Atteunators make it possible to use a head without the cab ?

[Uncle Remus]

Some people use atteunators so they don't need to use a cabinet at all. This way you can, if avalable, use the amps pre amp output for recording. You do need to use cabinet impulses to make it sound right though

I'm looking around for different ways of recording and keeping my options fairly open for the moment.

I came across the Koch 120 atteunator II earlier this week for a good price but it was the 8 ohm version since my cab (Laney) is a 16ohm I assumed it wouldn't work at all.

So i found this quote today ^^ and was wondering exactly what it means. What are impulses? If I get the 8ohm Koch will I be able to just run it through my Laney head that has an 8ohm jack?

Anyone use an atteunator? Any preferences as far as brands/models go?

Thanks in advance guys
Ben

 

[Regor]

Um, I have no idea what 'impulses' means... but I can tell you that unless the attenuator has a load (i.e. speaker cab) you can damage a tube amp. If you get an attenuator that has a load box built into it (First one that comes to mind is the Sequis Motherload), then you don't need the cab.

 

[Uncle Remus]

Yer sorry the Koch has a loadbox ^ *

 

[thebhef]

An impulse is something you'd record which simulates your speaker cab so that you don't actually have to use one. It's a big part of how a cab is modeled. That's all I really know about those.
The attenuator will need a load, like Mr. Yacht said. If you leave it open, you would probably kill your output transformer.

 

[Uncle Remus]

So if I use an attenuator with just my head where does the sound come from ?

/slightly nooby question

 

[cev]

You would use the 'send' from the effects loop, and hook that straight up to whatever recording interface you're using. That only gives you the sound of the preamp, so you need to model a poweramp and a cabinet somehow (which is why the post you quoted suggests using impulses).

 

[Uncle Remus]

i understand

Thanks

 

[Variant]

You would use the 'send' from the effects loop, and hook that straight up to whatever recording interface you're using. That only gives you the sound of the preamp, so you need to model a poweramp and a cabinet somehow (which is why the post you quoted suggests using impulses).

True, true, but, if you want to capture the characteristics that the power section adds (tonal shaping, power tube compression & distortion, sag, etc.) you'll want to take the signal out of your attenuator... that is IF it has one. I know the THD's can feed a signal form the Hot Plate itself.

 

[Scali]

You would use the 'send' from the effects loop, and hook that straight up to whatever recording interface you're using. That only gives you the sound of the preamp, so you need to model a poweramp and a cabinet somehow (which is why the post you quoted suggests using impulses).

In that case you probably don't need an attenuator anyway. You can just leave the cab connected and set the master volume to 0 (or with a serial effect loop you won't even need to do that, because if you don't plug anything into the fx return, no signal reaches the power section anyway). The master volume is after the effect loop, so it won't affect the preamp and fx send at all. In fact, my Marshall 6101 has a 'killswitch' for the poweramp, so you can just use the preamp output (either with or without its built-in cab simulator).

There are various kinds of attenuators though, some of which contain speaker simulators, so you can record them directly.

I personally think that if you want to skip your cab+mic setup anyway and rely on impulses or such to simulate it, that the whole tube amp won't do you too much good anyway. In my opinion the cab+mic is the most important part of the whole equation... If you're going to simulate that, you might aswell simulate the whole preamp, because I don't think it will sound more realistic with a real amp.
I've gotten better recordings out of a transistor or modeling preamp playing through an actual mic'ed up cab than I did with a real tube amp through speaker simulation software or hardware.

 

[zimbloth]

I personally think that if you want to skip your cab+mic setup anyway and rely on impulses or such to simulate it, that the whole tube amp won't do you too much good anyway. In my opinion the cab+mic is the most important part of the whole equation... If you're going to simulate that, you might aswell simulate the whole preamp, because I don't think it will sound more realistic with a real amp.
I've gotten better recordings out of a transistor or modeling preamp playing through an actual mic'ed up cab than I did with a real tube amp through speaker simulation software or hardware.

It's clear to me you're not too familiar with impulses, because it's not modeling or emulation, it's the real deal. It's simply the sonic signature of a room with all those crucial components factored in, in the form of a .wav file which is processed in a specialized plugin. It's nothing like cab modeling or digital amps, it's the real deal. I agree with your general sentiments though, but impulses have changed the game.

I just use my THD Hotplate and set it to 'load' when I record using impulses. Just be careful if you're not using a cab, make sure everything is always connected properly or else you can damage your amp.

