How good amp sim emulate real circuits? Amplitube '65 Twin reverb vs TH3 '65 Darkface vs LTSpice

[cprompt]

In fact I did try to read it, but it was too long and the pictures didn't match the text content and at best it was vague

I read it, took just a couple of minutes, the pictures fit right in and the text was clear and well written. If you found it long and vague, well, I don't know what to say really.

To echo what others have said, I find this fascinating and would be very interested to see how other ampsims stack up against your model - the Kazrog Thermionic stuff for example as they are among my favourites.

alteregoxxx - what would you think the margin of error between two identical amps would be? Do you think that the discrepancy between Amplitube and your model is down to just the fact that they modelled a physically different amp to the one you used, and that perhaps that is within the margin of error you would expect?

[stratum]

It's plenty of shorter and less vague topics in the forum, you could go comment those ones, no?

I'll give you a useful answer.

Accuracy of modelling is open to debate. Have a look at http://www.simulanalog.org/statevariable.pdf
This shows the real difficulty in modelling. It's just a pair of diodes and a capacitor and the modelling algorithm seeks the root of an equation by newton raphson for each sample. Extending this to an whole circuit necessarily requires compromises to be made because the generalized solution would be very inefficient.

Matching those curves that you plot wouldn't be too hard, but probably was pointless because they would change with respect to the cabinet/speaker you use, and I guess some ad-hoc adjustments by ear might have been made.

I read it, took just a couple of minutes, the pictures fit right in and the text was clear and well written. If you found it long and vague, well, I don't know what to say really.

Well, I know, I am coffee addict and didn't drink one for a long time. That may be the reason as to why I don't get it

[alteregoxxx]

alteregoxxx - what would you think the margin of error between two identical amps would be? Do you think that the discrepancy between Amplitube and your model is down to just the fact that they modelled a physically different amp to the one you used, and that perhaps that is within the margin of error you would expect?

Nope.
No way the circuit of a '65 TwinReverb can have that frequency response.

For sure two different real '65 TwinReverb circuits will not measure exactly the same, but for the specific case you're referring too, we're talking of a frequency response that is off by 20dB in comparison to the expected one. If they have modeled a TwinReverb 65 specimen, it for sure was a faulty one.

[jens]

It's plenty of shorter and less vague topics in the forum, you could go comment those ones, no?

I'll give you a useful answer.

Accuracy of modelling is open to debate. Have a look at http://www.simulanalog.org/statevariable.pdf

Now THAT is really funny!

You might not be aware of it, but Simulanalog was founded by Thomas Serafini before he became Overloud's - original and current - CTO many years ago.

[Jafo]

But, here

1)we're not rocking hard, so no"memory" effect, sorry

2)Fender Twin Reverb '65 electronic circuit has lots and lots of headroom, read no distortion, bias shift etc etc...Again, sorry

Consider me schooled.

(I really ought to know better than to let Guitarist Mike post... he really should just shut up and play his guitar, as the man said. Keysman Mike and Grammar Nazi Mike are much wiser and kinder.)

[stratum]

Now THAT is really funny!

It's not funny at all. You are comparing frequency responses of software models and your models, and as long as there isn't much distortion that would be OK, and you may rightly claim IKM models a faulty amp. OK.

When there is distortion solidstate analog models are worse than software models, because there are few things that you can do to change component behavior. If a silicon diodes current-voltage relationship follows an exponential function and if a triode's grid would follow a power function, good luck modelling that with analog solid state electronics.

A software model doesn't have this problem, the problem is a very different one, the solution doesn't scale to an whole circuit and your comparison method totally misses the point. I'd say it's irrelevant but you can nevertheless still fill a bug report to IKM and say their model is 20dB off the target. That's a reasonable complaint.

[alteregoxxx]

Now THAT is really funny!

It's not funny at all. You are comparing frequency responses of software models and your models, and as long as there isn't much distortion that would be OK, and you may rightly claim IKM models a faulty amp. OK.

When there is distortion solidstate analog models are worse than software models, because there are few things that you can do to change component behavior. If a silicon diodes current-voltage relationship follows an exponential function and if a triode's grid would follow a power function, good luck modelling that with solid state electronics.

A software model doesn't have this problem, the problem is a very different one, the solution doesn't scale to an whole circuit and your comparison method totally misses the point. I'd say it's irrelevant but you can nevertheless still fill a bug report to IKM and say their model is 20dB off the target. That's a reasonable complaint.

