KVR Audio

kylen · Post by **kylen** » Wed Oct 19, 2005 8:04 pm

I've seen a couple of folks mention this over at KVR as well as at least one developer who made an [obsolete] creamware plugin called V-Tube.

What's the problem with waveshaping - isn't that what happens at some point down in the dsp anyway? Or are we just talking in terms of user interface?

I'm using a few tube sweetening type VST plugins right now to try and model some amp and pre-am systems to sweeten my audio. If I look at a 1KHz waveform thru any of these it seems that sooner or later there is waveshaping whenever harmonics are produced! What do folks mean when demonizing waveshaping in this context?

pj geerlings · Post by **pj geerlings** » Wed Oct 19, 2005 8:13 pm

kylen wrote:I've seen a couple of folks mention this over at KVR as well as at least one developer who made an [obsolete] creamware plugin called V-Tube.

What's the problem with waveshaping - isn't that what happens at some point down in the dsp anyway? Or are we just talking in terms of user interface?

I'm using a few tube sweetening type VST plugins right now to try and model some amp and pre-am systems to sweeten my audio. If I look at a 1KHz waveform thru any of these it seems that sooner or later there is waveshaping whenever harmonics are produced! What do folks mean when demonizing waveshaping in this context?

Just MHO but I think there is more to it - the power-supply-sag-induced compression for one thing.

And maybe there is a general dissatisfaction with the available low-cost plugins.

My 2 cents

kylen · Post by **kylen** » Wed Oct 19, 2005 8:30 pm

pj geerlings wrote: Just MHO but I think there is more to it - the power-supply-sag-induced compression for one thing.

That's a good point - some amp systems are going to be slower or faster than others due to the PS. How would a pluginin simulate that - a simple compression single-band medium attack, slower release type of thing?

kylen · Post by **kylen** » Wed Oct 19, 2005 11:38 pm

At any rate - complicated as simulation of amplifier systems may be in a dsp I'm thinking I can focus on the harmonic distortion part with the idea that blending in a little sweetness is better that no sweetness at all. Granted the generation of the harmonic distortion can be more complicated in a physical tube amp system. I guess that's what the skepticism is all about - makes sense. The context of this is for audio restoration and remastering of my tape archives in the digital domain, in my DAW. Sweetening and repair in other words, recreating missing fullness, warmth and clarity thru the use of multi-band harmonics.

I know this might be a little dangerous but I'm trying to have a little fun here so I found a review that includes the standard measurement transfer curve of a Manley mono block using a 50Hz tone:
http://stereophile.com/tubepoweramps/670/index6.html

Not a whole lot of detail maybe for accurate modeling but a hella lotta fantasy for some spectral harmonic matching. The idea is to just look at the 50Hz harmonics and in general balance the odds and evens and see how it sounds across a full mix. Recreating a Manley mono-block, hehe probably not likely. To my ears, especially in ,multi-band mode, a lot better than simply pushing up a peaking band on an EQ to rebalance an old tape or recording.

Here's a bare bones picture of sony distortion (graph based waveshaper) that can do asymmetric (plus & minus) adjustments to the transfer curve:
Waveshaping Harmonic Balance

So - unless I'm mistaken some kind of waveshaper is underneath many of the harmonic distortion plugins. What else ? Trade secret - proprietary? Maybe so - that's not a problem for me since I'm a dsp plugin supporter and buyer. It is really tricky using a waveshaper(s) to generate certain patterns of harmonics...all I'm missing now are some really fine waveshaper presets: triodes, tetrodes, pentodes, single stage amps, dual-stage, 16-stage!

Aleksey Vaneev · Post by **Aleksey Vaneev** » Thu Oct 20, 2005 3:53 am

It should be added that waveshaping overdrive or distortion is based on voltage evaluation only. While tube distortion relies on voltage/current evaluation. At least tube 'transfer function' insists on that. In practice this creates a considerably less harsh distortion than what waveshaping can offer.

kylen · Post by **kylen** » Thu Oct 20, 2005 4:31 am

Thanks Aleksey - I was hoping you would stop by here for comment, and I already respect your work in this matter on the Tube Amp and Warmifier plugins I'm using.

