Speaker Impedance Curve: Why is this in the Amp Block instead of the Cab block?

[SinglecutGuy]

This is just something about which I'm curious. It's likely due to not understanding the concept, but I would think that the IR in the cab block should carry information along with it, that decides what the impedance of that particular cab is, and therefor the amplifier chosen in the amp block should/will react to the cab/IR in that fashion. Or have the option of different impedances of specific cabs as different IR's.

The part that confuses me, is why we'd have say, a limited 60+ impedance curves to chose from, relevant to "specific speaker cab" models, only to then add an actual Cab/IR to that amp, from 1,000's of actual IR's, where that relationship is not authentic? And when I said "limited" that is purely based on the difference between the number of Impedance Curves in the Axe, compared to the number of IR's in the Axe itself, much less what you can purchase from 3rd party vendors.

Am I correct that it's not possible in the real world to mimic the impedance curve of a Morgan 1x12 cab with say an actual 4x12 Recto cab? Nor do I understand why you'd want to mix and match those selections, rather than take the native impedance curve of the actual cabinet for a more "real" example.


Again, this is likely due to me not understanding the concepts, but I figured I'd ask.

 

[Dave Merrill]

It's modeling the effect of the speaker's impedance curve on the power amp.

Yes it's counterintuitive at first, but that's the reality of it.

 

[RevDrucifer]

If I remember right, it’s something along the lines of amp designers designing amps using a specific cab, so the pre-amp is tweaked to work with that specific cab. I could be wrong, but I seem to recall Cliff saying something similar to that. Since the cab is used to create the sound of the head, it seems more ‘important’ to put the impedance curves in the power-amp section rather than the cab.

 

[GlennO]

This is just something about which I'm curious. It's likely due to not understanding the concept, but I would think that the IR in the cab block should carry information along with it, that decides what the impedance of that particular cab is, and therefor the amplifier chosen in the amp block should/will react to the cab/IR in that fashion. Or have the option of different impedances of specific cabs as different IR's.

Yes, you’re correct, impedance curve should follow the IR selection. The impedance curve is a property of the speaker, not the amp. This also suggests IR makers should supply an impedance curve to go with their IR’s (That's not as complex as it might seem since you only need one curve per speaker in the library and the number of speakers in a library is usually much smaller than the number of IRs).

So, why is it in the amp block? The short answer is “expedience”. It’s easier to implement if you store it In the amp block settings, because that is where it is applied in the processing chain. It complicates the implementation of the amp block if you store it in the cab block where it belongs. But it can be done. Yes, there’s the case of using an external cabinet to consider, but there are better ways to solve that. And there's the complication of multiple IRs, but again it can be done.

I do not know how difficult this would be to implement in the Axe-FX, but the fact remains: the IC is conceptually a property of the cab block, not the the amp block. That suggests it should be stored/retrieved with the rest of the cab block data. Maybe this will change in a future Axe-FX.

 

[Dave Merrill]

Yes, you’re correct, impedance curve should follow the IR selection. The impedance curve is a property of the speaker, not the amp. This also suggests IR makers should supply an impedance curve to go with their IR’s (That's not as complex as it might seem since you only need one curve per speaker in the library and the number of speakers in a library is usually much smaller than the number of IRs).

So, why is it in the amp block? The short answer is “expedience”. It’s easier to implement if you store it In the amp block settings, because that is where it is applied in the processing chain. It complicates the implementation of the amp block if you store it in the cab block where it belongs. But it can be done. Yes, there’s the case of using an external cabinet to consider, but there are better ways to solve that. And there's the complication of multiple IRs, but again it can be done.

I do not know how difficult this would be to implement in the Axe-FX, but the fact remains: the IC is conceptually a property of the cab block, not the the amp block. That suggests it should be stored/retrieved with the rest of the cab block data. Maybe this will change in a future Axe-FX.

I don't think the impedance curve of a cab is something you hear directly in the real world. What you hear is how it affects the amp's output stage. That's why it's in the amp block.

 

[GlennO]

I don't think the impedance curve of a cab is something you hear directly in the real world. What you hear is how it affects the amp's output stage. That's why it's in the shop block.