 

[Scali]

It's clear to me you're not too familiar with impulses, because it's not modeling or emulation, it's the real deal. It's simply the sonic signature of a room with all those crucial components factored in, in the form of a .wav file which is processed in a specialized plugin. It's nothing like cab modeling or digital amps, it's the real deal. I agree with your general sentiments though, but impulses have changed the game.

How exactly is it different from modeling/emulation then?
The whole purpose of modeling is to model the entire 'sonic signature' of an amp, cab, microphone, room etc.
As far as I see it, the term 'impulse' was just coined to describe a certain kind of reverb/room modeling. It's still modeling/emulation, because it's not real. It's about as real as Star Trek's Holo-deck.
The idea is exactly the same as that of modeling of other types of effects/gear/phenomena.

In essence a Pod or similar device does the same thing... It first records a '.wav' of your clean guitar, then it processes it with the sonic signature of a certain amp, cab, mic, reverb and other effects, trying to factor in all crucial components. The end result sounds pretty close to what you'd expect of the real thing, since it uses pretty accurate models of these sonic signatures.

 

[7 Dying Trees]

Modelling uses maths to approximate a curve, with variables factored in, ie, treble/bass/mids/presence/gain/etc etc etc of an amp. You measure several 100 points or so, build the integral equations and use approximation to get to an equation that "approximates" the real life behaviour. You can use fast converging series like ortogonal polynomials (a nice one to look into if you like your maths and approximation as there are several you can use to fit different curves better, but look into it). Anyway, with all these methods there is in a processor a finite amount of coefficients you can use before you start running out of processing power, so you have to sacrifice "realness" to make the unit operate at a fast enough speed to process the audio so you don't get lag from the DSP chip.

With impulses what you effectively have is the impulse response of a "black box". This black box could be anything, a room reverb signature, a cab, a power amp, or, as in the case with what musicians terms as impulses a whole chain of a placed mic, mic type, cab, speaker type, room characteristics, power amp characteristics.

Why is this different? Well, with an impulse you no longer have the issue of making an approximate mathematical algorithm converge as closely as possible with the real thing. There are no parameters to feed in. What it represents is the entire system frequency response for that particular enviroment.

To create the impulse all you are doing is sending a noise sample through a system, or a sine wave (there are several ways of doing this, i can't remember, electronics at university is 10 years ago, but i think you use a white noise signal because of the equal energy across the frequency spectrum). So what you get is the reality, not a modelling approximation.

To be fair, yes, you can't go off and slightly move the mic 1cm to the right in an impulse or slightly change the angle, hence why you end up having a load of impulses and just choosing the one you like best.

However, overall, if you look at it, you are using a real "impulse response" of the system which represents what the equation graph of Voltage or Current vs frequency would look like for the entire circuit of power amp, cab and microphone as it were. Then, you just convolve this with the preamp output to get what the preamp would sound like given the fixed power amp, cab, speaker, microphone and room charactersitics.

There literally is no "modelling" involved, as per definition, modelling is an approximation of a real life system that takes in external parameters that change the impulse repsonse.

What we have here is a pure impulse response, so whilst you don't get the tweakability of modelling, you do get the real life characteristics without the approximation.

Of course, there's a lot you can tweak after the signal has gone through the impulse response, or you can tweak the impulse response itsself in some programs, but that's pretty much the same as eq-ing, which you end up doing anyway.

Hope that makes it clearer

I'd also say give it a go, it is a bit of a weird one to start, but say you take the pod and turn of cab modelling on that and use impulses instead, then people like mattayus on here have been getting excellent way more realistic sounding results just from that.

Another advantage is silent recording with an actual tube head as well.

 

[Scali]

Why is this different? Well, with an impulse you no longer have the issue of making an approximate mathematical algorithm converge as closely as possible with the real thing. There are no parameters to feed in. What it represents is the entire system frequency response for that particular enviroment.

I disagree.
Any mathematics performed by any computer is inherently discrete, and as such it is an approximate mathematical algorithm. You can never represent the 'entire system frequency response' since that is an infinite amount of frequencies and responses, requiring an infinite amount of processing power and memory to process.
So, we're back to the fact that they're both approximations.

To create the impulse all you are doing is sending a noise sample through a system, or a sine wave (there are several ways of doing this, i can't remember, electronics at university is 10 years ago, but i think you use a white noise signal because of the equal energy across the frequency spectrum). So what you get is the reality, not a modelling approximation.

