Analog cab sim

[caspercody]

Hello

I want to build an analog cab sim, anyone know a really good accurate schematic?

Also, anyway to create an analog cab sim based on a IR one might use the most?

Thanks
Rob

[fryingpan]

What I'd do:

1) a 2-pole high pass filter cutting below 150-200Hz for an open-back speaker (there are a few sweepable HPF schematics you can use)
2) a variable "proximity effect" that is merely a soft bass shelf (starting to boost high up, like 500-600Hz, 0-10dB)
3) a filter bank, with
- a narrow (Q = 2.4 or so) 4-5dB boost at 80-100Hz for speaker resonance (which, combined with the high output impedance of a tube amp, gives much of the "fatness" you feel with tube amps)
- a wide, shallow mid cut (let's say Q=0.6, -3/4dB centered around 600Hz)

and then, depending on the tone you have in mind (the following is for an "American" speaker, aka Fender:
- a wide, strong mid boost (let's say Q=0.7, 6-8dB centered around 2.5kHz)
- a few narrow boosts or cuts in the area between 1kHz and 5kHz
4) a steep 4-pole LPF around 6-8kHz (that's the physics of a guitar speaker)
5) possibly a smoother, sweepable 1-pole LPF between 3kHz and 8kHz (on-axis/off-axis)

Experiment a bit with your DAW of choice (maybe play your guitar on your amp and take the line out, if it has one, and record, then play around with an EQ). First design the EQ, and then move on to the actual circuit design (which is not that hard).

This is my idea of an EQ, for instance:



The left-most green curve is the HPF, the grey curve is the proximity boost.

[fryingpan]

Then, if you were to truly model a guitar speaker, you should also account for speaker distortion (some not-so-soft clipping before the final LPFs).

[Mark Hammer]

Insomuch as there are dozens upon dozens of different speakers used and an equal number of cab designs/dimensions, I don't know that there is anything one might call an accurate schematic, unless one specifies the cab and speakers, such that specific resonances and rolloffs can be identified and electronically mimicked as closely as possible.  But even there, is one trying to achieve a sound like an amp angled upwards, placed near a wall or in an open space, etc.?

[fryingpan]

Insomuch as there are dozens upon dozens of different speakers used and an equal number of cab designs/dimensions, I don't know that there is anything one might call an accurate schematic, unless one specifies the cab and speakers, such that specific resonances and rolloffs can be identified and electronically mimicked as closely as possible.  But even there, is one trying to achieve a sound like an amp angled upwards, placed near a wall or in an open space, etc.?

Cab sims, be they analog or digital (IRs) are all basically crude representations of the true tone of a cabinet. Most digital IR-based cab sims do not "capture" (well, it's the IRs they employ...) the room/space where the cabinet is supposedly playing (although they totally could! Even recording an IR on the fly), but they simply apply convolution of a close-micing setup (presumably at a low enough wattage with a linear enough amp). They can work in a pinch for live purposes (sometimes much better than micing, especially on very small stages or bad rooms). I wouldn't use them to record.

[GibsonGM]

To the original question: When not using IR's in my DAW, this one really gets the job done. Kudos to Rob!

I did get the PCB from him, great price, and made assembly much simpler.

https://www.diystompboxes.com/smfforum/index.php?topic=121846.0

[bluebunny]

Yep, agree.  Rob took the time to do a bit of tinkering with his sim (as is Rob's wont).

[merlinb]

https://valvewizard.co.uk/cabsim.html

[Rob Strand]

The problem with most speaker sims is they are usable but they don't sound like mic'd speakers.
An *accurate* cab sim is one which sounds like a speaker doesn't require post EQ.

This video shows the problem,

- jmp1 midrange sound, typically fixed by rolling back the mid EQ on the unit (can add a bit of bass too)
- Ultra G,  there is a darkness or lack of definition.  Lots of cab sims like this.
- speaker has a particular character not present on the cab sims.
  More highs but they don't sound bad.
  Obviously mic position changes things a lot and there's some pretty dark muddy recordings from speakers out there.   That case is missing from the video.

Then not all speakers sound the same, not a great example but you get the idea,

[fryingpan]

The problem with most speaker sims is they are usable but they don't sound like mic'd speakers.
An *accurate* cab sim is one which sounds like a speaker doesn't require post EQ.

This video shows the problem,

- jmp1 midrange sound, typically fixed by rolling back the mid EQ on the unit (can add a bit of bass too)
- Ultra G,  there is a darkness or lack of definition.  Lots of cab sims like this.
- speaker has a particular character not present on the cab sims.
  More highs but they don't sound bad.
  Obviously mic position changes things a lot and there's some pretty dark muddy recordings from speakers out these.   That case is missing from the video.