The calculation that employs the IC is in the amp block. That's why it is where it is on the Axe-FX. But the fact remains the IC is a property of the speaker, not the amp. So, ideally, it should follow your IR selection, not your amp selection.

 

[l4mpi]

As others have said it's in the power amp because that this is where it actually takes effect - the power amp output changes due to the impedance curve. And modelling internals aside, it can't be in the cab block for the simple reason that presets don't need to have a cab block, for example if you play through a real cab with a solid state power amp. You would still want to be able to set an impedance curve in this case, e.g. to match the curve for the real cab.

About the different amounts of IRs vs impedance curves: For starters, an impedance curve is a physical property of a guitar cab. It doesn't change with mic choice or placement. You can shoot an infinite number of different IRs with the same cab, but it has only a single impedance curve.

 

[GlennO]

can't

"Can't"? No . "More complicated to implement"? Yes, but there are software design patterns to be used for situations like this.

 

[Werpinater]

It effects the interaction of the power amp with the cab, which is part of the argument to have it in the amp block. It effects the way the amp reacts. In regards to the “authentic” curve for a cab, I use different IC’s for a different feel. My favorite IR’s are the AC20 cab, but if I want more low end thud for a Marshall amp I dial in a 4x12 IC with the AC20 IR’s. Or yesterday playing around with a Tweed Champ I wanted clearer low end and played with the IC selection to get that.

 

[l4mpi]

"Can't"? No . "More complicated to implement"? Yes, but there are software design patterns to be used for situations like this.

No, "can't". Because Cliff is not going to force people who play through a real cab to add a dummy cab block to their preset for selecting an impedance curve. And everything else you can think of, such as adding it globally, or per-preset "global" similar to the LFO settings, or dynamically add the control back to the amp block if no cab block is in the preset, means it's no longer "in the cab block".

 

[sprint]

I'm no expert and trying to understand also but as I understand from bits of info here/there:

• it's very much a 2-way relationship where a real amp and cab lock onto one another with a unique signature that affects overall response - so it's as much about amp as it is about cab and vise versa.
• the impedance curves provided are not exact as any given real cab via ss amp, or cab used in any given IR, could have their own unique resonances - we are picking the best fit whether using IR or SS amp + real cab with ability to fine tune.

Seems like this is one of the biggest challenges of modelling the whole process because the virtual amp has no way to "see" the actual impedance / resonance of a real cab via ss amp, or IR used. On top of that, the cabinet specs are hard to measure / elusive for most. I can't think of a way a modeller could do it any better than what Fractal has provided aside from some sort of standard where IR manufacturers accurately measure the impedance curves for every cab used in every capture and encode that curve info into every IR so that modellers can lock onto that info to model the amp/cab relationship. Probably not gonna happen any time soon.

I think the best improvement at this point would be for IR/speaker/cab sellers to do more to document what the impedance curve of a given IR should be exactly, and to have better published guidence for real cab owners to more easily get a handle on the low/hi resonances of their real cabs.

Many here also use the ICs as a tone shaping tool - no harm I guess though I think that can skew the modelling accuracy since, irl, it seems there is only one possible curve for a given amp/cab IC signature even though other possibilities may sound better to the user (+ with an unrealistic IC selected for a given cab, I suspect the amp algorithm could start reacting unrealistically for the amp/cab selected - but hey - if it sounds good...)

 

[laxu]

This is just something about which I'm curious. It's likely due to not understanding the concept, but I would think that the IR in the cab block should carry information along with it, that decides what the impedance of that particular cab is, and therefor the amplifier chosen in the amp block should/will react to the cab/IR in that fashion. Or have the option of different impedances of specific cabs as different IR's.

The part that confuses me, is why we'd have say, a limited 60+ impedance curves to chose from, relevant to "specific speaker cab" models, only to then add an actual Cab/IR to that amp, from 1,000's of actual IR's, where that relationship is not authentic? And when I said "limited" that is purely based on the difference between the number of Impedance Curves in the Axe, compared to the number of IR's in the Axe itself, much less what you can purchase from 3rd party vendors.

Am I correct that it's not possible in the real world to mimic the impedance curve of a Morgan 1x12 cab with say an actual 4x12 Recto cab? Nor do I understand why you'd want to mix and match those selections, rather than take the native impedance curve of the actual cabinet for a more "real" example.