It's not reality however. You take a few test samples and generalize the solution (the assumption is that white noise is the result of a combination of every frequency... but in practice the limited accuracy of discrete mathematics will break this assumption up), then apply it back to a different signal.
It's much like the re-lighting that is done in movies and such, where you try to distill the lighting solution (most commonly with a spherical harmonics solution) out of a certain image or set of images, and then re-apply this lighting to a different image or set of images, in order to make an object appear to be in an environment that it actually isn't, by simulating all the lights, reflections and shadows.

Ironically enough what you end up with is a set of (predetermined) polynomials and a matrix of constants, just like how you described modeling. It's the same concept.
I would expect that companies like Line6 also send signal generators through actual amps, and sample how various components respond. They might just tweak the matrix of constants and allow the user to modify certain constants (turning them into parameters) in order to not only sound like the actual amp, but also be able to use the amp's controls in a realistic way and allow you to do things like changing the microphone position or adjusting the room size, in order to be more flexible.
That doesn't mean that the original amp/cab/mic/room/etc models weren't based on samples taken from actual amps/cabs/mics/rooms/etc. They probably just found a clever way to blend them together for more flexibility.
They don't just pull those modeling equations and things out of thin air, you know. They wouldn't sound as realistic as they do, if they weren't based on actual sampled data, much like your impulses.

To be fair, yes, you can't go off and slightly move the mic 1cm to the right in an impulse or slightly change the angle, hence why you end up having a load of impulses and just choosing the one you like best.

Technically you could. You could add that extra dimension to your set of measured values, and interpolate between those aswell, just like you interpolate between other 'missing info' in your data set.

There literally is no "modelling" involved, as per definition, modelling is an approximation of a real life system that takes in external parameters that change the impulse repsonse.

Actually there is. Whoever 'sampled' the room characteristics in order to distill it into an impulse dataset was creating a mathematical model of that room. When you apply that model to your signal, you are creating an approximation of what your signal would have sounded like in that room.
In fact, any digital signal (audio, video or otherwise) is an approximation of the real thing by definition, with limited resolution.

What we have here is a pure impulse response, so whilst you don't get the tweakability of modelling, you do get the real life characteristics without the approximation.

Just because most modeling software/hardware allows you to tweak things doesn't mean that it's a requirement for modeling.
One could also argue that it's just a matter of defining your model. Apparently most amp models (but not my Zoom G9.2tt for example) choose to model the amp as a dynamic system, including not only gain, bass, treble knobs, but also other things, like microphone position etc. I'm quite sure that they arrived at these models in a very similar way to how room reverb impulses are sampled and stored, it's just that a room isn't as dynamic and tweakable as an amp is, so it doesn't make a lot of sense to try and model gain or eq knobs or anything.
You could however take two different microphone positions, and allow the user to blend between the two sets, which will probably give a reasonably close approximation of moving the microphone from spot A to spot B in the room. You could theoretically do this for as many positions and microphones as you'd like.

Hope that makes it clearer

Not really. I already knew how it worked, I just said I saw more similarities than differences between modeling and impulses.

I'd also say give it a go, it is a bit of a weird one to start, but say you take the pod and turn of cab modelling on that and use impulses instead, then people like mattayus on here have been getting excellent way more realistic sounding results just from that.

Thing is, are you talking about modeling in general, or just about a Pod?
Because certainly impulses can sound better than a Pod, but then so can many other modeling units or software.
For me, the best test is to use headphones. Then you get no real room ambience in your sound at all, unlike with speakers (which may make you think the artificial sound is better than it really is... I prefer to monitor via speakers rather than headphones when I record direct, because it gives me a more realistic sound). Somehow I can always tell modeling, impulses or whatever other direct recording from an actual mic'ed recording. It gets harder every year, but it's still not quite there yet.

 

[zimbloth]

Scali, you're wrong as usual. You don't know jack about impulses, period. Stop wasting our time by pretending otherwise. This isn't an opinion thing, you're factually wrong on absolutely every point James eloquently laid out for you.

Impulses are NOT MODELING, they're .wav files captured from micing the sound of a room. To make an impulse file you simply introduce a frequency sweep into your amp and mic it. Then you remove the sweep and all that's left is the natural sound of the room. That sound is captured in a .wav file and a specialized convolution reverb plugin processes the dry signal from a preamp along with the sound of the room, and boom. There it is and it works. The poweramp, mic, mic position, room acoustics, etc are all factored in naturally not emulated.

Please, I know you get off on being a contrarian but please stop chiming in on topics you clearly know nothing about. It brings nothing to the table.

 

[shadowgenesis]

Scali, you're wrong as usual. You don't know jack about impulses, period. Stop wasting our time by pretending otherwise. This isn't an opinion thing, you're factually wrong on absolutely every point James eloquently laid out for you.

