Axe-Fx III Firmware 10.00 Public Beta #2

Status
Not open for further replies.
It's a simple clipping circuit, copied from the Marshall 2203/2204. It's biased super cold, so the positive swing amplifies (although the AF is pretty low), but the negative swing drives into cutoff - the effect is not unlike a single-stage diode clipper.
Any triode stage is a "simple clipping circuit". IMO it's biased cold because if it were biased in the center it would cause so much blocking distortion that the amp would be unusable. You can't just look at a triode stage in isolation. You have to consider how it interacts with the other stages.

Most amps use a grid stopper between the last triode stage and the PI, often with another resistor to ground as a divider. Typical values are 220K. If the amp has a tone stack before the PI then that resistor is unnecessary as the tone stack provides the grid stopper action.

And I don't agree that it's copied from a JCM800. In a JCM800 that stage is before the final gain stage so it doesn't get driven nearly as hard. In a JCM800 that stage only clips for strong signals. For a typical mV signal the stage is only mildly distorting.
 
Hey, I'm a shredder but my brother n law is here and he is an incredible blues and country picker, what cab would go well with the Trainwreck? I want to give him a taste of the magic!
 
Kudos on cracking yet another frontier on the way to ultimate realness! The new algorithm definitely has a nice snarl!

Any chance the improvements for the plate impedance accuracy can be applied to the preamp modeling?
 
Hey, I'm a shredder but my brother n law is here and he is an incredible blues and country picker, what cab would go well with the Trainwreck? I want to give him a taste of the magic!
I like Legacy #132, 4x12 TV Mix #2
 
I feel like definitely there could be FAS models for this. Rather, more like having a FAS switch on the amp to toggle on/off the capability so there is not a huge increase in AMP models. Plus you'd be able to A/B them easily.
No idea if implementing an "ideal" mode for all amps would take up tons of resources/memory on the unit but it sounds like a lot of man hours to benefit such a niche group of people.

Agreed... and there is probably a design goal/constraint of keeping the amp models as lean/efficient as possible (while allowing enhancements/extensions like the gain enhancer).
 
I'm gonna have to pile on with this 10 beta. I've been a Fractal fan for a good while and have been very satisfied with the product, but the improvements in this issue and especially with the inclusion of the 'Gain Enhancer' Has just knocked it out of the park.

The Axe III is soooooo much fun to play now. And I'm finding amps I had dismissed in the past sounding fantastic.

Carry-on my good man Cliff! I don't know how you do it.
 
I've added a Tech Note about the Trainwreck Express tweaks.

I should make a recording of the real amp vs. the model with and without the tweaks. The real amp is a nasty thing. Very spitty as one would expect given the lack of a grid stopper resistor between the last triode stage and the PI.
This would be fantastic!
 
These last 2 updates are absolutely A joy to to play! The feel and sustain is so real and lively, its hard to put into words! I love getting updates and finding out what all is new, I hope that however many more updates there are, that they never lose this lively touch feeling of this Firmware.

Big thanks ,,love this !
 
Either that or a stickied post somewhere that talks about general tweaks you can do to amp models to make them better than the real thing. Cliff posts about these kinds of things every so often but I think it would be cool to have all of that info in one place. With having the info out there, people can tweak just the amps they like and leave everything else as is. No idea if implementing an "ideal" mode for all amps would take up tons of resources/memory on the unit but it sounds like a lot of man hours to benefit such a niche group of people.
Yes the wiki collects a lot of stuff but I wish there was some browsable knowledge base for the Axe there are so many amazingly cool tweaks and tricks Cliff and co toss out here. I try to collect them in a notes file on my computer but it would be so cool to have a better system. Regardless those tips are much appreciated!!
 
Yes the wiki collects a lot of stuff but I wish there was some browsable knowledge base for the Axe there are so many amazingly cool tweaks and tricks Cliff and co toss out here. I try to collect them in a notes file on my computer but it would be so cool to have a better system. Regardless those tips are much appreciated!!

Yek told me he may bring the Tips section back for that exact purpose.
 
It's been since 7.0 or so, since I keep thinking that it can't get substantially better than this....

...but this time...it really can't get much better!!!

....can it...?
 
And I don't agree that it's copied from a JCM800. In a JCM800 that stage is before the final gain stage so it doesn't get driven nearly as hard. In a JCM800 that stage only clips for strong signals. For a typical mV signal the stage is only mildly distorting.

Measure the voltage swing in both types of amp. It's not very different. In the Marshall, it's got a single triode stage in front of it; in the TW/Komet, it's got two, but it's also got the tone stack in-between, which drops a fair amount of signal.

Further, if it was done for the reason you give, there would be no reason for it to exist in the Komet Concorde (since it precedes a cathode follower). But it does. And switching it out (I built a clone years ago, in fact you may recall that I supplied you with the schematics for the K60 and Concorde) noticeably reduces the clipping at any given volume.

If you remove the triode we're talking about, you're left with what amounts to a BF/SF Fender non-reverb channel with Marshally tone stack values - triode, tone stack, triode, PI, power amp. I don't know why one would expect the TW to experience blocking that a Fender doesn't. Especially since the TWs have much smaller coupling caps into the power amp. The 220K resistors you mention in Fenderoids aren't grid stoppers; they're mixing resistors, and would be unnecessary if there were only one channel - like, for example, the Princeton/Princeton reverb. No resistor.

And all triodes aren't automatically 'clipping' circuits in the same sense - these are designed to clip asymmetrically with relatively little input signal. Biased to the midpoint, you'd have to hit it much harder to get any clipping at all.
 
Last edited:
It's been since 7.0 or so, since I keep thinking that it can't get substantially better than this....

...but this time...it really can't get much better!!!

....can it...?
Each of these advances usually spawns another advance. The work I did on the power amp modeling actually gave me an idea on how to improve the Drive modeling. I'm coding it now. If my suspicions are correct it will improve the accuracy and lower the CPU usage.
 
Status
Not open for further replies.
Back
Top Bottom