Again, this is likely due to me not understanding the concepts, but I figured I'd ask.

Like already said, it's the amp reacting to the speaker rather than the other way around. Which means different amps will also behave differently with that particular speaker.

IRs are more like "what this microphone at this position heard when a test signal for impulse response generation was played" rather than a representation of the speaker/cab itself.

Fractal just exposes the modeling of the speaker impedance curve for you to adjust if you so choose so you can get a tone you prefer or tone you think would be more accurate to the sound you would get with that particular amp-speaker combination. They default to the most common pairing for the amp type.

There most likely isn't a massive difference in impedance curves to make it sensible to provide hundreds of them plus you can always tweak that aspect with the Amp block's Speaker page yourself if you want.

 

[kplooksafterme]

I think it's in the correct place.

Introduction of this feature was also the biggest 'wow' moment for me in all my years of owning digital stuff. It is like the glue that sticks everything together. It affects sound and feel to such a degree that I can't imagine life without it now.

Hoping for the addition of more curves in future.

 

[LiamH]

Not sure if this adds much to thoughts already posted, but here's how it works in my mind. This may or may not work for anyone that hasn't studied electronics, and it was a while back that I was doing it "in anger", so please forgive any misconceptions.

The output transformer is an incredibly important part of any tube amp, and its features can influence both "voice" and "feel". It converts high voltage/high impedance signals to the low voltage/low impedance signals that can drive the voice coil of a speaker. While the voltage output from a transformer is stepped down in this instance, the reflected impedance of the speaker is stepped up by the output transformer, and reflected to the power tubes as something they have to drive. It's highly frequency dependent, and very much part of what's going on inside the amp rather than in the air in front of the speaker cone. IR's are designed/developed to deal with what happens in the air, but the speaker impedance curve deals with the way that the power tubes are being loaded electrically over an AC frequency range.

With all this in mind, the speaker impedance curve belongs very much inside the amp model, and has a huge influence on "feel", especially when the power amp is at limits of headroom and beyond. IRs deal much more with the sound and tone, and include the vagaries not just of the specific mic used to record the IR, but also any artefacts of mic positioning, and room reflections and resonances. While some of these might contribute to reflected load at the tubes, it cannot be by much.

I am not sure an IR could realistically include a speaker impedance curve by its very nature, (but if it can, permalink this post as I will share a claim of "prior art" with the OP.) Seriously though, although not entirely intuitive to non-electronic engineering people, the Fractal way of dealing with this definitely makes sense to me. With luck this post might help others.

Liam

 

[GlennO]

With all this in mind, the speaker impedance curve belongs very much inside the amp model,

Yes and no. The IC is applied to the signal in the amp block, but that’s missing the point. The issue is: should the selection of the IC follow the selection of the amp or the selection of the IR? The natural and logical answer is "the IR" because the fact is the IC is a property of the speaker, not the amp. In other words the IC and IR are a matched pair and when you select the IR, the matching IC should be loaded into the amp block.

FAS has advanced the technology by exposing IC selection to the user. Most modelers don't have IC selection at all. However, that doesn't mean there isn't room for improvement. An obvious way for this to grow in a future Axe-FX would be for the IC selection to follow the IR selection.

Is this the easy way to do it? No. If it was, everybody would be doing it. But advances often come from tackling problems that others shy away from because they are difficult.

 

[Greg Ferguson]

Not sure if this adds much to thoughts already posted, but here's how it works in my mind. This may or may not work for anyone that hasn't studied electronics, and it was a while back that I was doing it "in anger", so please forgive any misconceptions.

The output transformer is an incredibly important part of any tube amp, and its features can influence both "voice" and "feel". It converts high voltage/high impedance signals to the low voltage/low impedance signals that can drive the voice coil of a speaker. While the voltage output from a transformer is stepped down in this instance, the reflected impedance of the speaker is stepped up by the output transformer, and reflected to the power tubes as something they have to drive. It's highly frequency dependent, and very much part of what's going on inside the amp rather than in the air in front of the speaker cone. IR's are designed/developed to deal with what happens in the air, but the speaker impedance curve deals with the way that the power tubes are being loaded electrically over an AC frequency range.

