse Response). An IR represents the speaker's response to an impulse. By its nature, it cannot capture nonlinearities.
This is exactly it.

@woodbutcher65, someone will have come up with a whole new way to capture both an impulse response and the non linear characteristics of the full path. This means creating new algorithms and ways to capture those non linear characteristics, plus a way to store them and recall them. This is no small task.

I think the way Fractal is currently doing it, works. They have the non linear parts in the speaker section of the amp block and the linear part remains in the IR. This way the normal IR does not have to change and no real redo is needed.

Maybe open a wish in the wish list to add some of the non linear parts you feel are missing from the Speaker tab and see if Cliff has a want, for adding it.
 
If you did the first pic you posted you wired that coil out of phase from the North coil. It should still give you proper resistance but, can sound phasey or like a chorus is turned on when both coils are active.

If you wire them to the second picture and you start with the south coil (always start with the active coil when wiring pickups (for future reference). If you rewire them I am guessing the guitar will sound like a whole different beast.


Keep me posted on how this goes. I bet the IR issue pretty much clears up as well. You might have to start checking IRs you never thought would sound good because they were harsh before. LOL... That would be a good place to be, and more like what I would expect.

Okay, I haven't been able to test with a ton of IR's but I think I found the answer, which I detailed in this thread: https://forum.fractalaudio.com/thre...frequency-instead-of-changing-pickups.192601/

I first tried the newer wiring posted from Awesome Guitars, and it was definitely wrong; it was cutting out a coil, or out of phase, or something; I kept taking off the pickguard and putting it back on to see if I had screwed up the wiring somehow, but it was right every time; it was just a totally wrong, probably out of phase wiring. It's definitely mislabeled. So I re-rewired with a modification of the old wiring, but with one coil going into the other going the other direction. I cannot tell at this point which is actually North and which is South, but the rewire did actually sound better. The thing was, I was so frustrated with that pickup and the mislabeling, and the enormous time I spent troubleshooting for no good reason, I just decided to try something else I had been considering for some time, and that's what I talk about in that thread.

The best pickup swap I've ever made was with my Washburn Trevor Rabin, where I replaced the stock JB/59 set with a set of Saturday Night Specials. I had tried those Saturday Night Specials in my Sun Valley Super Shredder before, but they just sounded too thin and harsh overall. But I thought about lowering their resonant frequencies to be pleasing with this guitar.

The immediate result was that, with the handful of IRs I grabbed, I was starting to get the effect that Leon Todd and Brett Kingman have in their videos, where they just grab an IR with no smoothing or proximity, and they get world-class tones right away.

So, my guess from this is that, when you cannot get any IR to sound just right without smoothing, it may be that your pickups' resonant frequencies are dissonant to your guitar's resonant frequencies. And if you tune your pickup to harmonize, if you will, to your guitar's resonant frequencies, the overtones that are strongest from the coupling of your body and neck, you can effectively remove the harshness from your guitar at the source, the relationship between the pickup and the guitar. It's not that the guitar is inherently harsh, in my total layman's theory, but that the resonant frequency of your pickup just clashes with it, and that is sooo easy to alter. I'll write more when I have a chance to experiment with many IRs to see how my newly "harmonically tuned" guitar reacts across the board.
 
Okay, I haven't been able to test with a ton of IR's but I think I found the answer, which I detailed in this thread: https://forum.fractalaudio.com/thre...frequency-instead-of-changing-pickups.192601/

I first tried the newer wiring posted from Awesome Guitars, and it was definitely wrong; it was cutting out a coil, or out of phase, or something; I kept taking off the pickguard and putting it back on to see if I had screwed up the wiring somehow, but it was right every time; it was just a totally wrong, probably out of phase wiring. It's definitely mislabeled. So I re-rewired with a modification of the old wiring, but with one coil going into the other going the other direction. I cannot tell at this point which is actually North and which is South, but the rewire did actually sound better. The thing was, I was so frustrated with that pickup and the mislabeling, and the enormous time I spent troubleshooting for no good reason, I just decided to try something else I had been considering for some time, and that's what I talk about in that thread.

The best pickup swap I've ever made was with my Washburn Trevor Rabin, where I replaced the stock JB/59 set with a set of Saturday Night Specials. I had tried those Saturday Night Specials in my Sun Valley Super Shredder before, but they just sounded too thin and harsh overall. But I thought about lowering their resonant frequencies to be pleasing with this guitar.

The immediate result was that, with the handful of IRs I grabbed, I was starting to get the effect that Leon Todd and Brett Kingman have in their videos, where they just grab an IR with no smoothing or proximity, and they get world-class tones right away.

So, my guess from this is that, when you cannot get any IR to sound just right without smoothing, it may be that your pickups' resonant frequencies are dissonant to your guitar's resonant frequencies. And if you tune your pickup to harmonize, if you will, to your guitar's resonant frequencies, the overtones that are strongest from the coupling of your body and neck, you can effectively remove the harshness from your guitar at the source, the relationship between the pickup and the guitar. It's not that the guitar is inherently harsh, in my total layman's theory, but that the resonant frequency of your pickup just clashes with it, and that is sooo easy to alter. I'll write more when I have a chance to experiment with many IRs to see how my newly "harmonically tuned" guitar reacts across the board.
Yes, and no.

The pickups need to match the guitar, you are right on about that. The guitar will tell you what fits, just with what it is producing out of the box. Too bright and harsh and you can't dial it out. Grab a pickup that has a lower resonance to match the most likely bright wood (maple top most times). Of if you have a mahogany guitar with dark pickups, you can put brighter resonant pickups in it. Like the JB/59 combo for instance.

Glad you figured out what was going on. I have a feeling those pickups were just an issue from the get go. The fact that they had either changed the wiring of newer versions of the pickups, or just messed up early on... Say a lot.

P.S. Using a vol pot as a variable resistance is a great idea.
 
Last edited:
Back
Top Bottom