Man, I think you really need (lots of) the coffee you were talking of... ...or, maybe less...


First, the person you're quoting it's not me, the one who started the topic.
Second, it seems you keep spitting out "small pieces of things" you read and/or learned somewhere on the internet, in a completely uncorrelated fashion with what we're discussing here.
Exponental law, power law. Let's add in the Child-Langmuir Law and we're done, ok?

What's your problem with my approach? When you'll have found one, Can you explain technically why TH3 models adhere more or less perfectly to the LTSpice ones and the ones modeled in Amplitube did not? Maybe the Department of the University of California at Berkeley, when years ago created the first SPICE engine, made a secret agremeent with Overloud TH3, well before the latter know they would have ever been existed ? A conspiracy aiming in circuits frequency response faking, in a way that years and years later amp circuit simulated in LTSpice match with the ones made by Overloud but not with the ones in Amplitube?

Are you saying LTSPice produce inaccurate results? TH3 is inaccurate? Amplitube is inaccurate? The world is inaccurate? What? Please, enligten us!

[stratum]

You can figure out yourself what's wrong with your approach. It's waste of time to discuss because of your style and because of the damn coffee you keep quoting and other quotes that you mind.

[alteregoxxx]

You can figure out yourself what's wrong with your approach. It's waste of time to discuss because of your style and because of the damn coffee you keep quoting and other quotes that you mind.

Ok, thanks for your lessons. All of us have learnt a lot from you.
Now it's time to leave us with our senseless approach in discovering nothing. Please

[stratum]

Now it's time to leave us with our senseless approach in discovering nothing. Please

No problem.

[alteregoxxx]

But, here

1)we're not rocking hard, so no"memory" effect, sorry

2)Fender Twin Reverb '65 electronic circuit has lots and lots of headroom, read no distortion, bias shift etc etc...Again, sorry

Consider me schooled.

(I really ought to know better than to let Guitarist Mike post... he really should just shut up and play his guitar, as the man said. Keysman Mike and Grammar Nazi Mike are much wiser and kinder.)

I'm sorry, but english is not my native language, hence I've not understood if your is a way to say that I've been rude and could have offended you. If that's the case, I'm sorry, it was not my intention!

[jens]

I'd say it's irrelevant but you can nevertheless still fill a bug report to IKM and say their model is 20dB off the target. That's a reasonable complaint.

Thanks for revealing your agenda - that's settled then.

[stratum]

Thanks for revealing your agenda - that's settled then.

Pretty unlucky thread, but I have left due to the original poster's request.

[Jafo]

But, here

1)we're not rocking hard, so no"memory" effect, sorry

2)Fender Twin Reverb '65 electronic circuit has lots and lots of headroom, read no distortion, bias shift etc etc...Again, sorry

Consider me schooled.

(I really ought to know better than to let Guitarist Mike post... he really should just shut up and play his guitar, as the man said. Keysman Mike and Grammar Nazi Mike are much wiser and kinder.)

I'm sorry, but english is not my native language, hence I've not understood if your is a way to say that I've been rude and could have offended you. If that's the case, I'm sorry, it was not my intention!

Back at you! No offense taken; I was just being stupid, and you kindly pointed that out. Your explanation was well thought-out and well-delivered. The fault is entirely mine.

My approach to English (the only language I know well) can be overly stylistic and playful, and I sometimes forget that the knots of words I create can be difficult to unravel.

As for Guitarist/Keysman/Grammar Nazi Mike, they're all me... just different aspects of my personality. As a guitarist, I'm obsessed with little details that really don't matter, and I let that cloud my judgement. We guitarists are a strange breed!

So, to summarize: You have a great idea for a thread, and you've been far more thoughtful and professional than some of us deserve. (Well, more than I deserve, at least!) Thank you.

[GreyLion]

The detailed tech talk in this thread is mostly over my head, but I am curious --

I'd love to see you test out Peavey's Revalver 4, since supposedly, each amp model can generate a schematic that you can tweak at the component level. I wonder if you can simply switch around a few capacitors or resistors inside the appropriate model and create a perfectly matching curve.

Frustratingly, Peavey hides the names a bit, presumably because of trademark issues, but it looks like their Basic 100 amp can do the Twin thing. (Hints that it's modeled on a Bassman100, which should sortakinda get you close. It's extremely likely that someone here is much more knowledgeable about this than I am.) Revalver has a free version, and the add-on model is only $7.99, so it wouldn't be a huge expenditure.