I have lots of current/voltage curves and where to set tubes for their optimal and non-linear performance. That's a good thing to look at - thanks!

Do you think there's any use of looking at how a 50Hz or 1Khz signal performs in the amplifier system as a whole (like in the Manley Mono block example) - and then trying to "match" that type of harmonic performance. When I learn more about it I can think about how to match actual real-world performance where current and voltage don't have a fixed relationship. Wouldn't a static harmonic performance curve be a first step is what I'm wondering? Maybe better than nothing?

Aleksey Vaneev · Post by **Aleksey Vaneev** » Thu Oct 20, 2005 11:29 am

Theoretically, matching the harmonic structure may help get the same 'flavor', but in reality that's only a part of the story. Much more is usually going on in tubes, so it may simply not be possible to match tube sound with a waveshaper.

Jafo · Post by **Jafo** » Thu Oct 20, 2005 12:35 pm

Plus, isn't there something about how the harmonic structure changes with frequency and time? Not that I have any hope of ever understanding the math or physics involved, but I seem to recall Antti, Simulanalog, and the "Solid-State Tube" guy (who combined a bipolar transistor with a FET or something) going on for pages about that stuff.

I know that Ruby Tube is simply a waveshaper, and although it sounds really good, it just ain't the same thing.

There's definitely a (midrange, soft) compression involved also.

Actually, now that I think about it, the proportions of harmonics also change with gain; at low levels of drive, the sound seems to be mostly low- and even-order harmonics, while wiht higher gain, there is a large amount of odd- and slightly higher-order harmonics present. At least, that's what it sounds like to me; I have no hard data to back this up! But the character unquestionably changes somehow.

kylen · Post by **kylen** » Thu Oct 20, 2005 4:36 pm

Jafo wrote:Plus, isn't there something about how the harmonic structure changes with frequency and time? Not that I have any hope of ever understanding the math or physics involved, but I seem to recall Antti, Simulanalog, and the "Solid-State Tube" guy (who combined a bipolar transistor with a FET or something) going on for pages about that stuff.

I'm thinking multi-band may help here. At least to my ears it sounds less confusing when I break the harmonics up into 3 or 4 bands - depending on the material and amount of harmonic balancing needed.

Jafo wrote:Actually, now that I think about it, the proportions of harmonics also change with gain; at low levels of drive, the sound seems to be mostly low- and even-order harmonics, while wiht higher gain, there is a large amount of odd- and slightly higher-order harmonics present. At least, that's what it sounds like to me; I have no hard data to back this up! But the character unquestionably changes somehow.

Yes, good point - I've read that recently in my recent travels. AIPL Warm has a "Dynamics" slider that may try to control some of that - I'll have to check. Warmifier does something like that to a certain extent already in mode 2 I think - right Aleksey? At any rate I suppose different level signals would cause the tube and/or or system of tubs (amp in other words) to operate at different points on the current (I) transfer curve. So looking at a static 50Hz signal and the associated harmonic output at a single level (high level) gives only part of the picture for sure. I might be able to set up something in parallel - also some harmonic exciters have "mix" knobs that would allow part of the original dry signal back in untouched - the Sony distortion plugin does this. Putting a series or parallel compressor to simulate a little saturation at higher levels (or soft knees - I don't know what sounds better yet) would add some odd harmonics - Kjaerhus GCO-1 in classic mode does this according to my test last night, PSP also any many others I'm sure.

Thanks for the ideas & tips Jafo! Very interesting! I'm just trying to add and rebalance some harmonics to material that already has a certain amount in it. I'm thinking this would be more of a sweetening type thing - a type of tone control more that it would be an amplifier simulation although it would borrow certain components. Just dreaming actually!

kylen · Post by **kylen** » Thu Oct 20, 2005 4:47 pm

Aleksey Vaneev wrote:Theoretically, matching the harmonic structure may help get the same 'flavor', but in reality that's only a part of the story. Much more is usually going on in tubes, so it may simply not be possible to match tube sound with a waveshaper.