Then not all speakers sound the same, not a great example but you get the idea,

Keep in mind that mic'ing a cab that close, apart from the proximity effect, will also impart the microphone's tonality to the sound. (The SM57, I think, is adding a lot of that nasal quality you can hear in the mic'ed cabinet).

With an IR loader, you can presumably capture a cabinet in all its resonances, peaks and valleys, caused by (it's a guitar cabinet):

- the cabinets being relatively shoddy by hi-fi standards (so they resonate, they are badly sealed when they are closed-back, open-back means a lot of phase cancellation and interaction with the room, etc.)
- the speakers being low excursion (lots of distortion added by the woofers themselves), operated will into their breakup mode (the treble range) where the woofer has actually very little control over the movement of the cone

and they can also add a tiny bit of ambience.

My personal issue with IR loaders is that actually finding (or making) a good IR is not all that simple. And you do have added latency (sometimes it's immaterial, 2ms or so, equivalent to standing less than a metre away from the cabinet, but it's implementation-specific). Also, IRs assume a guitar speaker is a linear affair when it is far from being so. At the end of the day, an IR of a guitar speaker mostly boils down to an EQ (of the FIR variety). That's all it is. With as many bands as there are samples within, technically.

An analog version that sounds convincing does need to add in some "character" (which is why, especially in the 1-5kHz region, I would take the trouble to add in at least three or four variably narrow bands boosting and cutting). And you're not going to get the reverb (and no, adding realistic reverb is not all that simple in hardware, reverb really means digital - and once you go digital, you might as well do it all digital).

The Astrocab thingy does sound reasonably convincing (if a bit "flat", like all cab sims). The switches, the treble and air controls can be coaxed to yield an interesting response (although I find that rolling back the treble control makes things dark quickly).

[Rob Strand]

Keep in mind that mic'ing a cab that close, apart from the proximity effect, will also impart the microphone's tonality to the sound. (The SM57, I think, is adding a lot of that nasal quality you can hear in the mic'ed cabinet).

It's kind of a whole package and it's what is "normally" there.

- the cabinets being relatively shoddy by hi-fi standards (so they resonate, they are badly sealed when they are closed-back, open-back means a lot of phase cancellation and interaction with the room, etc.)

My personal issue with IR loaders is that actually finding (or making) a good IR is not all that simple. And you do have added latency (sometimes it's immaterial, 2ms or so, equivalent to standing less than a metre away from the cabinet, but it's implementation-specific). Also, IRs assume a guitar speaker is a linear affair when it is far from being so. At the end of the day, an IR of a guitar speaker mostly boils down to an EQ (of the FIR variety). That's all it is. With as many bands as there are samples within, technically.

In acoustics nothing is simple.

The biggest thing which helps the accurate measurement of mic'd guitar speakers is the mic'ing is normally done close.  That means it senses the near field behaviour of the speaker.   The signal there is quite strong so it inherently reduces the effect of other signal sources.  That includes cancellations from reflections.   The exception being room effects at low frequencies but luckily the limited guitar spectrum helps here too.    The response of an open back cabinet is quite complicated in a room and certainly doesn't sound close to the close mic'd signal.

HiFi people have been strugging with speaker measurements for years.   Interestingly Kippel developed a new technique to extract the speaker response from multiple measurement positions.  (FYI: It's based on mathematical projections, and spherical basis functions.   Makes sense if you that stuff but totally alien to non-mathematicians.)  Some of the guys there were using that technique to extract some impulses for guitar speakers and other stuff.   I'm not so sure about it replicating what a normal mic'd guitar speaker set-up produces.

An analog version that sounds convincing does need to add in some "character" (which is why, especially in the 1-5kHz region, I would take the trouble to add in at least three or four variably narrow bands boosting and cutting). And you're not going to get the reverb (and no, adding realistic reverb is not all that simple in hardware, reverb really means digital - and once you go digital, you might as well do it all digital).

Yes all good tweaks (and also your previous list).

To me mic'd speakers have a character.  Even crap setups.   This character is always missing from the cab sims.   I experimented with narrow notches some time back and it definitely made the cab sim sound more like a speaker.

[Jim Hagerman]

This has an adjustable treble peak, which you can just fix to 10k or so...

[fryingpan]

What helps to know, when designing a cab sim, is that the human ear apparently has a hard time discerning anything narrower than 1/3rd of an octave (when it comes to EQing), unless of course there is a very strong resonance (and even then, you may find that going extremely narrow to zero in on the offending frequency is an exercise in frustration, and you are better served by not such a narrow notch). In between 1 and 8kHz (where, if you look at speaker plots, you'll find most of the "jagged" response) there are three octaves, so 9 narrow band filters, and you don't even need that many because many speakers actually roll off much earlier (especially "British" speakers) and you don't necessarily need that fine a level of detail. It's not that hard to implement.

My suggestion is, again, play with a digital EQ first, then translate that into a design.

[caspercody]

Jim Hagerman, what does your schematic do?