Impulses are NOT MODELING, they're .wav files captured from micing the sound of a room. To make an impulse file you simply introduce a frequency sweep into your amp and mic it. Then you remove the sweep and all that's left is the natural sound of the room. That sound is captured in a .wav file and a specialized convolution reverb plugin processes the dry signal from a preamp along with the sound of the room, and boom. There it is and it works. The poweramp, mic, mic position, room acoustics, etc are all factored in naturally not emulated. "Boom, there it is and it works." Seriously?

Please, I know you get off on being a contrarian but please stop chiming in on topics you clearly know nothing about. It brings nothing to the table.

Ummm. Nick, you're a great guy and all, but I'm gonna go ahead and disagree with you here man. You reference convolution reverbs, which use impulses, but I'm pretty sure they are not the only way impulses are used. In fact, you were very vague in your description of how the process works.

As far as I know, an impulse is just a mathematical analysis of how a particular environment (room, signal chain, etc) changes the characteristics of a signal or sound. It could be physical (such as the acoustic qualities of a room), electrical (such as the power section of a tube amp), or a combination of the two.
You use controlled signals like white noise, pink noise, and sine waves through a certain environment and record the resulting sound to identify those changes. Then you apply those changes to your signal to recreate that environment.

Regardless how it works, it's a digital simulation of an environment. You can't, inside a computer, actually create a room and send a sound into and record it back. No matter HOW complex and accurate, you're still only using a mathematical process to try and recreate that environment.

 

[Variant]

Wait... a bit of a third party to all of this but are impules:

a. A sample of the ambiance of a room which is added into the guitar track by a VST triggered off the attack of the signal? (which is how I understood it)

b. An actual sample of a played note (possible plus room response) through an amp setup, again, mixed into the signal via a VST?
or...

c. A sample of a particular amp in a room which is then used by a VST to formulate an algorithm to process the input signal in reference to? In which case, that is indeed modeling.

 

[zimbloth]

Ummm. Nick, you're a great guy and all, but I'm gonna go ahead and disagree with you here man. You reference convolution reverbs, which use impulses, but I'm pretty sure they are not the only way impulses are used. In fact, you were very vague in your description of how the process works.

Yes, I agree my halfass explanation of how impulses work was not very good. I was too lazy to cut and paste a proper description.

You're right, on some level there probably are algorithms involved, but to claim it's exactly the same as how PODs or any modeler works is ridiculous. I've tried every modeler that exists pretty much, including running legit tube amps into modeling software, and none have the realism of using real-life captured impulses with tube pres. Anyone who has ever used a POD knows they don't sound anything like the real amps. I'm pretty certain there are differences in the technology.

Are you trying to say that it's exactly the same, only every other modeler simply emulates it poorly whereas impulse files are more accurate?

I'm not claiming to be an expert here, all I can tell you is in my experience impulses and modeling sound and feel nothing alike to me

 

[Variant]

From Wikipedia, the free encyclopedia

In audio signal processing, convolution reverb is a process for digitally simulating the reverberation of a physical or virtual space. It is based on the mathematical convolution operation, and uses a pre-recorded audio sample of the impulse response of the space being modelled. To apply the reverberation effect, the impulse-response recording is first stored in a digital signal-processing system. This is then convolved with the incoming audio signal to be processed.

An impulse response is a recording of the reverberation that is caused by an acoustic space when an ideal impulse is played. However, an ideal impulse is a mathematical construct, and cannot exist in reality, as it would have to be infinitesimally narrow in time. Therefore, approximations have to be used: the sound of an electric spark or a starter pistol shot, for instance. A recording of this approximated ideal impulse may be used directly as an impulse response.

Alternatively, one can excite a space with a longer sound (typically a sine sweep), then perform a deconvolution to produce an impulse response. This approach has the advantage that such sounds are less susceptible to distortion; however, it requires more sophisticated processing to produce a usable impulse response.

A third approach involves using maximum-length sequences, but this is difficult in practice because such sequences are highly susceptible to distortion.




Sounds like sound modeling to me.

 

[eaeolian]

OK, I hacked the flame war out of this thread. Nick, you know better. Scali, you're courting a perm, at this rate - how many times do you need to get banned before you get it? Nick is right about one thing - your post to 7DT makes you come off like a sanctimonius jerk, despite the fact that you are in essence correct.

 

[Scali]

Scali, you're courting a perm, at this rate - how many times do you need to get banned before you get it?

Excuse me? I was never banned before. I think you have me confused with someone else.