Yes a lot is going on for sure. Maybe that's all I'm after though is a little flavor. A little THD appropriately balanced and added in various critical bands - just enough to tickle my psychoacoustic mind. For fullness, warmth, presence, and air...using the richness of harmonics instead of a harmonically "flat" EQ as I normally would. If I was using a Manley analog EQ I would'nt have a choice - the harmonic color is in that thing naturally! Thanks for your comments Aleksey!

kylen · Post by **kylen** » Thu Oct 20, 2005 5:18 pm

OK - you guys are making me think about what I am REALLY trying to do...along with the fun experiments.

I am simply trying to rebalance the subbass, bass, mids, presence, treble in full mixes - we do this kind of thing every day either by mixing or during [re]mastering, right?

The way I want to do it in a DAW might be a little different since, in general, there is naturally no THD present in digital dsp (as in analog systems). By borrowing some elements that are naturally present in tube and tape systems that sound good psychoacoustically I merely mean to support the bass, mids, and treble already present in my audio by adding harmonic distortion to those bands. The THD I finally add may or may not match any amplifier or tape system but it must sound good psychoacoustically and not lead to "ear" fatigue. That's why I chose to add a certain mixture of even and odd harmonic distortion that occurs in nature and in devices tried and proven over time in audio recordings.

My Mission statement!

ED: At any rate now I know why you guys think waveshaping alone can't truely represent tube harmonics. Thanks!

antti · Post by **antti** » Thu Oct 20, 2005 7:28 pm

kylen wrote: What's the problem with waveshaping - isn't that what happens at some point down in the dsp anyway? Or are we just talking in terms of user interface?

The problem with waveshaping is that in almost all cases the developers do ONLY that. A real tube circuit is a combination of waveshaping and filtering inside feedback loops.

The feedback causes the bias, and thus gain, to vary depending on the signal (sticking a compressor there will not do the same!). It's a form of time-variant waveshaping.

annode · Post by **annode** » Thu Oct 20, 2005 8:17 pm

I think you`ll find lots of useful simulation theory and data at this site...
http://www.simulanalog.org/

Would you except this suggestion from a guy named "an(n)ode"?

kylen · Post by **kylen** » Fri Oct 21, 2005 5:09 am

Thanks for your comments antti and annode - that gives me some more ideas to research & think about, yes feedback loops - good point...hehe annode - great name for this topic!

I'm bothered by this statement I made earlier and have been thinking about it all day - something just doesn't seem right in it...

...in general, there is naturally no THD present in digital dsp (as in analog systems).

I need to think about that some more...maybe I'm confusing noise, linear, and non-linear performance...good time for a break while I do some more DAW testing and have some fun with harmonic distortion!

rogerperrin · Post by **rogerperrin** » Fri Oct 21, 2005 5:57 am

Hello Kylen, I'm not a synth writer/programmer, but a guitar player, so I thought I'd throw in my 2 cents. I'd just remind, that while we all talk about the magic of tubes and tube amps, there is more happening than just the tubes. There are the tubes, yes, but which tubes, the 12ax7's in preamp or the bigger power tubes. Also, the rectifier sag of the power supply, the audio output transformer, the speaker type and age, and the cabinet - large or small, open or closed back, plywood or particleboard. The speakers sound best if they are the older alnico magnet type instead of the newer ceramic magnets.

To me, the creamier the better in the distortion area. The distortion I hear in vst-ware is more digital noise than creamy. And it sounds awful. About 4 years ago I bought, repaired and sold about 12 old tube amps. In the process I got to play them all alot.

My favorites were the late 50's and 60's single stage tubers. The old amps in the 5 watt range just used like one tube for the preamp and the amplifier stage. They would have no gain or channel volume, just a master volume knob. The sweeter amps did not use the famous 12ax7's for preamp gain. But they sounded creamy. Jimmy Page, Eric Clapton, Jimi Hendrix - all used the smaller Supro Amps in their early days. And I wish I had never sold that old Supro I had. Lower the volume and it was a great jazz amp crank it a little to get near J.J. Cale and Eric Clapton, crank it a lot and I had the Stevie Ray Vaughn sound - all from one amp. If a dev could ever capture "creamy" in their distortion model, players would just drool, I promise.

Perhaps this

s

Why is waveshaping bad for simulating tube harmonics?