Drive

TomC · 30 Jun 2009

A Uio wrote: Hi,

What's the mathematical basis/definition of "Drive"?

Thanks, A

I guess 'drive' is something metaphorical like 'groove' that cannot be described with numbers.

Tom

AngularMomentum · 30 Jun 2009

Amen.

tektoog · 1 Jul 2009

Overdrive maybe???

see: http://en.wikipedia.org/wiki/Guitar_effects

Tzarls · 1 Jul 2009

Drive = Gain? Which, depending the situation, can generate distortion.

TomC · 1 Jul 2009

Tzarls wrote: Drive = Gain? Which, depending the situation, can generate distortion.

THAT's a drive: (and, BTW, drives like a charm)

sunsynth · 1 Jul 2009

HaHaHa Tom, my thoughts exactly!

bootsy · 1 Jul 2009

Yeah, but does it offer "Fahrvergnügen"? 🙂

stw · 1 Jul 2009

bootsy wrote: Yeah, but does it offer "Fahrvergnügen"? 🙂

not on a mathematical basis... 😈

TomC · 1 Jul 2009

bootsy wrote: Yeah, but does it offer "Fahrvergnügen"? 🙂

Are you kidding? This thing IS Fahrvergnügen

Tom

A Uio · 2 Jul 2009

After some investigation I found the Synthmaker Overdrive component is Gain followed by the code

streamin in; streamout out;

float x; float a,b;

x = in * 2; a = x - (x < 0 & (x+x)); b = (6+a*(3+a)); out = (x*b)/(a*b+12);

Which translates for x > 0 to

k = 2*in^3 + 3*in^2 + 3*in out = k / ( k + 3 )

a sort of limiting curve.

So it appears that drive/overdrive is simply gain with a limiting curve instead of a hard clip. Pretty easily described by numbers after all 🙂 I wonder if people ever make a multiband drive that handles frequency bands separately...

Cheers, A

Nu Audio Science · 2 Jul 2009

Yeah i made one in SM with Unk about 5 years ago or something heheheh Might remake it at some stage

NAS

A Uio · 2 Jul 2009

Interesting perhaps to consider the difference between Drive and Compression. Drive rounds off the tops/bottoms of waves, changing their frequency characteristics, converting signals into squarish waves. Whereas compression averages signal level (power) through a sliding time window and then reduces the whole signal by a scaling factor, with considerably less effect on the sound.

Cheers, A

Tzarls · 2 Jul 2009

Actually drive just aplies some gain to a signal. When the signal goes above a given maximum, it goes into overdrive, clipping the signal and making it more squarish. Technically speaking, the signal is also compressed.

theun · 2 Jul 2009

A Uio wrote: So it appears that drive/overdrive is simply gain with a limiting curve instead of a hard clip. Pretty easily described by numbers after all 🙂

The term "drive" comes from old valve amplifier design. The final tube of a valve amp (so the real power amp) is "driven" by the so called "driver" tube. See picture below, the 2 triodes of pre-amp and driver will be combined in e.g. one ECC83, the pentode could be an EL84 or EL34 in a practical design.

You'll find the same kind of design in guitar (tube) ampliers. In early guitar amp design (e.g. Vox and Fender designs) the idea behind the design was to make the (clean) sound of the guitar LOUD. Only when the volume control was put (too) high the power amp tube started to distort. So the nice sounding tube distortion sound could only be heard at high volume levels. Later, more sophisticated designs introduced volume pots at various stages of the amp. This is a very simplified layout of a Mesa Boogie amp.

By raising the volume pot level at the "drive" stage and lowering the volume level at the final stage it becomes possible to let the amp distort at much lower output volume levels. However, now it's not the poweramp tube that distorts, but the driver tube instead.

So yes, it's all about tube distortion (=non-linear behavior), and how to model that.

Cheers, theun

MegaHurtz · 16 Jul 2009

If you would moddel it on such design it would be something of a sine behavour i presume.

theun · 18 Jul 2009

A sine ??

Well, what you would have to model is tube distortion, which is essentially non-linear behavior. You can view a graph showing the typical tube amplification behavior here: http://www.vias.org/basicradio/img/basic_radio_hoag_Page_369-224.gif

To model this i'd start with a tanh function, use something like x-((x^3)/3)+((x^5)/5)- ....... to estimate. However even simplified calculations like that tend to use lots of CPU. So it might be better to store the coefficients in a lookup table ....

Regards, theun

MegaHurtz · 18 Jul 2009

I had a little run with it. Think im going to map actual tubes. This is a concept, does two things. some sort of real cheap class B, and class A. I Liked to make it work like a real one, "like you know" as in no minus. To finish it.. I think it really needs a nice yitter reduction. Butterworth did`nt perform. Too bad.. But for now with falling edges.

MegaHurtz · 19 Jul 2009

yes it`s definately better to do a convolution. (for as far it`s needed).No need for fir`s here. But this is fun to do.

theun · 19 Jul 2009

As I am not at my own machine -it's the holiday season- I can't have a look at your osm, MegaHurz 🙁

I forgot to mention that one of the articles in the current Computer Music Journal gives an overview on guitar amp modeling techniques. Normally you have to pay for each article, but this one is free, get your copy here: http://www.mitpressjournals.org/doi/pdf/10.1162/comj.2009.33.2.85

Even if you don't like reading this kind of scientific texts (though this one is not very difficult, I think), the list of references at the end gives quite some interesting internet links 🙂

Cheers!