With all this in mind, the speaker impedance curve belongs very much inside the amp model, and has a huge influence on "feel", especially when the power amp is at limits of headroom and beyond. IRs deal much more with the sound and tone, and include the vagaries not just of the specific mic used to record the IR, but also any artefacts of mic positioning, and room reflections and resonances. While some of these might contribute to reflected load at the tubes, it cannot be by much.

I am not sure an IR could realistically include a speaker impedance curve by its very nature, (but if it can, permalink this post as I will share a claim of "prior art" with the OP.) Seriously though, although not entirely intuitive to non-electronic engineering people, the Fractal way of dealing with this definitely makes sense to me. With luck this post might help others.

Liam

Nicely said!

The way I think of it, the speaker impedance curve is half of what affects the change in sound when we connect a different bare speaker to an amp; it affects the sound electrically. The materials and magnets of the speaker are the other half and affect the sound acoustically. Load that speaker into a cabinet and then it acoustically modifies the sound further.

It makes sense to me to keep the impedance curves attached to the amp block, because it’s modeling the electrical characteristics of the amp. Perhaps it makes sense to have a control for the curve in the editor in both blocks and let it gloss over the difference.

 

[unix-guy]

Nicely said!

The way I think of it, the speaker impedance curve is half of what affects the change in sound when we connect a different bare speaker to an amp; it affects the sound electrically. The materials and magnets of the speaker are the other half and affect the sound acoustically. Load that speaker into a cabinet and then it acoustically modifies the sound further.

It makes sense to me to keep the impedance curves attached to the amp block, because it’s modeling the electrical characteristics of the amp. Perhaps it makes sense to have a control for the curve in the editor in both blocks and let it gloss over the difference.

The modeled Speaker Impedance Curve is not the bare speaker. It's the combination of the speaker and the cabinet...

 

[sprint]

Putting on my old database admin hat, I can also see that IC is a property of the Cab IR / real cab, so there's a "data modelling" case to be made to put IC info in the cab block with the IR (or with a totally flat IR in the case of a real cab) - but putting the setting in one place or another does not improve much right now imo (maybe some differences of data entry) - because the real challenge is not where the setting is housed, but it's that the precise IC for a given Cab IR / Real Cab are not readily known in many (possibly most) cases - Fractal has provided 60 or so good guesses which is more than any other modeller does to get users in the ballpark of the true IC of whatever SS Amp + Real Cab / Cab IR they have (unless the user just uses IC as a "what sounds best eq" in which case the IC used could be way off from realistically modelled) - true progress beyond these 60 ICs would be to get more IC precision in play for any given cab IR / SSAmp+Real Cab which is what happens when a real cab is plugged into a real tube amp (a precise reaction takes place) - that seems beyond Fractal's capability to control as they just don't have the info which is probably best gathered by IR makers (if they did this, now the data modelling becomes more important because we'd have the amp modelling picking up the IC info embedded into the IR by the IR maker). I'd bet a key reason the setting currently exists in the amp block is just that the amp block is where all the other genius bits are located where as the cab block is mostly seen as a gateway to third party IRs - not Fractal's sweet spot of development: amp modelling.

 

[RDeraz]

If you really want to get crazy, why not just select resistive load and create your own IC. It's fully adjustable to whatever your heart desires.

 

[GlennO]

Putting on my old database admin hat, I can also see that IC is a property of the Cab IR / real cab, so there's a "data modelling" case to be made to put IC info in the cab block with the IR (or with a totally flat IR in the case of a real cab) - but putting the setting in one place or another does not solve much right now imo (maybe some differences of data entry) - because the real challenge is not where the setting is housed, but it's that the precise IC for a given Cab IR / Real Cab are not readily known in many (possibly most) cases -

Yes, there are two halves to the problem. Both need to be solved.

1) The interface should be changed so the IC is selected automatically when selecting the IR
2) The IR/IC pairs need to be defined so it is known what IC goes with which IR.

Is it challenging to solve a problem where there are two inter-dependent components to the solution? Yes, but the best advancements often come from tackling challenging problems. After all, in the early days of the Axe-FX, I'm sure people said: what's the point of having user-loadable IR's when there are so few IR's available? In time, both halves of the solution emerged and amp modeling advanced significantly because of that. I think it's possible we'll see a similar advancement here with the coupling of IR's and IC's.