MegaHurtz · 19 Jul 2009

I love this stuff. The Yamaha pattent`s will be next.

MegaHurtz · 20 Jul 2009

I could`nt resist building the two models mentioned in the text. Sow here they are.

stw · 20 Jul 2009

MegaHurtz wrote: I could`nt resist building the two models mentioned in the text. Sow here they are.

Some slightly optimized code version...

MegaHurtz · 20 Jul 2009

n1, like this it`s hardly stealing 😀. Im working on a triode model now, only it`s getting kinda confusing. I dunno how many volts this stuff needs. what leaks and not. Hardly able to validate if anything is correct. like use natural log. ah whell it`s pretty versatile when it`s there.

theun · 21 Jul 2009

You guys are so fast .... I'll have a look at your stuff when back @ my own machine ...

Cheers

hansdump · 6 Aug 2009

This thread is really amazing... although it is so old... but thanks.

So i got a question about the lookup-table-nonlinearity mentioned in the paper.

Is there anybody outta there who can explain to me a way to make a table-look up in samplerate domain?

Maybe without interpolation?

Of course all that in Synthmaker.

Thanks,

H.

Dozius · 6 Aug 2009

stw wrote: MegaHurtz wrote: I could`nt resist building the two models mentioned in the text. Sow here they are.

Some slightly optimized code version...

slightly more optimized in asm

MegaHurtz · 9 Aug 2009

Interesting, i figured not much that could be done in asm. let`s see what you did. pz!

MegaHurtz · 11 Aug 2009

Basicly you yust replace the old with the new level. So what you can do is for every veritcal "bit" in the sound you can have an index in the buffer. You should have no need for interpolation if every step of 2^bit is covered.

ronaldpassion · 5 Sept 2009

Distortion and OD i default OSM is good enough...

I am looking another algorithm or any DSP literature regarding guitar preamp/distortion, any one know?

Or does any one in here know about transistor/capacitor value(in farad) convert to Hz (Freq) ??? I am trying to understand real amp schematic by know THE FREQ selection they have....

Thanx 🙂

robrokken · 5 Sept 2009

hahah RP I was just gonna make a post about this, as I am wondering the same thing also lol

~Rob.

MegaHurtz · 20 Sept 2009

Korens triode model is a really good one. There you can mimic tubes. It`s a heavy build nevertheless. It can be done if you know about eulers log and signum. If completed you have to know a little electronics but ittl be alright.

The nature of an amp sim is the distortion it makes, you could state that in "digital audio land" it doesnt matter anymore what kind of heritage you have. But soully the sound it makes. If you want flexibility you should check out the amp sim in Audition/Cool edit for your inspiration. It`s the real deal.

Cheers.

robrokken · 25 Sept 2009

Is there currently any analyzer I can use to see the transfer curves of these distortions, or any distortions that I make, like a tube ??? Really trying to get something that ramps up slowly.. But need analyzation tools to do it.

~Rob.

hansdump · 25 Sept 2009

So far for that... Converting x y Oscilloscope from Reaktor

But if you look at any curve you will not really get what is inside of it.

And the other thing is that the lissajous figures give you an additional information about the phase behaviour of the circuit... so if you get it to record this additional dimension please tell me if succeeded... I would be very interested... 😈

H.

ronaldpassion · 26 Sept 2009

robrokken wrote: Is there currently any analyzer I can use to see the transfer curves of these distortions, or any distortions that I make, like a tube ??? Really trying to get something that ramps up slowly.. But need analyzation tools to do it.

~Rob.

Analyzer my friend???

i use christian w budde analyzer a lot Good for testing our plugin, see the waveshaping curve, the HD etc...

Thanx 🙂

hansdump · 29 Sept 2009

Although it is nice to know how other plugs work and this is a very nice tool what I didn´t know before my advice is to build as much as you can on your own. If one tries to avoid problems that aren´t clear there could appear much more bigger problems... 😉

H.

robrokken · 29 Sept 2009

download/file.php?id=2270

If that was flipped upsidedown so that it gradually goes up.... on a little steeper angle, that would be a great distortion...

~Rob.

MegaHurtz · 15 Oct 2009

in*in

squareal · 25 Mar 2010

Hi

Could someone please explain in layman terms about the basic tube amp on the first page. In particular about the code for drive. I 'm a bit of a noob and the math kinda freaks me out. If not then could anyone point me in the right direction (websites etc)

Much appreciated.

Jay · 25 Mar 2010

Hi squareal Welcome

I'm like you m8 when it comes to maths so i cant help you with explaning asm code but there is masses of stuff related to Synthmaker and dsp in general on our wiki - [old link] and have a look at the links section of the wiki for many many sites related to dsp, asm etc etc

Sorry i cant help you directly but i'm sure one of the Guru's will offer you some insight/help but, try these places in the meantime 😀

Best Regards

squareal · 25 Mar 2010

Thanks Jay

I'll look into the info you supplied.

Its good to know that Im not the only one here who is slighlty dyslexic with math.